Release 0.6.6

- New command-line tool: vitastor-cli
- Implement layer (snapshot/clone) merge and delete
- Remove 'bool' from the C header
- Fix a very rare flusher stall
- More diagnostics now printed for slow ops in the log

CMakeLists.txt

@@ -2,6 +2,6 @@ cmake_minimum_required(VERSION 2.8)
 
 project(vitastor)
 
-set(VERSION "0.6.3")
+set(VERSION "0.6.6")
 
 add_subdirectory(src)

README-ru.md

@@ -22,7 +22,6 @@ Vitastor на данный момент находится в статусе п
 
 Однако следующее уже реализовано:
 
-0.5.x (стабильная версия):
 - Базовая часть - надёжное кластерное блочное хранилище без единой точки отказа
 - Производительность ;-D
 - Несколько схем отказоустойчивости: репликация, XOR n+1 (1 диск чётности), коды коррекции ошибок
@@ -41,21 +40,23 @@ Vitastor на данный момент находится в статусе п
 - Драйвер диска для QEMU (собирается вне дерева исходников QEMU)
 - Драйвер диска для утилиты тестирования производительности fio (также собирается вне дерева исходников fio)
 - NBD-прокси для монтирования образов ядром ("блочное устройство в режиме пользователя")
-- Утилита удаления образов/инодов (vitastor-rm)
+- Утилита для удаления образов/инодов (vitastor-cli rm)
 - Пакеты для Debian и CentOS
-
-0.6.x (master-ветка):
 - Статистика операций ввода/вывода и занятого места в разрезе инодов
 - Именование инодов через хранение их метаданных в etcd
 - Снапшоты и copy-on-write клоны
 - Сглаживание производительности случайной записи в SSD+HDD конфигурациях
 - Поддержка RDMA/RoCEv2 через libibverbs
+- CSI-плагин для Kubernetes
+- Базовая поддержка OpenStack: драйвер Cinder, патчи для Nova и libvirt
+- Слияние снапшотов (vitastor-cli {snap-rm,flatten,merge})
 
 ## Планы развития
 
+- Поддержка удаления снапшотов (слияния слоёв)
 - Более корректные скрипты разметки дисков и автоматического запуска OSD
 - Другие инструменты администрирования
-- Плагины для OpenStack, Kubernetes, OpenNebula, Proxmox и других облачных систем
+- Плагины для OpenNebula, Proxmox и других облачных систем
 - iSCSI-прокси
 - Более быстрое переключение при отказах
 - Фоновая проверка целостности без контрольных сумм (сверка реплик)
@@ -372,7 +373,7 @@ Vitastor с однопоточной NBD прокси на том же стен
 - Установите gcc и g++ 8.x или новее.
 - Склонируйте данный репозиторий с подмодулями: `git clone https://yourcmc.ru/git/vitalif/vitastor/`.
 - Желательно пересобрать QEMU с патчем, который делает необязательным запуск через LD_PRELOAD.
-  См `qemu-*.*-vitastor.patch` - выберите версию, наиболее близкую вашей версии QEMU.
+  См `patches/qemu-*.*-vitastor.patch` - выберите версию, наиболее близкую вашей версии QEMU.
 - Установите QEMU 3.0 или новее, возьмите исходные коды установленного пакета, начните его пересборку,
   через некоторое время остановите её и скопируйте следующие заголовки:
    - `<qemu>/include` &rarr; `<vitastor>/qemu/include`
@@ -491,10 +492,10 @@ qemu-system-x86_64 -enable-kvm -m 1024
 
 ### Удалить образ
 
-Используйте утилиту vitastor-rm. Например:
+Используйте утилиту vitastor-cli rm. Например:
 
 ```
-vitastor-rm --etcd_address 10.115.0.10:2379/v3 --pool 1 --inode 1 --parallel_osds 16 --iodepth 32
+vitastor-cli rm --etcd_address 10.115.0.10:2379/v3 --pool 1 --inode 1 --parallel_osds 16 --iodepth 32
 ```
 
 ### NBD
@@ -511,6 +512,21 @@ vitastor-nbd map --etcd_address 10.115.0.10:2379/v3 --image testimg
 Для обращения по номеру инода, аналогично другим командам, можно использовать опции
 `--pool <POOL> --inode <INODE> --size <SIZE>` вместо `--image testimg`.
 
+### Kubernetes
+
+У Vitastor есть CSI-плагин для Kubernetes, поддерживающий RWO-тома.
+
+Для установки возьмите манифесты из директории [csi/deploy/](csi/deploy/), поместите
+вашу конфигурацию подключения к Vitastor в [csi/deploy/001-csi-config-map.yaml](001-csi-config-map.yaml),
+настройте StorageClass в [csi/deploy/009-storage-class.yaml](009-storage-class.yaml)
+и примените все `NNN-*.yaml` к вашей инсталляции Kubernetes.
+
+```
+for i in ./???-*.yaml; do kubectl apply -f $i; done
+```
+
+После этого вы сможете создавать PersistentVolume. Пример смотрите в файле [csi/deploy/example-pvc.yaml](csi/deploy/example-pvc.yaml).
+
 ## Известные проблемы
 
 - Запросы удаления объектов могут в данный момент приводить к "неполным" объектам в EC-пулах,

README.md

@@ -16,7 +16,6 @@ with configurable redundancy (replication or erasure codes/XOR).
 Vitastor is currently a pre-release, a lot of features are missing and you can still expect
 breaking changes in the future. However, the following is implemented:
 
-0.5.x (stable):
 - Basic part: highly-available block storage with symmetric clustering and no SPOF
 - Performance ;-D
 - Multiple redundancy schemes: Replication, XOR n+1, Reed-Solomon erasure codes
@@ -35,21 +34,23 @@ breaking changes in the future. However, the following is implemented:
 - QEMU driver (built out-of-tree)
 - Loadable fio engine for benchmarks (also built out-of-tree)
 - NBD proxy for kernel mounts
-- Inode removal tool (vitastor-rm)
+- Inode removal tool (vitastor-cli rm)
 - Packaging for Debian and CentOS
-
-0.6.x (master):
 - Per-inode I/O and space usage statistics
 - Inode metadata storage in etcd
 - Snapshots and copy-on-write image clones
 - Write throttling to smooth random write workloads in SSD+HDD configurations
 - RDMA/RoCEv2 support via libibverbs
+- CSI plugin for Kubernetes
+- Basic OpenStack support: Cinder driver, Nova and libvirt patches
+- Snapshot merge tool (vitastor-cli {snap-rm,flatten,merge})
 
 ## Roadmap
 
+- Snapshot deletion (layer merge) support
 - Better OSD creation and auto-start tools
 - Other administrative tools
-- Plugins for OpenStack, Kubernetes, OpenNebula, Proxmox and other cloud systems
+- Plugins for OpenNebula, Proxmox and other cloud systems
 - iSCSI proxy
 - Faster failover
 - Scrubbing without checksums (verification of replicas)
@@ -340,7 +341,7 @@ Vitastor with single-thread NBD on the same hardware:
       * For QEMU 2.0+: `<qemu>/qapi-types.h` &rarr; `<vitastor>/qemu/b/qemu/qapi-types.h`
    - `config-host.h` and `qapi` are required because they contain generated headers
 - You can also rebuild QEMU with a patch that makes LD_PRELOAD unnecessary to load vitastor driver.
-  See `qemu-*.*-vitastor.patch`.
+  See `patches/qemu-*.*-vitastor.patch`.
 - Install fio 3.7 or later, get its source and symlink it into `<vitastor>/fio`.
 - Build & install Vitastor with `mkdir build && cd build && cmake .. && make -j8 && make install`.
   Pay attention to the `QEMU_PLUGINDIR` cmake option - it must be set to `qemu-kvm` on RHEL.
@@ -461,6 +462,21 @@ It will output the device name, like /dev/nbd0 which you can then format and mou
 
 Again, you can use `--pool <POOL> --inode <INODE> --size <SIZE>` insteaf of `--image <IMAGE>` if you want.
 
+### Kubernetes
+
+Vitastor has a CSI plugin for Kubernetes which supports RWO volumes.
+
+To deploy it, take manifests from [csi/deploy/](csi/deploy/) directory, put your
+Vitastor configuration in [csi/deploy/001-csi-config-map.yaml](001-csi-config-map.yaml),
+configure storage class in [csi/deploy/009-storage-class.yaml](009-storage-class.yaml)
+and apply all `NNN-*.yaml` manifests to your Kubernetes installation:
+
+```
+for i in ./???-*.yaml; do kubectl apply -f $i; done
+```
+
+After that you'll be able to create PersistentVolumes. See example in [csi/deploy/example-pvc.yaml](csi/deploy/example-pvc.yaml).
+
 ## Known Problems
 
 - Object deletion requests may currently lead to 'incomplete' objects in EC pools

csi/.dockerignore

@@ -0,0 +1,3 @@
+vitastor-csi
+go.sum
+Dockerfile

csi/Dockerfile

@@ -0,0 +1,32 @@
+# Compile stage
+FROM golang:buster AS build
+
+ADD go.mod /app/
+RUN cd /app; CGO_ENABLED=1 GOOS=linux GOARCH=amd64 go mod download -x
+ADD . /app
+RUN perl -i -e '$/ = undef; while(<>) { s/\n\s*(\{\s*\n)/$1\n/g; s/\}(\s*\n\s*)else\b/$1} else/g; print; }' `find /app -name '*.go'`
+RUN cd /app; CGO_ENABLED=1 GOOS=linux GOARCH=amd64 go build -o vitastor-csi
+
+# Final stage
+FROM debian:buster
+
+LABEL maintainers="Vitaliy Filippov <vitalif@yourcmc.ru>"
+LABEL description="Vitastor CSI Driver"
+
+ENV NODE_ID=""
+ENV CSI_ENDPOINT=""
+
+RUN apt-get update && \
+    apt-get install -y wget && \
+    wget -q -O /etc/apt/trusted.gpg.d/vitastor.gpg https://vitastor.io/debian/pubkey.gpg && \
+    (echo deb http://vitastor.io/debian buster main > /etc/apt/sources.list.d/vitastor.list) && \
+    (echo deb http://deb.debian.org/debian buster-backports main > /etc/apt/sources.list.d/backports.list) && \
+    (echo "APT::Install-Recommends false;" > /etc/apt/apt.conf) && \
+    apt-get update && \
+    apt-get install -y e2fsprogs xfsprogs vitastor kmod && \
+    apt-get clean && \
+    (echo options nbd nbds_max=128 > /etc/modprobe.d/nbd.conf)
+
+COPY --from=build /app/vitastor-csi /bin/
+
+ENTRYPOINT ["/bin/vitastor-csi"]

csi/Makefile

@@ -0,0 +1,9 @@
+VERSION ?= v0.6.6
+
+all: build push
+
+build:
+	@docker build --rm -t vitalif/vitastor-csi:$(VERSION) .
+
+push:
+	@docker push vitalif/vitastor-csi:$(VERSION)

csi/deploy/000-csi-namespace.yaml

@@ -0,0 +1,5 @@
+---
+apiVersion: v1
+kind: Namespace
+metadata:
+  name: vitastor-system

csi/deploy/001-csi-config-map.yaml

@@ -0,0 +1,9 @@
+---
+apiVersion: v1
+kind: ConfigMap
+data:
+  vitastor.conf: |-
+    {"etcd_address":"http://192.168.7.2:2379","etcd_prefix":"/vitastor"}
+metadata:
+  namespace: vitastor-system
+  name: vitastor-config

csi/deploy/002-csi-nodeplugin-rbac.yaml

@@ -0,0 +1,37 @@
+---
+apiVersion: v1
+kind: ServiceAccount
+metadata:
+  namespace: vitastor-system
+  name: vitastor-csi-nodeplugin
+---
+kind: ClusterRole
+apiVersion: rbac.authorization.k8s.io/v1
+metadata:
+  namespace: vitastor-system
+  name: vitastor-csi-nodeplugin
+rules:
+  - apiGroups: [""]
+    resources: ["nodes"]
+    verbs: ["get"]
+  # allow to read Vault Token and connection options from the Tenants namespace
+  - apiGroups: [""]
+    resources: ["secrets"]
+    verbs: ["get"]
+  - apiGroups: [""]
+    resources: ["configmaps"]
+    verbs: ["get"]
+---
+kind: ClusterRoleBinding
+apiVersion: rbac.authorization.k8s.io/v1
+metadata:
+  namespace: vitastor-system
+  name: vitastor-csi-nodeplugin
+subjects:
+  - kind: ServiceAccount
+    name: vitastor-csi-nodeplugin
+    namespace: vitastor-system
+roleRef:
+  kind: ClusterRole
+  name: vitastor-csi-nodeplugin
+  apiGroup: rbac.authorization.k8s.io

csi/deploy/003-csi-nodeplugin-psp.yaml

@@ -0,0 +1,72 @@
+---
+apiVersion: policy/v1beta1
+kind: PodSecurityPolicy
+metadata:
+  namespace: vitastor-system
+  name: vitastor-csi-nodeplugin-psp
+spec:
+  allowPrivilegeEscalation: true
+  allowedCapabilities:
+    - 'SYS_ADMIN'
+  fsGroup:
+    rule: RunAsAny
+  privileged: true
+  hostNetwork: true
+  hostPID: true
+  runAsUser:
+    rule: RunAsAny
+  seLinux:
+    rule: RunAsAny
+  supplementalGroups:
+    rule: RunAsAny
+  volumes:
+    - 'configMap'
+    - 'emptyDir'
+    - 'projected'
+    - 'secret'
+    - 'downwardAPI'
+    - 'hostPath'
+  allowedHostPaths:
+    - pathPrefix: '/dev'
+      readOnly: false
+    - pathPrefix: '/run/mount'
+      readOnly: false
+    - pathPrefix: '/sys'
+      readOnly: false
+    - pathPrefix: '/lib/modules'
+      readOnly: true
+    - pathPrefix: '/var/lib/kubelet/pods'
+      readOnly: false
+    - pathPrefix: '/var/lib/kubelet/plugins/csi.vitastor.io'
+      readOnly: false
+    - pathPrefix: '/var/lib/kubelet/plugins_registry'
+      readOnly: false
+    - pathPrefix: '/var/lib/kubelet/plugins'
+      readOnly: false
+
+---
+kind: Role
+apiVersion: rbac.authorization.k8s.io/v1
+metadata:
+  namespace: vitastor-system
+  name: vitastor-csi-nodeplugin-psp
+rules:
+  - apiGroups: ['policy']
+    resources: ['podsecuritypolicies']
+    verbs: ['use']
+    resourceNames: ['vitastor-csi-nodeplugin-psp']
+
+---
+kind: RoleBinding
+apiVersion: rbac.authorization.k8s.io/v1
+metadata:
+  namespace: vitastor-system
+  name: vitastor-csi-nodeplugin-psp
+subjects:
+  - kind: ServiceAccount
+    name: vitastor-csi-nodeplugin
+    namespace: vitastor-system
+roleRef:
+  kind: Role
+  name: vitastor-csi-nodeplugin-psp
+  apiGroup: rbac.authorization.k8s.io

csi/deploy/004-csi-nodeplugin.yaml

@@ -0,0 +1,140 @@
+---
+kind: DaemonSet
+apiVersion: apps/v1
+metadata:
+  namespace: vitastor-system
+  name: csi-vitastor
+spec:
+  selector:
+    matchLabels:
+      app: csi-vitastor
+  template:
+    metadata:
+      namespace: vitastor-system
+      labels:
+        app: csi-vitastor
+    spec:
+      serviceAccountName: vitastor-csi-nodeplugin
+      hostNetwork: true
+      hostPID: true
+      priorityClassName: system-node-critical
+      # to use e.g. Rook orchestrated cluster, and mons' FQDN is
+      # resolved through k8s service, set dns policy to cluster first
+      dnsPolicy: ClusterFirstWithHostNet
+      containers:
+        - name: driver-registrar
+          # This is necessary only for systems with SELinux, where
+          # non-privileged sidecar containers cannot access unix domain socket
+          # created by privileged CSI driver container.
+          securityContext:
+            privileged: true
+          image: k8s.gcr.io/sig-storage/csi-node-driver-registrar:v2.2.0
+          args:
+            - "--v=5"
+            - "--csi-address=/csi/csi.sock"
+            - "--kubelet-registration-path=/var/lib/kubelet/plugins/csi.vitastor.io/csi.sock"
+          env:
+            - name: KUBE_NODE_NAME
+              valueFrom:
+                fieldRef:
+                  fieldPath: spec.nodeName
+          volumeMounts:
+            - name: socket-dir
+              mountPath: /csi
+            - name: registration-dir
+              mountPath: /registration
+        - name: csi-vitastor
+          securityContext:
+            privileged: true
+            capabilities:
+              add: ["SYS_ADMIN"]
+            allowPrivilegeEscalation: true
+          image: vitalif/vitastor-csi:v0.6.6
+          args:
+            - "--node=$(NODE_ID)"
+            - "--endpoint=$(CSI_ENDPOINT)"
+          env:
+            - name: NODE_ID
+              valueFrom:
+                fieldRef:
+                  fieldPath: spec.nodeName
+            - name: CSI_ENDPOINT
+              value: unix:///csi/csi.sock
+          imagePullPolicy: "IfNotPresent"
+          ports:
+          - containerPort: 9898
+            name: healthz
+            protocol: TCP
+          livenessProbe:
+            failureThreshold: 5
+            httpGet:
+              path: /healthz
+              port: healthz
+            initialDelaySeconds: 10
+            timeoutSeconds: 3
+            periodSeconds: 2
+          volumeMounts:
+            - name: socket-dir
+              mountPath: /csi
+            - mountPath: /dev
+              name: host-dev
+            - mountPath: /sys
+              name: host-sys
+            - mountPath: /run/mount
+              name: host-mount
+            - mountPath: /lib/modules
+              name: lib-modules
+              readOnly: true
+            - name: vitastor-config
+              mountPath: /etc/vitastor
+            - name: plugin-dir
+              mountPath: /var/lib/kubelet/plugins
+              mountPropagation: "Bidirectional"
+            - name: mountpoint-dir
+              mountPath: /var/lib/kubelet/pods
+              mountPropagation: "Bidirectional"
+        - name: liveness-probe
+          securityContext:
+            privileged: true
+          image: quay.io/k8scsi/livenessprobe:v1.1.0
+          args:
+            - "--csi-address=$(CSI_ENDPOINT)"
+            - "--health-port=9898"
+          env:
+            - name: CSI_ENDPOINT
+              value: unix://csi/csi.sock
+          volumeMounts:
+          - mountPath: /csi
+            name: socket-dir
+      volumes:
+        - name: socket-dir
+          hostPath:
+            path: /var/lib/kubelet/plugins/csi.vitastor.io
+            type: DirectoryOrCreate
+        - name: plugin-dir
+          hostPath:
+            path: /var/lib/kubelet/plugins
+            type: Directory
+        - name: mountpoint-dir
+          hostPath:
+            path: /var/lib/kubelet/pods
+            type: DirectoryOrCreate
+        - name: registration-dir
+          hostPath:
+            path: /var/lib/kubelet/plugins_registry/
+            type: Directory
+        - name: host-dev
+          hostPath:
+            path: /dev
+        - name: host-sys
+          hostPath:
+            path: /sys
+        - name: host-mount
+          hostPath:
+            path: /run/mount
+        - name: lib-modules
+          hostPath:
+            path: /lib/modules
+        - name: vitastor-config
+          configMap:
+            name: vitastor-config

csi/deploy/005-csi-provisioner-rbac.yaml

@@ -0,0 +1,102 @@
+---
+apiVersion: v1
+kind: ServiceAccount
+metadata:
+  namespace: vitastor-system
+  name: vitastor-csi-provisioner
+
+---
+kind: ClusterRole
+apiVersion: rbac.authorization.k8s.io/v1
+metadata:
+  namespace: vitastor-system
+  name: vitastor-external-provisioner-runner
+rules:
+  - apiGroups: [""]
+    resources: ["nodes"]
+    verbs: ["get", "list", "watch"]
+  - apiGroups: [""]
+    resources: ["secrets"]
+    verbs: ["get", "list", "watch"]
+  - apiGroups: [""]
+    resources: ["events"]
+    verbs: ["list", "watch", "create", "update", "patch"]
+  - apiGroups: [""]
+    resources: ["persistentvolumes"]
+    verbs: ["get", "list", "watch", "create", "update", "delete", "patch"]
+  - apiGroups: [""]
+    resources: ["persistentvolumeclaims"]
+    verbs: ["get", "list", "watch", "update"]
+  - apiGroups: [""]
+    resources: ["persistentvolumeclaims/status"]
+    verbs: ["update", "patch"]
+  - apiGroups: ["storage.k8s.io"]
+    resources: ["storageclasses"]
+    verbs: ["get", "list", "watch"]
+  - apiGroups: ["snapshot.storage.k8s.io"]
+    resources: ["volumesnapshots"]
+    verbs: ["get", "list"]
+  - apiGroups: ["snapshot.storage.k8s.io"]
+    resources: ["volumesnapshotcontents"]
+    verbs: ["create", "get", "list", "watch", "update", "delete"]
+  - apiGroups: ["snapshot.storage.k8s.io"]
+    resources: ["volumesnapshotclasses"]
+    verbs: ["get", "list", "watch"]
+  - apiGroups: ["storage.k8s.io"]
+    resources: ["volumeattachments"]
+    verbs: ["get", "list", "watch", "update", "patch"]
+  - apiGroups: ["storage.k8s.io"]
+    resources: ["volumeattachments/status"]
+    verbs: ["patch"]
+  - apiGroups: ["storage.k8s.io"]
+    resources: ["csinodes"]
+    verbs: ["get", "list", "watch"]
+  - apiGroups: ["snapshot.storage.k8s.io"]
+    resources: ["volumesnapshotcontents/status"]
+    verbs: ["update"]
+  - apiGroups: [""]
+    resources: ["configmaps"]
+    verbs: ["get"]
+---
+kind: ClusterRoleBinding
+apiVersion: rbac.authorization.k8s.io/v1
+metadata:
+  namespace: vitastor-system
+  name: vitastor-csi-provisioner-role
+subjects:
+  - kind: ServiceAccount
+    name: vitastor-csi-provisioner
+    namespace: vitastor-system
+roleRef:
+  kind: ClusterRole
+  name: vitastor-external-provisioner-runner
+  apiGroup: rbac.authorization.k8s.io
+
+---
+kind: Role
+apiVersion: rbac.authorization.k8s.io/v1
+metadata:
+  namespace: vitastor-system
+  name: vitastor-external-provisioner-cfg
+rules:
+  - apiGroups: [""]
+    resources: ["configmaps"]
+    verbs: ["get", "list", "watch", "create", "update", "delete"]
+  - apiGroups: ["coordination.k8s.io"]
+    resources: ["leases"]
+    verbs: ["get", "watch", "list", "delete", "update", "create"]
+
+---
+kind: RoleBinding
+apiVersion: rbac.authorization.k8s.io/v1
+metadata:
+  name: vitastor-csi-provisioner-role-cfg
+  namespace: vitastor-system
+subjects:
+  - kind: ServiceAccount
+    name: vitastor-csi-provisioner
+    namespace: vitastor-system
+roleRef:
+  kind: Role
+  name: vitastor-external-provisioner-cfg
+  apiGroup: rbac.authorization.k8s.io

csi/deploy/006-csi-provisioner-psp.yaml

@@ -0,0 +1,60 @@
+---
+apiVersion: policy/v1beta1
+kind: PodSecurityPolicy
+metadata:
+  namespace: vitastor-system
+  name: vitastor-csi-provisioner-psp
+spec:
+  allowPrivilegeEscalation: true
+  allowedCapabilities:
+    - 'SYS_ADMIN'
+  fsGroup:
+    rule: RunAsAny
+  privileged: true
+  runAsUser:
+    rule: RunAsAny
+  seLinux:
+    rule: RunAsAny
+  supplementalGroups:
+    rule: RunAsAny
+  volumes:
+    - 'configMap'
+    - 'emptyDir'
+    - 'projected'
+    - 'secret'
+    - 'downwardAPI'
+    - 'hostPath'
+  allowedHostPaths:
+    - pathPrefix: '/dev'
+      readOnly: false
+    - pathPrefix: '/sys'
+      readOnly: false
+    - pathPrefix: '/lib/modules'
+      readOnly: true
+
+---
+kind: Role
+apiVersion: rbac.authorization.k8s.io/v1
+metadata:
+  namespace: vitastor-system
+  name: vitastor-csi-provisioner-psp
+rules:
+  - apiGroups: ['policy']
+    resources: ['podsecuritypolicies']
+    verbs: ['use']
+    resourceNames: ['vitastor-csi-provisioner-psp']
+
+---
+kind: RoleBinding
+apiVersion: rbac.authorization.k8s.io/v1
+metadata:
+  name: vitastor-csi-provisioner-psp
+  namespace: vitastor-system
+subjects:
+  - kind: ServiceAccount
+    name: vitastor-csi-provisioner
+    namespace: vitastor-system
+roleRef:
+  kind: Role
+  name: vitastor-csi-provisioner-psp
+  apiGroup: rbac.authorization.k8s.io

csi/deploy/007-csi-provisioner.yaml

@@ -0,0 +1,159 @@
+---
+kind: Service
+apiVersion: v1
+metadata:
+  namespace: vitastor-system
+  name: csi-vitastor-provisioner
+  labels:
+    app: csi-metrics
+spec:
+  selector:
+    app: csi-vitastor-provisioner
+  ports:
+    - name: http-metrics
+      port: 8080
+      protocol: TCP
+      targetPort: 8680
+
+---
+kind: Deployment
+apiVersion: apps/v1
+metadata:
+  namespace: vitastor-system
+  name: csi-vitastor-provisioner
+spec:
+  replicas: 3
+  selector:
+    matchLabels:
+      app: csi-vitastor-provisioner
+  template:
+    metadata:
+      namespace: vitastor-system
+      labels:
+        app: csi-vitastor-provisioner
+    spec:
+      affinity:
+        podAntiAffinity:
+          requiredDuringSchedulingIgnoredDuringExecution:
+            - labelSelector:
+                matchExpressions:
+                  - key: app
+                    operator: In
+                    values:
+                      - csi-vitastor-provisioner
+              topologyKey: "kubernetes.io/hostname"
+      serviceAccountName: vitastor-csi-provisioner
+      priorityClassName: system-cluster-critical
+      containers:
+        - name: csi-provisioner
+          image: k8s.gcr.io/sig-storage/csi-provisioner:v2.2.0
+          args:
+            - "--csi-address=$(ADDRESS)"
+            - "--v=5"
+            - "--timeout=150s"
+            - "--retry-interval-start=500ms"
+            - "--leader-election=true"
+            #  set it to true to use topology based provisioning
+            - "--feature-gates=Topology=false"
+            # if fstype is not specified in storageclass, ext4 is default
+            - "--default-fstype=ext4"
+            - "--extra-create-metadata=true"
+          env:
+            - name: ADDRESS
+              value: unix:///csi/csi-provisioner.sock
+          imagePullPolicy: "IfNotPresent"
+          volumeMounts:
+            - name: socket-dir
+              mountPath: /csi
+        - name: csi-snapshotter
+          image: k8s.gcr.io/sig-storage/csi-snapshotter:v4.0.0
+          args:
+            - "--csi-address=$(ADDRESS)"
+            - "--v=5"
+            - "--timeout=150s"
+            - "--leader-election=true"
+          env:
+            - name: ADDRESS
+              value: unix:///csi/csi-provisioner.sock
+          imagePullPolicy: "IfNotPresent"
+          securityContext:
+            privileged: true
+          volumeMounts:
+            - name: socket-dir
+              mountPath: /csi
+        - name: csi-attacher
+          image: k8s.gcr.io/sig-storage/csi-attacher:v3.1.0
+          args:
+            - "--v=5"
+            - "--csi-address=$(ADDRESS)"
+            - "--leader-election=true"
+            - "--retry-interval-start=500ms"
+          env:
+            - name: ADDRESS
+              value: /csi/csi-provisioner.sock
+          imagePullPolicy: "IfNotPresent"
+          volumeMounts:
+            - name: socket-dir
+              mountPath: /csi
+        - name: csi-resizer
+          image: k8s.gcr.io/sig-storage/csi-resizer:v1.1.0
+          args:
+            - "--csi-address=$(ADDRESS)"
+            - "--v=5"
+            - "--timeout=150s"
+            - "--leader-election"
+            - "--retry-interval-start=500ms"
+            - "--handle-volume-inuse-error=false"
+          env:
+            - name: ADDRESS
+              value: unix:///csi/csi-provisioner.sock
+          imagePullPolicy: "IfNotPresent"
+          volumeMounts:
+            - name: socket-dir
+              mountPath: /csi
+        - name: csi-vitastor
+          securityContext:
+            privileged: true
+            capabilities:
+              add: ["SYS_ADMIN"]
+          image: vitalif/vitastor-csi:v0.6.6
+          args:
+            - "--node=$(NODE_ID)"
+            - "--endpoint=$(CSI_ENDPOINT)"
+          env:
+            - name: NODE_ID
+              valueFrom:
+                fieldRef:
+                  fieldPath: spec.nodeName
+            - name: CSI_ENDPOINT
+              value: unix:///csi/csi-provisioner.sock
+          imagePullPolicy: "IfNotPresent"
+          volumeMounts:
+            - name: socket-dir
+              mountPath: /csi
+            - mountPath: /dev
+              name: host-dev
+            - mountPath: /sys
+              name: host-sys
+            - mountPath: /lib/modules
+              name: lib-modules
+              readOnly: true
+            - name: vitastor-config
+              mountPath: /etc/vitastor
+      volumes:
+        - name: host-dev
+          hostPath:
+            path: /dev
+        - name: host-sys
+          hostPath:
+            path: /sys
+        - name: lib-modules
+          hostPath:
+            path: /lib/modules
+        - name: socket-dir
+          emptyDir: {
+            medium: "Memory"
+          }
+        - name: vitastor-config
+          configMap:
+            name: vitastor-config

csi/deploy/008-csi-driver.yaml

@@ -0,0 +1,11 @@
+---
+# if Kubernetes version is less than 1.18 change
+# apiVersion to storage.k8s.io/v1betav1
+apiVersion: storage.k8s.io/v1
+kind: CSIDriver
+metadata:
+  namespace: vitastor-system
+  name: csi.vitastor.io
+spec:
+  attachRequired: true
+  podInfoOnMount: false

csi/deploy/009-storage-class.yaml

@@ -0,0 +1,19 @@
+---
+apiVersion: storage.k8s.io/v1
+kind: StorageClass
+metadata:
+  namespace: vitastor-system
+  name: vitastor
+  annotations:
+    storageclass.kubernetes.io/is-default-class: "true"
+provisioner: csi.vitastor.io
+volumeBindingMode: Immediate
+parameters:
+  etcdVolumePrefix: ""
+  poolId: "1"
+  # you can choose other configuration file if you have it in the config map
+  #configPath: "/etc/vitastor/vitastor.conf"
+  # you can also specify etcdUrl here, maybe to connect to another Vitastor cluster
+  # multiple etcdUrls may be specified, delimited by comma
+  #etcdUrl: "http://192.168.7.2:2379"
+  #etcdPrefix: "/vitastor"

csi/deploy/example-pvc.yaml

@@ -0,0 +1,12 @@
+---
+apiVersion: v1
+kind: PersistentVolumeClaim
+metadata:
+  name: test-vitastor-pvc
+spec:
+  storageClassName: vitastor
+  accessModes:
+    - ReadWriteOnce
+  resources:
+    requests:
+      storage: 10Gi

csi/go.mod

@@ -0,0 +1,35 @@
+module vitastor.io/csi
+
+go 1.15
+
+require (
+	github.com/container-storage-interface/spec v1.4.0
+	github.com/coreos/bbolt v0.0.0-00010101000000-000000000000 // indirect
+	github.com/coreos/etcd v3.3.25+incompatible // indirect
+	github.com/coreos/go-semver v0.3.0 // indirect
+	github.com/coreos/go-systemd v0.0.0-20191104093116-d3cd4ed1dbcf // indirect
+	github.com/coreos/pkg v0.0.0-20180928190104-399ea9e2e55f // indirect
+	github.com/dustin/go-humanize v1.0.0 // indirect
+	github.com/golang/glog v0.0.0-20160126235308-23def4e6c14b
+	github.com/gorilla/websocket v1.4.2 // indirect
+	github.com/grpc-ecosystem/go-grpc-middleware v1.3.0 // indirect
+	github.com/grpc-ecosystem/go-grpc-prometheus v1.2.0 // indirect
+	github.com/grpc-ecosystem/grpc-gateway v1.16.0 // indirect
+	github.com/jonboulle/clockwork v0.2.2 // indirect
+	github.com/kubernetes-csi/csi-lib-utils v0.9.1
+	github.com/soheilhy/cmux v0.1.5 // indirect
+	github.com/tmc/grpc-websocket-proxy v0.0.0-20201229170055-e5319fda7802 // indirect
+	github.com/xiang90/probing v0.0.0-20190116061207-43a291ad63a2 // indirect
+	go.etcd.io/bbolt v0.0.0-00010101000000-000000000000 // indirect
+	go.etcd.io/etcd v3.3.25+incompatible
+	golang.org/x/net v0.0.0-20201202161906-c7110b5ffcbb
+	google.golang.org/grpc v1.33.1
+	k8s.io/klog v1.0.0
+	k8s.io/utils v0.0.0-20210305010621-2afb4311ab10
+)
+
+replace github.com/coreos/bbolt => go.etcd.io/bbolt v1.3.5
+
+replace go.etcd.io/bbolt => github.com/coreos/bbolt v1.3.5
+
+replace google.golang.org/grpc => google.golang.org/grpc v1.25.1

csi/src/config.go

@@ -0,0 +1,22 @@
+// Copyright (c) Vitaliy Filippov, 2019+
+// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
+
+package vitastor
+
+const (
+    vitastorCSIDriverName    = "csi.vitastor.io"
+    vitastorCSIDriverVersion = "0.6.6"
+)
+
+// Config struct fills the parameters of request or user input
+type Config struct
+{
+    Endpoint string
+    NodeID   string
+}
+
+// NewConfig returns config struct to initialize new driver
+func NewConfig() *Config
+{
+    return &Config{}
+}

csi/src/controllerserver.go

@@ -0,0 +1,530 @@
+// Copyright (c) Vitaliy Filippov, 2019+
+// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
+
+package vitastor
+
+import (
+    "context"
+    "encoding/json"
+    "strings"
+    "bytes"
+    "strconv"
+    "time"
+    "fmt"
+    "os"
+    "os/exec"
+    "io/ioutil"
+
+    "github.com/kubernetes-csi/csi-lib-utils/protosanitizer"
+    "k8s.io/klog"
+
+    "google.golang.org/grpc/codes"
+    "google.golang.org/grpc/status"
+
+    "go.etcd.io/etcd/clientv3"
+
+    "github.com/container-storage-interface/spec/lib/go/csi"
+)
+
+const (
+    KB int64 = 1024
+    MB int64 = 1024 * KB
+    GB int64 = 1024 * MB
+    TB int64 = 1024 * GB
+    ETCD_TIMEOUT time.Duration = 15*time.Second
+)
+
+type InodeIndex struct
+{
+    Id uint64 `json:"id"`
+    PoolId uint64 `json:"pool_id"`
+}
+
+type InodeConfig struct
+{
+    Name string `json:"name"`
+    Size uint64 `json:"size,omitempty"`
+    ParentPool uint64 `json:"parent_pool,omitempty"`
+    ParentId uint64 `json:"parent_id,omitempty"`
+    Readonly bool `json:"readonly,omitempty"`
+}
+
+type ControllerServer struct
+{
+    *Driver
+}
+
+// NewControllerServer create new instance controller
+func NewControllerServer(driver *Driver) *ControllerServer
+{
+    return &ControllerServer{
+        Driver: driver,
+    }
+}
+
+func GetConnectionParams(params map[string]string) (map[string]string, []string, string)
+{
+    ctxVars := make(map[string]string)
+    configPath := params["configPath"]
+    if (configPath == "")
+    {
+        configPath = "/etc/vitastor/vitastor.conf"
+    }
+    else
+    {
+        ctxVars["configPath"] = configPath
+    }
+    config := make(map[string]interface{})
+    if configFD, err := os.Open(configPath); err == nil
+    {
+        defer configFD.Close()
+        data, _ := ioutil.ReadAll(configFD)
+        json.Unmarshal(data, &config)
+    }
+    // Try to load prefix & etcd URL from the config
+    var etcdUrl []string
+    if (params["etcdUrl"] != "")
+    {
+        ctxVars["etcdUrl"] = params["etcdUrl"]
+        etcdUrl = strings.Split(params["etcdUrl"], ",")
+    }
+    if (len(etcdUrl) == 0)
+    {
+        switch config["etcd_address"].(type)
+        {
+        case string:
+            etcdUrl = strings.Split(config["etcd_address"].(string), ",")
+        case []string:
+            etcdUrl = config["etcd_address"].([]string)
+        }
+    }
+    etcdPrefix := params["etcdPrefix"]
+    if (etcdPrefix == "")
+    {
+        etcdPrefix, _ = config["etcd_prefix"].(string)
+        if (etcdPrefix == "")
+        {
+            etcdPrefix = "/vitastor"
+        }
+    }
+    else
+    {
+        ctxVars["etcdPrefix"] = etcdPrefix
+    }
+    return ctxVars, etcdUrl, etcdPrefix
+}
+
+// Create the volume
+func (cs *ControllerServer) CreateVolume(ctx context.Context, req *csi.CreateVolumeRequest) (*csi.CreateVolumeResponse, error)
+{
+    klog.Infof("received controller create volume request %+v", protosanitizer.StripSecrets(req))
+    if (req == nil)
+    {
+        return nil, status.Errorf(codes.InvalidArgument, "request cannot be empty")
+    }
+    if (req.GetName() == "")
+    {
+        return nil, status.Error(codes.InvalidArgument, "name is a required field")
+    }
+    volumeCapabilities := req.GetVolumeCapabilities()
+    if (volumeCapabilities == nil)
+    {
+        return nil, status.Error(codes.InvalidArgument, "volume capabilities is a required field")
+    }
+
+    etcdVolumePrefix := req.Parameters["etcdVolumePrefix"]
+    poolId, _ := strconv.ParseUint(req.Parameters["poolId"], 10, 64)
+    if (poolId == 0)
+    {
+        return nil, status.Error(codes.InvalidArgument, "poolId is missing in storage class configuration")
+    }
+
+    volName := etcdVolumePrefix + req.GetName()
+    volSize := 1 * GB
+    if capRange := req.GetCapacityRange(); capRange != nil
+    {
+        volSize = ((capRange.GetRequiredBytes() + MB - 1) / MB) * MB
+    }
+
+    // FIXME: The following should PROBABLY be implemented externally in a management tool
+
+    ctxVars, etcdUrl, etcdPrefix := GetConnectionParams(req.Parameters)
+    if (len(etcdUrl) == 0)
+    {
+        return nil, status.Error(codes.InvalidArgument, "no etcdUrl in storage class configuration and no etcd_address in vitastor.conf")
+    }
+
+    // Connect to etcd
+    cli, err := clientv3.New(clientv3.Config{
+        DialTimeout: ETCD_TIMEOUT,
+        Endpoints: etcdUrl,
+    })
+    if (err != nil)
+    {
+        return nil, status.Error(codes.Internal, "failed to connect to etcd at "+strings.Join(etcdUrl, ",")+": "+err.Error())
+    }
+    defer cli.Close()
+
+    var imageId uint64 = 0
+    for
+    {
+        // Check if the image exists
+        ctx, cancel := context.WithTimeout(context.Background(), ETCD_TIMEOUT)
+        resp, err := cli.Get(ctx, etcdPrefix+"/index/image/"+volName)
+        cancel()
+        if (err != nil)
+        {
+            return nil, status.Error(codes.Internal, "failed to read key from etcd: "+err.Error())
+        }
+        if (len(resp.Kvs) > 0)
+        {
+            kv := resp.Kvs[0]
+            var v InodeIndex
+            err := json.Unmarshal(kv.Value, &v)
+            if (err != nil)
+            {
+                return nil, status.Error(codes.Internal, "invalid /index/image/"+volName+" key in etcd: "+err.Error())
+            }
+            poolId = v.PoolId
+            imageId = v.Id
+            inodeCfgKey := fmt.Sprintf("/config/inode/%d/%d", poolId, imageId)
+            ctx, cancel := context.WithTimeout(context.Background(), ETCD_TIMEOUT)
+            resp, err := cli.Get(ctx, etcdPrefix+inodeCfgKey)
+            cancel()
+            if (err != nil)
+            {
+                return nil, status.Error(codes.Internal, "failed to read key from etcd: "+err.Error())
+            }
+            if (len(resp.Kvs) == 0)
+            {
+                return nil, status.Error(codes.Internal, "missing "+inodeCfgKey+" key in etcd")
+            }
+            var inodeCfg InodeConfig
+            err = json.Unmarshal(resp.Kvs[0].Value, &inodeCfg)
+            if (err != nil)
+            {
+                return nil, status.Error(codes.Internal, "invalid "+inodeCfgKey+" key in etcd: "+err.Error())
+            }
+            if (inodeCfg.Size < uint64(volSize))
+            {
+                return nil, status.Error(codes.Internal, "image "+volName+" is already created, but size is less than expected")
+            }
+        }
+        else
+        {
+            // Find a free ID
+            // Create image metadata in a transaction verifying that the image doesn't exist yet AND ID is still free
+            maxIdKey := fmt.Sprintf("%s/index/maxid/%d", etcdPrefix, poolId)
+            ctx, cancel := context.WithTimeout(context.Background(), ETCD_TIMEOUT)
+            resp, err := cli.Get(ctx, maxIdKey)
+            cancel()
+            if (err != nil)
+            {
+                return nil, status.Error(codes.Internal, "failed to read key from etcd: "+err.Error())
+            }
+            var modRev int64
+            var nextId uint64
+            if (len(resp.Kvs) > 0)
+            {
+                var err error
+                nextId, err = strconv.ParseUint(string(resp.Kvs[0].Value), 10, 64)
+                if (err != nil)
+                {
+                    return nil, status.Error(codes.Internal, maxIdKey+" contains invalid ID")
+                }
+                modRev = resp.Kvs[0].ModRevision
+                nextId++
+            }
+            else
+            {
+                nextId = 1
+            }
+            inodeIdxJson, _ := json.Marshal(InodeIndex{
+                Id: nextId,
+                PoolId: poolId,
+            })
+            inodeCfgJson, _ := json.Marshal(InodeConfig{
+                Name: volName,
+                Size: uint64(volSize),
+            })
+            ctx, cancel = context.WithTimeout(context.Background(), ETCD_TIMEOUT)
+            txnResp, err := cli.Txn(ctx).If(
+                clientv3.Compare(clientv3.ModRevision(fmt.Sprintf("%s/index/maxid/%d", etcdPrefix, poolId)), "=", modRev),
+                clientv3.Compare(clientv3.CreateRevision(fmt.Sprintf("%s/index/image/%s", etcdPrefix, volName)), "=", 0),
+                clientv3.Compare(clientv3.CreateRevision(fmt.Sprintf("%s/config/inode/%d/%d", etcdPrefix, poolId, nextId)), "=", 0),
+            ).Then(
+                clientv3.OpPut(fmt.Sprintf("%s/index/maxid/%d", etcdPrefix, poolId), fmt.Sprintf("%d", nextId)),
+                clientv3.OpPut(fmt.Sprintf("%s/index/image/%s", etcdPrefix, volName), string(inodeIdxJson)),
+                clientv3.OpPut(fmt.Sprintf("%s/config/inode/%d/%d", etcdPrefix, poolId, nextId), string(inodeCfgJson)),
+            ).Commit()
+            cancel()
+            if (err != nil)
+            {
+                return nil, status.Error(codes.Internal, "failed to commit transaction in etcd: "+err.Error())
+            }
+            if (txnResp.Succeeded)
+            {
+                imageId = nextId
+                break
+            }
+            // Start over if the transaction fails
+        }
+    }
+
+    ctxVars["name"] = volName
+    volumeIdJson, _ := json.Marshal(ctxVars)
+    return &csi.CreateVolumeResponse{
+        Volume: &csi.Volume{
+            // Ugly, but VolumeContext isn't passed to DeleteVolume :-(
+            VolumeId: string(volumeIdJson),
+            CapacityBytes: volSize,
+        },
+    }, nil
+}
+
+// DeleteVolume deletes the given volume
+func (cs *ControllerServer) DeleteVolume(ctx context.Context, req *csi.DeleteVolumeRequest) (*csi.DeleteVolumeResponse, error)
+{
+    klog.Infof("received controller delete volume request %+v", protosanitizer.StripSecrets(req))
+    if (req == nil)
+    {
+        return nil, status.Error(codes.InvalidArgument, "request cannot be empty")
+    }
+
+    ctxVars := make(map[string]string)
+    err := json.Unmarshal([]byte(req.VolumeId), &ctxVars)
+    if (err != nil)
+    {
+        return nil, status.Error(codes.Internal, "volume ID not in JSON format")
+    }
+    volName := ctxVars["name"]
+
+    _, etcdUrl, etcdPrefix := GetConnectionParams(ctxVars)
+    if (len(etcdUrl) == 0)
+    {
+        return nil, status.Error(codes.InvalidArgument, "no etcdUrl in storage class configuration and no etcd_address in vitastor.conf")
+    }
+
+    cli, err := clientv3.New(clientv3.Config{
+        DialTimeout: ETCD_TIMEOUT,
+        Endpoints: etcdUrl,
+    })
+    if (err != nil)
+    {
+        return nil, status.Error(codes.Internal, "failed to connect to etcd at "+strings.Join(etcdUrl, ",")+": "+err.Error())
+    }
+    defer cli.Close()
+
+    // Find inode by name
+    ctx, cancel := context.WithTimeout(context.Background(), ETCD_TIMEOUT)
+    resp, err := cli.Get(ctx, etcdPrefix+"/index/image/"+volName)
+    cancel()
+    if (err != nil)
+    {
+        return nil, status.Error(codes.Internal, "failed to read key from etcd: "+err.Error())
+    }
+    if (len(resp.Kvs) == 0)
+    {
+        return nil, status.Error(codes.NotFound, "volume "+volName+" does not exist")
+    }
+    var idx InodeIndex
+    err = json.Unmarshal(resp.Kvs[0].Value, &idx)
+    if (err != nil)
+    {
+        return nil, status.Error(codes.Internal, "invalid /index/image/"+volName+" key in etcd: "+err.Error())
+    }
+
+    // Get inode config
+    inodeCfgKey := fmt.Sprintf("%s/config/inode/%d/%d", etcdPrefix, idx.PoolId, idx.Id)
+    ctx, cancel = context.WithTimeout(context.Background(), ETCD_TIMEOUT)
+    resp, err = cli.Get(ctx, inodeCfgKey)
+    cancel()
+    if (err != nil)
+    {
+        return nil, status.Error(codes.Internal, "failed to read key from etcd: "+err.Error())
+    }
+    if (len(resp.Kvs) == 0)
+    {
+        return nil, status.Error(codes.NotFound, "volume "+volName+" does not exist")
+    }
+    var inodeCfg InodeConfig
+    err = json.Unmarshal(resp.Kvs[0].Value, &inodeCfg)
+    if (err != nil)
+    {
+        return nil, status.Error(codes.Internal, "invalid "+inodeCfgKey+" key in etcd: "+err.Error())
+    }
+
+    // Delete inode data by invoking vitastor-cli
+    args := []string{
+        "rm", "--etcd_address", strings.Join(etcdUrl, ","),
+        "--pool", fmt.Sprintf("%d", idx.PoolId),
+        "--inode", fmt.Sprintf("%d", idx.Id),
+    }
+    if (ctxVars["configPath"] != "")
+    {
+        args = append(args, "--config_path", ctxVars["configPath"])
+    }
+    c := exec.Command("/usr/bin/vitastor-cli", args...)
+    var stderr bytes.Buffer
+    c.Stdout = nil
+    c.Stderr = &stderr
+    err = c.Run()
+    stderrStr := string(stderr.Bytes())
+    if (err != nil)
+    {
+        klog.Errorf("vitastor-cli rm failed: %s, status %s\n", stderrStr, err)
+        return nil, status.Error(codes.Internal, stderrStr+" (status "+err.Error()+")")
+    }
+
+    // Delete inode config in etcd
+    ctx, cancel = context.WithTimeout(context.Background(), ETCD_TIMEOUT)
+    txnResp, err := cli.Txn(ctx).Then(
+        clientv3.OpDelete(fmt.Sprintf("%s/index/image/%s", etcdPrefix, volName)),
+        clientv3.OpDelete(fmt.Sprintf("%s/config/inode/%d/%d", etcdPrefix, idx.PoolId, idx.Id)),
+    ).Commit()
+    cancel()
+    if (err != nil)
+    {
+        return nil, status.Error(codes.Internal, "failed to delete keys in etcd: "+err.Error())
+    }
+    if (!txnResp.Succeeded)
+    {
+        return nil, status.Error(codes.Internal, "failed to delete keys in etcd: transaction failed")
+    }
+
+    return &csi.DeleteVolumeResponse{}, nil
+}
+
+// ControllerPublishVolume return Unimplemented error
+func (cs *ControllerServer) ControllerPublishVolume(ctx context.Context, req *csi.ControllerPublishVolumeRequest) (*csi.ControllerPublishVolumeResponse, error)
+{
+    return nil, status.Error(codes.Unimplemented, "")
+}
+
+// ControllerUnpublishVolume return Unimplemented error
+func (cs *ControllerServer) ControllerUnpublishVolume(ctx context.Context, req *csi.ControllerUnpublishVolumeRequest) (*csi.ControllerUnpublishVolumeResponse, error)
+{
+    return nil, status.Error(codes.Unimplemented, "")
+}
+
+// ValidateVolumeCapabilities checks whether the volume capabilities requested are supported.
+func (cs *ControllerServer) ValidateVolumeCapabilities(ctx context.Context, req *csi.ValidateVolumeCapabilitiesRequest) (*csi.ValidateVolumeCapabilitiesResponse, error)
+{
+    klog.Infof("received controller validate volume capability request %+v", protosanitizer.StripSecrets(req))
+    if (req == nil)
+    {
+        return nil, status.Errorf(codes.InvalidArgument, "request is nil")
+    }
+    volumeID := req.GetVolumeId()
+    if (volumeID == "")
+    {
+        return nil, status.Error(codes.InvalidArgument, "volumeId is nil")
+    }
+    volumeCapabilities := req.GetVolumeCapabilities()
+    if (volumeCapabilities == nil)
+    {
+        return nil, status.Error(codes.InvalidArgument, "volumeCapabilities is nil")
+    }
+
+    var volumeCapabilityAccessModes []*csi.VolumeCapability_AccessMode
+    for _, mode := range []csi.VolumeCapability_AccessMode_Mode{
+        csi.VolumeCapability_AccessMode_SINGLE_NODE_WRITER,
+        csi.VolumeCapability_AccessMode_MULTI_NODE_MULTI_WRITER,
+    } {
+        volumeCapabilityAccessModes = append(volumeCapabilityAccessModes, &csi.VolumeCapability_AccessMode{Mode: mode})
+    }
+
+    capabilitySupport := false
+    for _, capability := range volumeCapabilities
+    {
+        for _, volumeCapabilityAccessMode := range volumeCapabilityAccessModes
+        {
+            if (volumeCapabilityAccessMode.Mode == capability.AccessMode.Mode)
+            {
+                capabilitySupport = true
+            }
+        }
+    }
+
+    if (!capabilitySupport)
+    {
+        return nil, status.Errorf(codes.NotFound, "%v not supported", req.GetVolumeCapabilities())
+    }
+
+    return &csi.ValidateVolumeCapabilitiesResponse{
+        Confirmed: &csi.ValidateVolumeCapabilitiesResponse_Confirmed{
+            VolumeCapabilities: req.VolumeCapabilities,
+        },
+    }, nil
+}
+
+// ListVolumes returns a list of volumes
+func (cs *ControllerServer) ListVolumes(ctx context.Context, req *csi.ListVolumesRequest) (*csi.ListVolumesResponse, error)
+{
+    return nil, status.Error(codes.Unimplemented, "")
+}
+
+// GetCapacity returns the capacity of the storage pool
+func (cs *ControllerServer) GetCapacity(ctx context.Context, req *csi.GetCapacityRequest) (*csi.GetCapacityResponse, error)
+{
+    return nil, status.Error(codes.Unimplemented, "")
+}
+
+// ControllerGetCapabilities returns the capabilities of the controller service.
+func (cs *ControllerServer) ControllerGetCapabilities(ctx context.Context, req *csi.ControllerGetCapabilitiesRequest) (*csi.ControllerGetCapabilitiesResponse, error)
+{
+    functionControllerServerCapabilities := func(cap csi.ControllerServiceCapability_RPC_Type) *csi.ControllerServiceCapability
+    {
+        return &csi.ControllerServiceCapability{
+            Type: &csi.ControllerServiceCapability_Rpc{
+                Rpc: &csi.ControllerServiceCapability_RPC{
+                    Type: cap,
+                },
+            },
+        }
+    }
+
+    var controllerServerCapabilities []*csi.ControllerServiceCapability
+    for _, capability := range []csi.ControllerServiceCapability_RPC_Type{
+        csi.ControllerServiceCapability_RPC_CREATE_DELETE_VOLUME,
+        csi.ControllerServiceCapability_RPC_LIST_VOLUMES,
+        csi.ControllerServiceCapability_RPC_EXPAND_VOLUME,
+        csi.ControllerServiceCapability_RPC_CREATE_DELETE_SNAPSHOT,
+    } {
+        controllerServerCapabilities = append(controllerServerCapabilities, functionControllerServerCapabilities(capability))
+    }
+
+    return &csi.ControllerGetCapabilitiesResponse{
+        Capabilities: controllerServerCapabilities,
+    }, nil
+}
+
+// CreateSnapshot create snapshot of an existing PV
+func (cs *ControllerServer) CreateSnapshot(ctx context.Context, req *csi.CreateSnapshotRequest) (*csi.CreateSnapshotResponse, error)
+{
+    return nil, status.Error(codes.Unimplemented, "")
+}
+
+// DeleteSnapshot delete provided snapshot of a PV
+func (cs *ControllerServer) DeleteSnapshot(ctx context.Context, req *csi.DeleteSnapshotRequest) (*csi.DeleteSnapshotResponse, error)
+{
+    return nil, status.Error(codes.Unimplemented, "")
+}
+
+// ListSnapshots list the snapshots of a PV
+func (cs *ControllerServer) ListSnapshots(ctx context.Context, req *csi.ListSnapshotsRequest) (*csi.ListSnapshotsResponse, error)
+{
+    return nil, status.Error(codes.Unimplemented, "")
+}
+
+// ControllerExpandVolume resizes a volume
+func (cs *ControllerServer) ControllerExpandVolume(ctx context.Context, req *csi.ControllerExpandVolumeRequest) (*csi.ControllerExpandVolumeResponse, error)
+{
+    return nil, status.Error(codes.Unimplemented, "")
+}
+
+// ControllerGetVolume get volume info
+func (cs *ControllerServer) ControllerGetVolume(ctx context.Context, req *csi.ControllerGetVolumeRequest) (*csi.ControllerGetVolumeResponse, error)
+{
+    return nil, status.Error(codes.Unimplemented, "")
+}

csi/src/grpc.go

@@ -0,0 +1,137 @@
+/*
+Copyright 2017 The Kubernetes Authors.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+*/
+
+package vitastor
+
+import (
+    "fmt"
+    "net"
+    "os"
+    "strings"
+    "sync"
+
+    "github.com/golang/glog"
+    "golang.org/x/net/context"
+    "google.golang.org/grpc"
+
+    "github.com/container-storage-interface/spec/lib/go/csi"
+    "github.com/kubernetes-csi/csi-lib-utils/protosanitizer"
+)
+
+// Defines Non blocking GRPC server interfaces
+type NonBlockingGRPCServer interface {
+    // Start services at the endpoint
+    Start(endpoint string, ids csi.IdentityServer, cs csi.ControllerServer, ns csi.NodeServer)
+    // Waits for the service to stop
+    Wait()
+    // Stops the service gracefully
+    Stop()
+    // Stops the service forcefully
+    ForceStop()
+}
+
+func NewNonBlockingGRPCServer() NonBlockingGRPCServer {
+    return &nonBlockingGRPCServer{}
+}
+
+// NonBlocking server
+type nonBlockingGRPCServer struct {
+    wg     sync.WaitGroup
+    server *grpc.Server
+}
+
+func (s *nonBlockingGRPCServer) Start(endpoint string, ids csi.IdentityServer, cs csi.ControllerServer, ns csi.NodeServer) {
+
+    s.wg.Add(1)
+
+    go s.serve(endpoint, ids, cs, ns)
+
+    return
+}
+
+func (s *nonBlockingGRPCServer) Wait() {
+    s.wg.Wait()
+}
+
+func (s *nonBlockingGRPCServer) Stop() {
+    s.server.GracefulStop()
+}
+
+func (s *nonBlockingGRPCServer) ForceStop() {
+    s.server.Stop()
+}
+
+func (s *nonBlockingGRPCServer) serve(endpoint string, ids csi.IdentityServer, cs csi.ControllerServer, ns csi.NodeServer) {
+
+    proto, addr, err := ParseEndpoint(endpoint)
+    if err != nil {
+        glog.Fatal(err.Error())
+    }
+
+    if proto == "unix" {
+        addr = "/" + addr
+        if err := os.Remove(addr); err != nil && !os.IsNotExist(err) {
+            glog.Fatalf("Failed to remove %s, error: %s", addr, err.Error())
+        }
+    }
+
+    listener, err := net.Listen(proto, addr)
+    if err != nil {
+        glog.Fatalf("Failed to listen: %v", err)
+    }
+
+    opts := []grpc.ServerOption{
+        grpc.UnaryInterceptor(logGRPC),
+    }
+    server := grpc.NewServer(opts...)
+    s.server = server
+
+    if ids != nil {
+        csi.RegisterIdentityServer(server, ids)
+    }
+    if cs != nil {
+        csi.RegisterControllerServer(server, cs)
+    }
+    if ns != nil {
+        csi.RegisterNodeServer(server, ns)
+    }
+
+    glog.Infof("Listening for connections on address: %#v", listener.Addr())
+
+    server.Serve(listener)
+}
+
+func ParseEndpoint(ep string) (string, string, error) {
+    if strings.HasPrefix(strings.ToLower(ep), "unix://") || strings.HasPrefix(strings.ToLower(ep), "tcp://") {
+        s := strings.SplitN(ep, "://", 2)
+        if s[1] != "" {
+            return s[0], s[1], nil
+        }
+    }
+    return "", "", fmt.Errorf("Invalid endpoint: %v", ep)
+}
+
+func logGRPC(ctx context.Context, req interface{}, info *grpc.UnaryServerInfo, handler grpc.UnaryHandler) (interface{}, error) {
+    glog.V(3).Infof("GRPC call: %s", info.FullMethod)
+    glog.V(5).Infof("GRPC request: %s", protosanitizer.StripSecrets(req))
+    resp, err := handler(ctx, req)
+    if err != nil {
+        glog.Errorf("GRPC error: %v", err)
+    } else {
+        glog.V(5).Infof("GRPC response: %s", protosanitizer.StripSecrets(resp))
+    }
+    return resp, err
+}

csi/src/identityserver.go

@@ -0,0 +1,60 @@
+// Copyright (c) Vitaliy Filippov, 2019+
+// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
+
+package vitastor
+
+import (
+    "context"
+
+    "github.com/kubernetes-csi/csi-lib-utils/protosanitizer"
+    "k8s.io/klog"
+
+    "github.com/container-storage-interface/spec/lib/go/csi"
+)
+
+// IdentityServer struct of Vitastor CSI driver with supported methods of CSI identity server spec.
+type IdentityServer struct
+{
+    *Driver
+}
+
+// NewIdentityServer create new instance identity
+func NewIdentityServer(driver *Driver) *IdentityServer
+{
+    return &IdentityServer{
+        Driver: driver,
+    }
+}
+
+// GetPluginInfo returns metadata of the plugin
+func (is *IdentityServer) GetPluginInfo(ctx context.Context, req *csi.GetPluginInfoRequest) (*csi.GetPluginInfoResponse, error)
+{
+    klog.Infof("received identity plugin info request %+v", protosanitizer.StripSecrets(req))
+    return &csi.GetPluginInfoResponse{
+        Name:          vitastorCSIDriverName,
+        VendorVersion: vitastorCSIDriverVersion,
+    }, nil
+}
+
+// GetPluginCapabilities returns available capabilities of the plugin
+func (is *IdentityServer) GetPluginCapabilities(ctx context.Context, req *csi.GetPluginCapabilitiesRequest) (*csi.GetPluginCapabilitiesResponse, error)
+{
+    klog.Infof("received identity plugin capabilities request %+v", protosanitizer.StripSecrets(req))
+    return &csi.GetPluginCapabilitiesResponse{
+        Capabilities: []*csi.PluginCapability{
+            {
+                Type: &csi.PluginCapability_Service_{
+                    Service: &csi.PluginCapability_Service{
+                        Type: csi.PluginCapability_Service_CONTROLLER_SERVICE,
+                    },
+                },
+            },
+        },
+    }, nil
+}
+
+// Probe returns the health and readiness of the plugin
+func (is *IdentityServer) Probe(ctx context.Context, req *csi.ProbeRequest) (*csi.ProbeResponse, error)
+{
+    return &csi.ProbeResponse{}, nil
+}

csi/src/nodeserver.go

@@ -0,0 +1,279 @@
+// Copyright (c) Vitaliy Filippov, 2019+
+// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
+
+package vitastor
+
+import (
+    "context"
+    "os"
+    "os/exec"
+    "encoding/json"
+    "strings"
+    "bytes"
+
+    "google.golang.org/grpc/codes"
+    "google.golang.org/grpc/status"
+    "k8s.io/utils/mount"
+    utilexec "k8s.io/utils/exec"
+
+    "github.com/container-storage-interface/spec/lib/go/csi"
+    "github.com/kubernetes-csi/csi-lib-utils/protosanitizer"
+    "k8s.io/klog"
+)
+
+// NodeServer struct of Vitastor CSI driver with supported methods of CSI node server spec.
+type NodeServer struct
+{
+    *Driver
+    mounter mount.Interface
+}
+
+// NewNodeServer create new instance node
+func NewNodeServer(driver *Driver) *NodeServer
+{
+    return &NodeServer{
+        Driver: driver,
+        mounter: mount.New(""),
+    }
+}
+
+// NodeStageVolume mounts the volume to a staging path on the node.
+func (ns *NodeServer) NodeStageVolume(ctx context.Context, req *csi.NodeStageVolumeRequest) (*csi.NodeStageVolumeResponse, error)
+{
+    return &csi.NodeStageVolumeResponse{}, nil
+}
+
+// NodeUnstageVolume unstages the volume from the staging path
+func (ns *NodeServer) NodeUnstageVolume(ctx context.Context, req *csi.NodeUnstageVolumeRequest) (*csi.NodeUnstageVolumeResponse, error)
+{
+    return &csi.NodeUnstageVolumeResponse{}, nil
+}
+
+func Contains(list []string, s string) bool
+{
+    for i := 0; i < len(list); i++
+    {
+        if (list[i] == s)
+        {
+            return true
+        }
+    }
+    return false
+}
+
+// NodePublishVolume mounts the volume mounted to the staging path to the target path
+func (ns *NodeServer) NodePublishVolume(ctx context.Context, req *csi.NodePublishVolumeRequest) (*csi.NodePublishVolumeResponse, error)
+{
+    klog.Infof("received node publish volume request %+v", protosanitizer.StripSecrets(req))
+
+    targetPath := req.GetTargetPath()
+
+    // Check that it's not already mounted
+    free, error := mount.IsNotMountPoint(ns.mounter, targetPath)
+    if (error != nil)
+    {
+        if (os.IsNotExist(error))
+        {
+            error := os.MkdirAll(targetPath, 0777)
+            if (error != nil)
+            {
+                return nil, status.Error(codes.Internal, error.Error())
+            }
+            free = true
+        }
+        else
+        {
+            return nil, status.Error(codes.Internal, error.Error())
+        }
+    }
+    if (!free)
+    {
+        return &csi.NodePublishVolumeResponse{}, nil
+    }
+
+    ctxVars := make(map[string]string)
+    err := json.Unmarshal([]byte(req.VolumeId), &ctxVars)
+    if (err != nil)
+    {
+        return nil, status.Error(codes.Internal, "volume ID not in JSON format")
+    }
+    volName := ctxVars["name"]
+
+    _, etcdUrl, etcdPrefix := GetConnectionParams(ctxVars)
+    if (len(etcdUrl) == 0)
+    {
+        return nil, status.Error(codes.InvalidArgument, "no etcdUrl in storage class configuration and no etcd_address in vitastor.conf")
+    }
+
+    // Map NBD device
+    // FIXME: Check if already mapped
+    args := []string{
+        "map", "--etcd_address", strings.Join(etcdUrl, ","),
+        "--etcd_prefix", etcdPrefix,
+        "--image", volName,
+    };
+    if (ctxVars["configPath"] != "")
+    {
+        args = append(args, "--config_path", ctxVars["configPath"])
+    }
+    if (req.GetReadonly())
+    {
+        args = append(args, "--readonly", "1")
+    }
+    c := exec.Command("/usr/bin/vitastor-nbd", args...)
+    var stdout, stderr bytes.Buffer
+    c.Stdout, c.Stderr = &stdout, &stderr
+    err = c.Run()
+    stdoutStr, stderrStr := string(stdout.Bytes()), string(stderr.Bytes())
+    if (err != nil)
+    {
+        klog.Errorf("vitastor-nbd map failed: %s, status %s\n", stdoutStr+stderrStr, err)
+        return nil, status.Error(codes.Internal, stdoutStr+stderrStr+" (status "+err.Error()+")")
+    }
+    devicePath := strings.TrimSpace(stdoutStr)
+
+    // Check existing format
+    diskMounter := &mount.SafeFormatAndMount{Interface: ns.mounter, Exec: utilexec.New()}
+    existingFormat, err := diskMounter.GetDiskFormat(devicePath)
+    if (err != nil)
+    {
+        klog.Errorf("failed to get disk format for path %s, error: %v", err)
+        // unmap NBD device
+        unmapOut, unmapErr := exec.Command("/usr/bin/vitastor-nbd", "unmap", devicePath).CombinedOutput()
+        if (unmapErr != nil)
+        {
+            klog.Errorf("failed to unmap NBD device %s: %s, error: %v", devicePath, unmapOut, unmapErr)
+        }
+        return nil, err
+    }
+
+    // Format the device (ext4 or xfs)
+    fsType := req.GetVolumeCapability().GetMount().GetFsType()
+    isBlock := req.GetVolumeCapability().GetBlock() != nil
+    opt := req.GetVolumeCapability().GetMount().GetMountFlags()
+    opt = append(opt, "_netdev")
+    if ((req.VolumeCapability.AccessMode.Mode == csi.VolumeCapability_AccessMode_MULTI_NODE_READER_ONLY ||
+        req.VolumeCapability.AccessMode.Mode == csi.VolumeCapability_AccessMode_SINGLE_NODE_READER_ONLY) &&
+        !Contains(opt, "ro"))
+    {
+        opt = append(opt, "ro")
+    }
+    if (fsType == "xfs")
+    {
+        opt = append(opt, "nouuid")
+    }
+    readOnly := Contains(opt, "ro")
+    if (existingFormat == "" && !readOnly)
+    {
+        args := []string{}
+        switch fsType
+        {
+            case "ext4":
+                args = []string{"-m0", "-Enodiscard,lazy_itable_init=1,lazy_journal_init=1", devicePath}
+            case "xfs":
+                args = []string{"-K", devicePath}
+        }
+        if (len(args) > 0)
+        {
+            cmdOut, cmdErr := diskMounter.Exec.Command("mkfs."+fsType, args...).CombinedOutput()
+            if (cmdErr != nil)
+            {
+                klog.Errorf("failed to run mkfs error: %v, output: %v", cmdErr, string(cmdOut))
+                // unmap NBD device
+                unmapOut, unmapErr := exec.Command("/usr/bin/vitastor-nbd", "unmap", devicePath).CombinedOutput()
+                if (unmapErr != nil)
+                {
+                    klog.Errorf("failed to unmap NBD device %s: %s, error: %v", devicePath, unmapOut, unmapErr)
+                }
+                return nil, status.Error(codes.Internal, cmdErr.Error())
+            }
+        }
+    }
+    if (isBlock)
+    {
+        opt = append(opt, "bind")
+        err = diskMounter.Mount(devicePath, targetPath, fsType, opt)
+    }
+    else
+    {
+        err = diskMounter.FormatAndMount(devicePath, targetPath, fsType, opt)
+    }
+    if (err != nil)
+    {
+        klog.Errorf(
+            "failed to mount device path (%s) to path (%s) for volume (%s) error: %s",
+            devicePath, targetPath, volName, err,
+        )
+        // unmap NBD device
+        unmapOut, unmapErr := exec.Command("/usr/bin/vitastor-nbd", "unmap", devicePath).CombinedOutput()
+        if (unmapErr != nil)
+        {
+            klog.Errorf("failed to unmap NBD device %s: %s, error: %v", devicePath, unmapOut, unmapErr)
+        }
+        return nil, status.Error(codes.Internal, err.Error())
+    }
+    return &csi.NodePublishVolumeResponse{}, nil
+}
+
+// NodeUnpublishVolume unmounts the volume from the target path
+func (ns *NodeServer) NodeUnpublishVolume(ctx context.Context, req *csi.NodeUnpublishVolumeRequest) (*csi.NodeUnpublishVolumeResponse, error)
+{
+    klog.Infof("received node unpublish volume request %+v", protosanitizer.StripSecrets(req))
+    targetPath := req.GetTargetPath()
+    devicePath, refCount, err := mount.GetDeviceNameFromMount(ns.mounter, targetPath)
+    if (err != nil)
+    {
+        if (os.IsNotExist(err))
+        {
+            return nil, status.Error(codes.NotFound, "Target path not found")
+        }
+        return nil, status.Error(codes.Internal, err.Error())
+    }
+    if (devicePath == "")
+    {
+        return nil, status.Error(codes.NotFound, "Volume not mounted")
+    }
+    // unmount
+    err = mount.CleanupMountPoint(targetPath, ns.mounter, false)
+    if (err != nil)
+    {
+        return nil, status.Error(codes.Internal, err.Error())
+    }
+    // unmap NBD device
+    if (refCount == 1)
+    {
+        unmapOut, unmapErr := exec.Command("/usr/bin/vitastor-nbd", "unmap", devicePath).CombinedOutput()
+        if (unmapErr != nil)
+        {
+            klog.Errorf("failed to unmap NBD device %s: %s, error: %v", devicePath, unmapOut, unmapErr)
+        }
+    }
+    return &csi.NodeUnpublishVolumeResponse{}, nil
+}
+
+// NodeGetVolumeStats returns volume capacity statistics available for the volume
+func (ns *NodeServer) NodeGetVolumeStats(ctx context.Context, req *csi.NodeGetVolumeStatsRequest) (*csi.NodeGetVolumeStatsResponse, error)
+{
+    return nil, status.Error(codes.Unimplemented, "")
+}
+
+// NodeExpandVolume expanding the file system on the node
+func (ns *NodeServer) NodeExpandVolume(ctx context.Context, req *csi.NodeExpandVolumeRequest) (*csi.NodeExpandVolumeResponse, error)
+{
+    return nil, status.Error(codes.Unimplemented, "")
+}
+
+// NodeGetCapabilities returns the supported capabilities of the node server
+func (ns *NodeServer) NodeGetCapabilities(ctx context.Context, req *csi.NodeGetCapabilitiesRequest) (*csi.NodeGetCapabilitiesResponse, error)
+{
+    return &csi.NodeGetCapabilitiesResponse{}, nil
+}
+
+// NodeGetInfo returns NodeGetInfoResponse for CO.
+func (ns *NodeServer) NodeGetInfo(ctx context.Context, req *csi.NodeGetInfoRequest) (*csi.NodeGetInfoResponse, error)
+{
+    klog.Infof("received node get info request %+v", protosanitizer.StripSecrets(req))
+    return &csi.NodeGetInfoResponse{
+        NodeId: ns.NodeID,
+    }, nil
+}

csi/src/server.go

@@ -0,0 +1,36 @@
+// Copyright (c) Vitaliy Filippov, 2019+
+// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
+
+package vitastor
+
+import (
+    "k8s.io/klog"
+)
+
+type Driver struct
+{
+    *Config
+}
+
+// NewDriver create new instance driver
+func NewDriver(config *Config) (*Driver, error)
+{
+    if (config == nil)
+    {
+        klog.Errorf("Vitastor CSI driver initialization failed")
+        return nil, nil
+    }
+    driver := &Driver{
+        Config: config,
+    }
+    klog.Infof("Vitastor CSI driver initialized")
+    return driver, nil
+}
+
+// Start server
+func (driver *Driver) Run()
+{
+    server := NewNonBlockingGRPCServer()
+    server.Start(driver.Endpoint, NewIdentityServer(driver), NewControllerServer(driver), NewNodeServer(driver))
+    server.Wait()
+}

csi/vitastor-csi.go

@@ -0,0 +1,39 @@
+// Copyright (c) Vitaliy Filippov, 2019+
+// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
+
+package main
+
+import (
+    "flag"
+    "fmt"
+    "os"
+    "k8s.io/klog"
+    "vitastor.io/csi/src"
+)
+
+func main()
+{
+    var config = vitastor.NewConfig()
+    flag.StringVar(&config.Endpoint, "endpoint", "", "CSI endpoint")
+    flag.StringVar(&config.NodeID, "node", "", "Node ID")
+    flag.Parse()
+    if (config.Endpoint == "")
+    {
+        config.Endpoint = os.Getenv("CSI_ENDPOINT")
+    }
+    if (config.NodeID == "")
+    {
+        config.NodeID = os.Getenv("NODE_ID")
+    }
+    if (config.Endpoint == "" && config.NodeID == "")
+    {
+        fmt.Fprintf(os.Stderr, "Please set -endpoint and -node / CSI_ENDPOINT & NODE_ID env vars\n")
+        os.Exit(1)
+    }
+    drv, err := vitastor.NewDriver(config)
+    if (err != nil)
+    {
+        klog.Fatalln(err)
+    }
+    drv.Run()
+}

debian/changelog

@@ -1,4 +1,4 @@
-vitastor (0.6.3-1) unstable; urgency=medium
+vitastor (0.6.6-1) unstable; urgency=medium
 
   * RDMA support
   * Bugfixes

debian/patched-qemu.Dockerfile

@@ -11,6 +11,10 @@ RUN if [ "$REL" = "buster" ]; then \
         echo 'Package: *' >> /etc/apt/preferences; \
         echo 'Pin: release a=buster-backports' >> /etc/apt/preferences; \
         echo 'Pin-Priority: 500' >> /etc/apt/preferences; \
+        echo >> /etc/apt/preferences; \
+        echo 'Package: libglvnd* libgles* libglx* libgl1 libegl* libopengl* mesa*' >> /etc/apt/preferences; \
+        echo 'Pin: release a=buster-backports' >> /etc/apt/preferences; \
+        echo 'Pin-Priority: 50' >> /etc/apt/preferences; \
     fi; \
     grep '^deb ' /etc/apt/sources.list | perl -pe 's/^deb/deb-src/' >> /etc/apt/sources.list; \
     echo 'APT::Install-Recommends false;' >> /etc/apt/apt.conf; \
@@ -20,20 +24,22 @@ RUN apt-get update
 RUN apt-get -y install qemu fio liburing1 liburing-dev libgoogle-perftools-dev devscripts
 RUN apt-get -y build-dep qemu
 RUN apt-get -y build-dep fio
+# To build a custom version
+#RUN cp /root/packages/qemu-orig/* /root
 RUN apt-get --download-only source qemu
 RUN apt-get --download-only source fio
 
-ADD qemu-5.0-vitastor.patch qemu-5.1-vitastor.patch /root/vitastor/
+ADD patches/qemu-5.0-vitastor.patch patches/qemu-5.1-vitastor.patch /root/vitastor/patches/
 RUN set -e; \
     mkdir -p /root/packages/qemu-$REL; \
     rm -rf /root/packages/qemu-$REL/*; \
     cd /root/packages/qemu-$REL; \
     dpkg-source -x /root/qemu*.dsc; \
     if [ -d /root/packages/qemu-$REL/qemu-5.0 ]; then \
-        cp /root/vitastor/qemu-5.0-vitastor.patch /root/packages/qemu-$REL/qemu-5.0/debian/patches; \
+        cp /root/vitastor/patches/qemu-5.0-vitastor.patch /root/packages/qemu-$REL/qemu-5.0/debian/patches; \
         echo qemu-5.0-vitastor.patch >> /root/packages/qemu-$REL/qemu-5.0/debian/patches/series; \
     else \
-        cp /root/vitastor/qemu-5.1-vitastor.patch /root/packages/qemu-$REL/qemu-*/debian/patches; \
+        cp /root/vitastor/patches/qemu-5.1-vitastor.patch /root/packages/qemu-$REL/qemu-*/debian/patches; \
         P=`ls -d /root/packages/qemu-$REL/qemu-*/debian/patches`; \
         echo qemu-5.1-vitastor.patch >> $P/series; \
     fi; \

debian/vitastor.Dockerfile

@@ -40,10 +40,10 @@ RUN set -e -x; \
     mkdir -p /root/packages/vitastor-$REL; \
     rm -rf /root/packages/vitastor-$REL/*; \
     cd /root/packages/vitastor-$REL; \
-    cp -r /root/vitastor vitastor-0.6.3; \
-    ln -s /root/packages/qemu-$REL/qemu-*/ vitastor-0.6.3/qemu; \
-    ln -s /root/fio-build/fio-*/ vitastor-0.6.3/fio; \
-    cd vitastor-0.6.3; \
+    cp -r /root/vitastor vitastor-0.6.6; \
+    ln -s /root/packages/qemu-$REL/qemu-*/ vitastor-0.6.6/qemu; \
+    ln -s /root/fio-build/fio-*/ vitastor-0.6.6/fio; \
+    cd vitastor-0.6.6; \
     FIO=$(head -n1 fio/debian/changelog | perl -pe 's/^.*\((.*?)\).*$/$1/'); \
     QEMU=$(head -n1 qemu/debian/changelog | perl -pe 's/^.*\((.*?)\).*$/$1/'); \
     sh copy-qemu-includes.sh; \
@@ -59,8 +59,8 @@ RUN set -e -x; \
     echo "dep:fio=$FIO" > debian/substvars; \
     echo "dep:qemu=$QEMU" >> debian/substvars; \
     cd /root/packages/vitastor-$REL; \
-    tar --sort=name --mtime='2020-01-01' --owner=0 --group=0 --exclude=debian -cJf vitastor_0.6.3.orig.tar.xz vitastor-0.6.3; \
-    cd vitastor-0.6.3; \
+    tar --sort=name --mtime='2020-01-01' --owner=0 --group=0 --exclude=debian -cJf vitastor_0.6.6.orig.tar.xz vitastor-0.6.6; \
+    cd vitastor-0.6.6; \
     V=$(head -n1 debian/changelog | perl -pe 's/^.*\((.*?)\).*$/$1/'); \
     DEBFULLNAME="Vitaliy Filippov <vitalif@yourcmc.ru>" dch -D $REL -v "$V""$REL" "Rebuild for $REL"; \
     DEB_BUILD_OPTIONS=nocheck dpkg-buildpackage --jobs=auto -sa; \

docker/Dockerfile

@@ -0,0 +1,9 @@
+# Build Docker image with Vitastor packages
+
+FROM debian:bullseye
+
+ADD vitastor.list /etc/apt/sources.list.d
+ADD vitastor.gpg /etc/apt/trusted.gpg.d
+ADD vitastor.pref /etc/apt/preferences.d
+ADD apt.conf /etc/apt/
+RUN apt-get update && apt-get -y install vitastor qemu-system-x86 qemu-system-common && apt-get clean

docker/apt.conf

@@ -0,0 +1 @@
+APT::Install-Recommends false;

docker/vitastor.list

@@ -0,0 +1 @@
+deb http://vitastor.io/debian bullseye main

docker/vitastor.pref

@@ -0,0 +1,3 @@
+Package: *
+Pin: origin "vitastor.io"
+Pin-Priority: 1000

mon/lp-optimizer.js

@@ -244,6 +244,7 @@ async function optimize_change({ prev_pgs: prev_int_pgs, osd_tree, pg_size = 3,
     {
         return null;
     }
+    // FIXME: use parity_chunks with parity_space instead of pg_minsize
     const pg_effsize = Math.min(pg_minsize, Object.keys(osd_tree).length)
         + Math.max(0, Math.min(pg_size, Object.keys(osd_tree).length) - pg_minsize) * parity_space;
     const pg_count = prev_int_pgs.length;

mon/mon.js

@@ -36,13 +36,15 @@ const etcd_allow = new RegExp('^'+[
     'history/last_clean_pgs',
     'inode/stats/[1-9]\\d*/[1-9]\\d*',
     'stats',
+    'index/image/.*',
+    'index/maxid/[1-9]\\d*',
 ].join('$|^')+'$');
 
 const etcd_tree = {
     config: {
         /* global: {
             // WARNING: NOT ALL OF THESE ARE ACTUALLY CONFIGURABLE HERE
-            // THIS IS JUST A POOR'S MAN CONFIG DOCUMENTATION
+            // THIS IS JUST A POOR MAN'S CONFIG DOCUMENTATION
             // etcd connection
             config_path: "/etc/vitastor/vitastor.conf",
             etcd_address: "10.0.115.10:2379/v3",
@@ -257,14 +259,26 @@ const etcd_tree = {
     },
     inode: {
         stats: {
-            /* <inode_t>: {
-                raw_used: uint64_t, // raw used bytes on OSDs
-                read: { count: uint64_t, usec: uint64_t, bytes: uint64_t },
-                write: { count: uint64_t, usec: uint64_t, bytes: uint64_t },
-                delete: { count: uint64_t, usec: uint64_t, bytes: uint64_t },
+            /* <pool_id>: {
+                <inode_t>: {
+                    raw_used: uint64_t, // raw used bytes on OSDs
+                    read: { count: uint64_t, usec: uint64_t, bytes: uint64_t },
+                    write: { count: uint64_t, usec: uint64_t, bytes: uint64_t },
+                    delete: { count: uint64_t, usec: uint64_t, bytes: uint64_t },
+                },
             }, */
         },
     },
+    pool: {
+        stats: {
+            /* <pool_id>: {
+                used_raw_tb: float, // used raw space in the pool
+                total_raw_tb: float, // maximum amount of space in the pool
+                raw_to_usable: float, // raw to usable ratio
+                space_efficiency: float, // 0..1
+            } */
+        },
+    },
     stats: {
         /* op_stats: {
             <string>: { count: uint64_t, usec: uint64_t, bytes: uint64_t },
@@ -287,6 +301,17 @@ const etcd_tree = {
     history: {
         last_clean_pgs: {},
     },
+    index: {
+        image: {
+            /* <name>: {
+                id: uint64_t,
+                pool_id: uint64_t,
+            }, */
+        },
+        maxid: {
+            /* <pool_id>: uint64_t, */
+        },
+    },
 };
 
 // FIXME Split into several files
@@ -361,6 +386,11 @@ class Mon
         {
             this.config.mon_stats_timeout = 100;
         }
+        this.config.mon_stats_interval = Number(this.config.mon_stats_interval) || 5000;
+        if (this.config.mon_stats_interval < 100)
+        {
+            this.config.mon_stats_interval = 100;
+        }
         // After this number of seconds, a dead OSD will be removed from PG distribution
         this.config.osd_out_time = Number(this.config.osd_out_time) || 0;
         if (!this.config.osd_out_time)
@@ -1025,6 +1055,17 @@ class Mon
                     } });
                 }
                 LPOptimizer.print_change_stats(optimize_result);
+                const pg_effsize = Math.min(pool_cfg.pg_size, Object.keys(pool_tree).length);
+                this.state.pool.stats[pool_id] = {
+                    used_raw_tb: (this.state.pool.stats[pool_id]||{}).used_raw_tb || 0,
+                    total_raw_tb: optimize_result.space,
+                    raw_to_usable: pg_effsize / (pool_cfg.pg_size - (pool_cfg.parity_chunks||0)),
+                    space_efficiency: optimize_result.space/(optimize_result.total_space||1),
+                };
+                etcd_request.success.push({ requestPut: {
+                    key: b64(this.etcd_prefix+'/pool/stats/'+pool_id),
+                    value: b64(JSON.stringify(this.state.pool.stats[pool_id])),
+                } });
                 this.save_new_pgs_txn(etcd_request, pool_id, up_osds, real_prev_pgs, optimize_result.int_pgs, pg_history);
             }
             this.state.config.pgs.hash = tree_hash;
@@ -1131,7 +1172,7 @@ class Mon
         }, this.config.mon_change_timeout || 1000);
     }
 
-    sum_stats()
+    sum_op_stats()
     {
         const op_stats = {}, subop_stats = {}, recovery_stats = {};
         for (const osd in this.state.osd.stats)
@@ -1192,18 +1233,31 @@ class Mon
             write: { count: 0n, usec: 0n, bytes: 0n },
             delete: { count: 0n, usec: 0n, bytes: 0n },
         });
+        for (const pool_id in this.state.config.pools)
+        {
+            this.state.pool.stats[pool_id] = this.state.pool.stats[pool_id] || {};
+            this.state.pool.stats[pool_id].used_raw_tb = 0n;
+        }
         for (const osd_num in this.state.osd.space)
         {
             for (const pool_id in this.state.osd.space[osd_num])
             {
+                this.state.pool.stats[pool_id] = this.state.pool.stats[pool_id] || { used_raw_tb: 0n };
                 inode_stats[pool_id] = inode_stats[pool_id] || {};
                 for (const inode_num in this.state.osd.space[osd_num][pool_id])
                 {
+                    const u = BigInt(this.state.osd.space[osd_num][pool_id][inode_num]||0);
                     inode_stats[pool_id][inode_num] = inode_stats[pool_id][inode_num] || inode_stub();
-                    inode_stats[pool_id][inode_num].raw_used += BigInt(this.state.osd.space[osd_num][pool_id][inode_num]||0);
+                    inode_stats[pool_id][inode_num].raw_used += u;
+                    this.state.pool.stats[pool_id].used_raw_tb += u;
                 }
             }
         }
+        for (const pool_id in this.state.config.pools)
+        {
+            const used = this.state.pool.stats[pool_id].used_raw_tb;
+            this.state.pool.stats[pool_id].used_raw_tb = Number(used)/1024/1024/1024/1024;
+        }
         for (const osd_num in this.state.osd.inodestats)
         {
             const ist = this.state.osd.inodestats[osd_num];
@@ -1275,7 +1329,7 @@ class Mon
     async update_total_stats()
     {
         const txn = [];
-        const stats = this.sum_stats();
+        const stats = this.sum_op_stats();
         const object_counts = this.sum_object_counts();
         const inode_stats = this.sum_inode_stats();
         this.fix_stat_overflows(stats, (this.prev_stats = this.prev_stats || {}));
@@ -1294,6 +1348,13 @@ class Mon
                 } });
             }
         }
+        for (const pool_id in this.state.pool.stats)
+        {
+            txn.push({ requestPut: {
+                key: b64(this.etcd_prefix+'/pool/stats/'+pool_id),
+                value: b64(JSON.stringify(this.state.pool.stats[pool_id])),
+            } });
+        }
         if (txn.length)
         {
             await this.etcd_call('/kv/txn', { success: txn }, this.config.etcd_mon_timeout, 0);
@@ -1307,11 +1368,17 @@ class Mon
             clearTimeout(this.stats_timer);
             this.stats_timer = null;
         }
+        let sleep = (this.stats_update_next||0) - Date.now();
+        if (sleep < this.config.mon_stats_timeout)
+        {
+            sleep = this.config.mon_stats_timeout;
+        }
         this.stats_timer = setTimeout(() =>
         {
             this.stats_timer = null;
+            this.stats_update_next = Date.now() + this.config.mon_stats_interval;
             this.update_total_stats().catch(console.error);
-        }, this.config.mon_stats_timeout || 1000);
+        }, sleep);
     }
 
     parse_kv(kv)

patches/cinder-vitastor.py

@@ -0,0 +1,948 @@
+# Vitastor Driver for OpenStack Cinder
+#
+# --------------------------------------------
+# Install as cinder/volume/drivers/vitastor.py
+# --------------------------------------------
+#
+# Copyright 2020 Vitaliy Filippov
+#
+# Licensed under the Apache License, Version 2.0 (the "License"); you may
+# not use this file except in compliance with the License. You may obtain
+# a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
+# WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
+# License for the specific language governing permissions and limitations
+# under the License.
+"""Cinder Vitastor Driver"""
+
+import binascii
+import base64
+import errno
+import json
+import math
+import os
+import tempfile
+
+from castellan import key_manager
+from oslo_config import cfg
+from oslo_log import log as logging
+from oslo_service import loopingcall
+from oslo_concurrency import processutils
+from oslo_utils import encodeutils
+from oslo_utils import excutils
+from oslo_utils import fileutils
+from oslo_utils import units
+import six
+from six.moves.urllib import request
+
+from cinder import exception
+from cinder.i18n import _
+from cinder.image import image_utils
+from cinder import interface
+from cinder import objects
+from cinder.objects import fields
+from cinder import utils
+from cinder.volume import configuration
+from cinder.volume import driver
+from cinder.volume import volume_utils
+
+VERSION = '0.6.6'
+
+LOG = logging.getLogger(__name__)
+
+VITASTOR_OPTS = [
+    cfg.StrOpt(
+        'vitastor_config_path',
+        default='/etc/vitastor/vitastor.conf',
+        help='Vitastor configuration file path'
+    ),
+    cfg.StrOpt(
+        'vitastor_etcd_address',
+        default='',
+        help='Vitastor etcd address(es)'),
+    cfg.StrOpt(
+        'vitastor_etcd_prefix',
+        default='/vitastor',
+        help='Vitastor etcd prefix'
+    ),
+    cfg.StrOpt(
+        'vitastor_pool_id',
+        default='',
+        help='Vitastor pool ID to use for volumes'
+    ),
+    # FIXME exclusive_cinder_pool ?
+]
+
+CONF = cfg.CONF
+CONF.register_opts(VITASTOR_OPTS, group = configuration.SHARED_CONF_GROUP)
+
+class VitastorDriverException(exception.VolumeDriverException):
+    message = _("Vitastor Cinder driver failure: %(reason)s")
+
+@interface.volumedriver
+class VitastorDriver(driver.CloneableImageVD,
+    driver.ManageableVD, driver.ManageableSnapshotsVD,
+    driver.BaseVD):
+    """Implements Vitastor volume commands."""
+
+    cfg = {}
+    _etcd_urls = []
+
+    def __init__(self, active_backend_id = None, *args, **kwargs):
+        super(VitastorDriver, self).__init__(*args, **kwargs)
+        self.configuration.append_config_values(VITASTOR_OPTS)
+
+    @classmethod
+    def get_driver_options(cls):
+        additional_opts = cls._get_oslo_driver_opts(
+            'reserved_percentage',
+            'max_over_subscription_ratio',
+            'volume_dd_blocksize'
+        )
+        return VITASTOR_OPTS + additional_opts
+
+    def do_setup(self, context):
+        """Performs initialization steps that could raise exceptions."""
+        super(VitastorDriver, self).do_setup(context)
+        # Make sure configuration is in UTF-8
+        for attr in [ 'config_path', 'etcd_address', 'etcd_prefix', 'pool_id' ]:
+            val = self.configuration.safe_get('vitastor_'+attr)
+            if val is not None:
+                self.cfg[attr] = utils.convert_str(val)
+        self.cfg = self._load_config(self.cfg)
+
+    def _load_config(self, cfg):
+        # Try to load configuration file
+        try:
+            f = open(cfg['config_path'] or '/etc/vitastor/vitastor.conf')
+            conf = json.loads(f.read())
+            f.close()
+            for k in conf:
+                cfg[k] = cfg.get(k, conf[k])
+        except:
+            pass
+        if isinstance(cfg['etcd_address'], str):
+            cfg['etcd_address'] = cfg['etcd_address'].split(',')
+        # Sanitize etcd URLs
+        for i, etcd_url in enumerate(cfg['etcd_address']):
+            ssl = False
+            if etcd_url.lower().startswith('http://'):
+                etcd_url = etcd_url[7:]
+            elif etcd_url.lower().startswith('https://'):
+                etcd_url = etcd_url[8:]
+                ssl = True
+            if etcd_url.find('/') < 0:
+                etcd_url += '/v3'
+            if ssl:
+                etcd_url = 'https://'+etcd_url
+            else:
+                etcd_url = 'http://'+etcd_url
+            cfg['etcd_address'][i] = etcd_url
+        return cfg
+
+    def check_for_setup_error(self):
+        """Returns an error if prerequisites aren't met."""
+
+    def _encode_etcd_key(self, key):
+        if not isinstance(key, bytes):
+            key = str(key).encode('utf-8')
+        return base64.b64encode(self.cfg['etcd_prefix'].encode('utf-8')+b'/'+key).decode('utf-8')
+
+    def _encode_etcd_value(self, value):
+        if not isinstance(value, bytes):
+            value = str(value).encode('utf-8')
+        return base64.b64encode(value).decode('utf-8')
+
+    def _encode_etcd_requests(self, obj):
+        for v in obj:
+            for rt in v:
+                if 'key' in v[rt]:
+                    v[rt]['key'] = self._encode_etcd_key(v[rt]['key'])
+                if 'range_end' in v[rt]:
+                    v[rt]['range_end'] = self._encode_etcd_key(v[rt]['range_end'])
+                if 'value' in v[rt]:
+                    v[rt]['value'] = self._encode_etcd_value(v[rt]['value'])
+
+    def _etcd_txn(self, params):
+        if 'compare' in params:
+            for v in params['compare']:
+                if 'key' in v:
+                    v['key'] = self._encode_etcd_key(v['key'])
+        if 'failure' in params:
+            self._encode_etcd_requests(params['failure'])
+        if 'success' in params:
+            self._encode_etcd_requests(params['success'])
+        body = json.dumps(params).encode('utf-8')
+        headers = {
+            'Content-Type': 'application/json'
+        }
+        err = None
+        for etcd_url in self.cfg['etcd_address']:
+            try:
+                resp = request.urlopen(request.Request(etcd_url+'/kv/txn', body, headers), timeout = 5)
+                data = json.loads(resp.read())
+                if 'responses' not in data:
+                    data['responses'] = []
+                for i, resp in enumerate(data['responses']):
+                    if 'response_range' in resp:
+                        if 'kvs' not in resp['response_range']:
+                            resp['response_range']['kvs'] = []
+                        for kv in resp['response_range']['kvs']:
+                            kv['key'] = base64.b64decode(kv['key'].encode('utf-8')).decode('utf-8')
+                            if kv['key'].startswith(self.cfg['etcd_prefix']+'/'):
+                                kv['key'] = kv['key'][len(self.cfg['etcd_prefix'])+1 : ]
+                            kv['value'] = json.loads(base64.b64decode(kv['value'].encode('utf-8')))
+                    if len(resp.keys()) != 1:
+                        LOG.exception('unknown responses['+str(i)+'] format: '+json.dumps(resp))
+                    else:
+                        resp = data['responses'][i] = resp[list(resp.keys())[0]]
+                return data
+            except Exception as e:
+                LOG.exception('error calling etcd transaction: '+body.decode('utf-8')+'\nerror: '+str(e))
+                err = e
+        raise err
+
+    def _etcd_foreach(self, prefix, add_fn):
+        total = 0
+        batch = 1000
+        begin = prefix+'/'
+        while True:
+            resp = self._etcd_txn({ 'success': [
+                { 'request_range': {
+                    'key': begin,
+                    'range_end': prefix+'0',
+                    'limit': batch+1,
+                } },
+            ] })
+            i = 0
+            while i < batch and i < len(resp['responses'][0]['kvs']):
+                kv = resp['responses'][0]['kvs'][i]
+                add_fn(kv)
+                i += 1
+            if len(resp['responses'][0]['kvs']) <= batch:
+                break
+            begin = resp['responses'][0]['kvs'][batch]['key']
+        return total
+
+    def _update_volume_stats(self):
+        location_info = json.dumps({
+            'config': self.configuration.vitastor_config_path,
+            'etcd_address': self.configuration.vitastor_etcd_address,
+            'etcd_prefix': self.configuration.vitastor_etcd_prefix,
+            'pool_id': self.configuration.vitastor_pool_id,
+        })
+
+        stats = {
+            'vendor_name': 'Vitastor',
+            'driver_version': self.VERSION,
+            'storage_protocol': 'vitastor',
+            'total_capacity_gb': 'unknown',
+            'free_capacity_gb': 'unknown',
+            # FIXME check if safe_get is required
+            'reserved_percentage': self.configuration.safe_get('reserved_percentage'),
+            'multiattach': True,
+            'thin_provisioning_support': True,
+            'max_over_subscription_ratio': self.configuration.safe_get('max_over_subscription_ratio'),
+            'location_info': location_info,
+            'backend_state': 'down',
+            'volume_backend_name': self.configuration.safe_get('volume_backend_name') or 'vitastor',
+            'replication_enabled': False,
+        }
+
+        try:
+            pool_stats = self._etcd_txn({ 'success': [
+                { 'request_range': { 'key': 'pool/stats/'+str(self.cfg['pool_id']) } }
+            ] })
+            total_provisioned = 0
+            def add_total(kv):
+                nonlocal total_provisioned
+                if kv['key'].find('@') >= 0:
+                    total_provisioned += kv['value']['size']
+            self._etcd_foreach('config/inode/'+str(self.cfg['pool_id']), lambda kv: add_total(kv))
+            stats['provisioned_capacity_gb'] = round(total_provisioned/1024.0/1024.0/1024.0, 2)
+            pool_stats = pool_stats['responses'][0]['kvs']
+            if len(pool_stats):
+                pool_stats = pool_stats[0]
+                stats['free_capacity_gb'] = round(1024.0*(pool_stats['total_raw_tb']-pool_stats['used_raw_tb'])/pool_stats['raw_to_usable'], 2)
+                stats['total_capacity_gb'] = round(1024.0*pool_stats['total_raw_tb'], 2)
+            stats['backend_state'] = 'up'
+        except Exception as e:
+            # just log and return unknown capacities
+            LOG.exception('error getting vitastor pool stats: '+str(e))
+
+        self._stats = stats
+
+    def _next_id(self, resp):
+        if len(resp['kvs']) == 0:
+            return (1, 0)
+        else:
+            return (1 + resp['kvs'][0]['value'], resp['kvs'][0]['mod_revision'])
+
+    def create_volume(self, volume):
+        """Creates a logical volume."""
+
+        size = int(volume.size) * units.Gi
+        # FIXME: Check if convert_str is really required
+        vol_name = utils.convert_str(volume.name)
+        if vol_name.find('@') >= 0 or vol_name.find('/') >= 0:
+            raise exception.VolumeBackendAPIException(data = '@ and / are forbidden in volume and snapshot names')
+
+        LOG.debug("creating volume '%s'", vol_name)
+
+        self._create_image(vol_name, { 'size': size })
+
+        if volume.encryption_key_id:
+            self._create_encrypted_volume(volume, volume.obj_context)
+
+        volume_update = {}
+        return volume_update
+
+    def _create_encrypted_volume(self, volume, context):
+        """Create a new LUKS encrypted image directly in Vitastor."""
+        vol_name = utils.convert_str(volume.name)
+        f, opts = self._encrypt_opts(volume, context)
+        # FIXME: Check if it works at all :-)
+        self._execute(
+            'qemu-img', 'convert', '-f', 'luks', *opts,
+            'vitastor:image='+vol_name.replace(':', '\\:')+self._qemu_args(),
+            '%sM' % (volume.size * 1024)
+        )
+        f.close()
+
+    def _encrypt_opts(self, volume, context):
+        encryption = volume_utils.check_encryption_provider(self.db, volume, context)
+        # Fetch the key associated with the volume and decode the passphrase
+        keymgr = key_manager.API(CONF)
+        key = keymgr.get(context, encryption['encryption_key_id'])
+        passphrase = binascii.hexlify(key.get_encoded()).decode('utf-8')
+        # Decode the dm-crypt style cipher spec into something qemu-img can use
+        cipher_spec = image_utils.decode_cipher(encryption['cipher'], encryption['key_size'])
+        tmp_dir = volume_utils.image_conversion_dir()
+        f = tempfile.NamedTemporaryFile(prefix = 'luks_', dir = tmp_dir)
+        f.write(passphrase)
+        f.flush()
+        return (f, [
+            '--object', 'secret,id=luks_sec,format=raw,file=%(passfile)s' % {'passfile': f.name},
+            '-o', 'key-secret=luks_sec,cipher-alg=%(cipher_alg)s,cipher-mode=%(cipher_mode)s,ivgen-alg=%(ivgen_alg)s' % cipher_spec,
+        ])
+
+    def create_snapshot(self, snapshot):
+        """Creates a volume snapshot."""
+
+        vol_name = utils.convert_str(snapshot.volume_name)
+        snap_name = utils.convert_str(snapshot.name)
+        if snap_name.find('@') >= 0 or snap_name.find('/') >= 0:
+            raise exception.VolumeBackendAPIException(data = '@ and / are forbidden in volume and snapshot names')
+        self._create_snapshot(vol_name, vol_name+'@'+snap_name)
+
+    def snapshot_revert_use_temp_snapshot(self):
+        """Disable the use of a temporary snapshot on revert."""
+        return False
+
+    def revert_to_snapshot(self, context, volume, snapshot):
+        """Revert a volume to a given snapshot."""
+
+        # FIXME Delete the image, then recreate it from the snapshot
+
+    def delete_snapshot(self, snapshot):
+        """Deletes a snapshot."""
+
+        vol_name = utils.convert_str(snapshot.volume_name)
+        snap_name = utils.convert_str(snapshot.name)
+
+        # Find the snapshot
+        resp = self._etcd_txn({ 'success': [
+            { 'request_range': { 'key': 'index/image/'+vol_name+'@'+snap_name } },
+        ] })
+        if len(resp['responses'][0]['kvs']) == 0:
+            raise exception.SnapshotNotFound(snapshot_id = snap_name)
+        inode_id = int(resp['responses'][0]['kvs'][0]['value']['id'])
+        pool_id = int(resp['responses'][0]['kvs'][0]['value']['pool_id'])
+        parents = {}
+        parents[(pool_id << 48) | (inode_id & 0xffffffffffff)] = True
+
+        # Check if there are child volumes
+        children = self._child_count(parents)
+        if children > 0:
+            raise exception.SnapshotIsBusy(snapshot_name = snap_name)
+
+        # FIXME: We can't delete snapshots because we can't merge layers yet
+        raise exception.VolumeBackendAPIException(data = 'Snapshot delete (layer merge) is not implemented yet')
+
+    def _child_count(self, parents):
+        children = 0
+        def add_child(kv):
+            nonlocal children
+            children += self._check_parent(kv, parents)
+        self._etcd_foreach('config/inode', lambda kv: add_child(kv))
+        return children
+
+    def _check_parent(self, kv, parents):
+        if 'parent_id' not in kv['value']:
+            return 0
+        parent_id = kv['value']['parent_id']
+        _, _, pool_id, inode_id = kv['key'].split('/')
+        parent_pool_id = pool_id
+        if 'parent_pool_id' in kv['value'] and kv['value']['parent_pool_id']:
+            parent_pool_id = kv['value']['parent_pool_id']
+        inode = (int(pool_id) << 48) | (int(inode_id) & 0xffffffffffff)
+        parent = (int(parent_pool_id) << 48) | (int(parent_id) & 0xffffffffffff)
+        if parent in parents and inode not in parents:
+            return 1
+        return 0
+
+    def create_cloned_volume(self, volume, src_vref):
+        """Create a cloned volume from another volume."""
+
+        size = int(volume.size) * units.Gi
+        src_name = utils.convert_str(src_vref.name)
+        dest_name = utils.convert_str(volume.name)
+        if dest_name.find('@') >= 0 or dest_name.find('/') >= 0:
+            raise exception.VolumeBackendAPIException(data = '@ and / are forbidden in volume and snapshot names')
+
+        # FIXME Do full copy if requested (cfg.disable_clone)
+
+        if src_vref.admin_metadata.get('readonly') == 'True':
+            # source volume is a volume-image cache entry or other readonly volume
+            # clone without intermediate snapshot
+            src = self._get_image(src_name)
+            LOG.debug("creating image '%s' from '%s'", dest_name, src_name)
+            new_cfg = self._create_image(dest_name, {
+                'size': size,
+                'parent_id': src['idx']['id'],
+                'parent_pool_id': src['idx']['pool_id'],
+            })
+            return {}
+
+        clone_snap = "%s@%s.clone_snap" % (src_name, dest_name)
+        make_img = True
+        if (volume.display_name and
+            volume.display_name.startswith('image-') and
+            src_vref.project_id != volume.project_id):
+            # idiotic openstack creates image-volume cache entries
+            # as clones of normal VM volumes... :-X prevent it :-D
+            clone_snap = dest_name
+            make_img = False
+
+        LOG.debug("creating layer '%s' under '%s'", clone_snap, src_name)
+        new_cfg = self._create_snapshot(src_name, clone_snap, True)
+        if make_img:
+            # Then create a clone from it
+            new_cfg = self._create_image(dest_name, {
+                'size': size,
+                'parent_id': new_cfg['parent_id'],
+                'parent_pool_id': new_cfg['parent_pool_id'],
+            })
+
+        return {}
+
+    def create_volume_from_snapshot(self, volume, snapshot):
+        """Creates a cloned volume from an existing snapshot."""
+
+        vol_name = utils.convert_str(volume.name)
+        snap_name = utils.convert_str(snapshot.name)
+
+        snap = self._get_image(vol_name+'@'+snap_name)
+        if not snap:
+            raise exception.SnapshotNotFound(snapshot_id = snap_name)
+        snap_inode_id = int(resp['responses'][0]['kvs'][0]['value']['id'])
+        snap_pool_id = int(resp['responses'][0]['kvs'][0]['value']['pool_id'])
+
+        size = snap['cfg']['size']
+        if int(volume.size):
+            size = int(volume.size) * units.Gi
+        new_cfg = self._create_image(vol_name, {
+            'size': size,
+            'parent_id': snap['idx']['id'],
+            'parent_pool_id': snap['idx']['pool_id'],
+        })
+
+        return {}
+
+    def _vitastor_args(self):
+        args = []
+        for k in [ 'config_path', 'etcd_address', 'etcd_prefix' ]:
+            v = self.configuration.safe_get('vitastor_'+k)
+            if v:
+                args.extend(['--'+k, v])
+        return args
+
+    def _qemu_args(self):
+        args = ''
+        for k in [ 'config_path', 'etcd_address', 'etcd_prefix' ]:
+            v = self.configuration.safe_get('vitastor_'+k)
+            kk = k
+            if kk == 'etcd_address':
+                # FIXME use etcd_address in qemu driver
+                kk = 'etcd_host'
+            if v:
+                args += ':'+kk+'='+v.replace(':', '\\:')
+        return args
+
+    def delete_volume(self, volume):
+        """Deletes a logical volume."""
+
+        vol_name = utils.convert_str(volume.name)
+
+        # Find the volume and all its snapshots
+        range_end = b'index/image/' + vol_name.encode('utf-8')
+        range_end = range_end[0 : len(range_end)-1] + six.int2byte(range_end[len(range_end)-1] + 1)
+        resp = self._etcd_txn({ 'success': [
+            { 'request_range': { 'key': 'index/image/'+vol_name, 'range_end': range_end } },
+        ] })
+        if len(resp['responses'][0]['kvs']) == 0:
+            # already deleted
+            LOG.info("volume %s no longer exists in backend", vol_name)
+            return
+        layers = resp['responses'][0]['kvs']
+        layer_ids = {}
+        for kv in layers:
+            inode_id = int(kv['value']['id'])
+            pool_id = int(kv['value']['pool_id'])
+            inode_pool_id = (pool_id << 48) | (inode_id & 0xffffffffffff)
+            layer_ids[inode_pool_id] = True
+
+        # Check if the volume has clones and raise 'busy' if so
+        children = self._child_count(layer_ids)
+        if children > 0:
+            raise exception.VolumeIsBusy(volume_name = vol_name)
+
+        # Clear data
+        for kv in layers:
+            args = [
+                'vitastor-cli', 'rm', '--pool', str(kv['value']['pool_id']),
+                '--inode', str(kv['value']['id']), '--progress', '0',
+                *(self._vitastor_args())
+            ]
+            try:
+                self._execute(*args)
+            except processutils.ProcessExecutionError as exc:
+                LOG.error("Failed to remove layer "+kv['key']+": "+exc)
+                raise exception.VolumeBackendAPIException(data = exc.stderr)
+
+        # Delete all layers from etcd
+        requests = []
+        for kv in layers:
+            requests.append({ 'request_delete_range': { 'key': kv['key'] } })
+            requests.append({ 'request_delete_range': { 'key': 'config/inode/'+str(kv['value']['pool_id'])+'/'+str(kv['value']['id']) } })
+        self._etcd_txn({ 'success': requests })
+
+    def retype(self, context, volume, new_type, diff, host):
+        """Change extra type specifications for a volume."""
+
+        # FIXME Maybe (in the future) support multiple pools as different types
+        return True, {}
+
+    def ensure_export(self, context, volume):
+        """Synchronously recreates an export for a logical volume."""
+        pass
+
+    def create_export(self, context, volume, connector):
+        """Exports the volume."""
+        pass
+
+    def remove_export(self, context, volume):
+        """Removes an export for a logical volume."""
+        pass
+
+    def _create_image(self, vol_name, cfg):
+        pool_s = str(self.cfg['pool_id'])
+        image_id = 0
+        while image_id == 0:
+            # check if the image already exists and find a free ID
+            resp = self._etcd_txn({ 'success': [
+                { 'request_range': { 'key': 'index/image/'+vol_name } },
+                { 'request_range': { 'key': 'index/maxid/'+pool_s } },
+            ] })
+            if len(resp['responses'][0]['kvs']) > 0:
+                # already exists
+                raise exception.VolumeBackendAPIException(data = 'Volume '+vol_name+' already exists')
+            image_id, id_mod = self._next_id(resp['responses'][1])
+            # try to create the image
+            resp = self._etcd_txn({ 'compare': [
+                { 'target': 'MOD', 'mod_revision': id_mod, 'key': 'index/maxid/'+pool_s },
+                { 'target': 'VERSION', 'version': 0, 'key': 'index/image/'+vol_name },
+                { 'target': 'VERSION', 'version': 0, 'key': 'config/inode/'+pool_s+'/'+str(image_id) },
+            ], 'success': [
+                { 'request_put': { 'key': 'index/maxid/'+pool_s, 'value': image_id } },
+                { 'request_put': { 'key': 'index/image/'+vol_name, 'value': json.dumps({
+                    'id': image_id, 'pool_id': self.cfg['pool_id']
+                }) } },
+                { 'request_put': { 'key': 'config/inode/'+pool_s+'/'+str(image_id), 'value': json.dumps({
+                    **cfg, 'name': vol_name,
+                }) } },
+            ] })
+            if not resp.get('succeeded'):
+                # repeat
+                image_id = 0
+
+    def _create_snapshot(self, vol_name, snap_vol_name, allow_existing = False):
+        while True:
+            # check if the image already exists and snapshot doesn't
+            resp = self._etcd_txn({ 'success': [
+                { 'request_range': { 'key': 'index/image/'+vol_name } },
+                { 'request_range': { 'key': 'index/image/'+snap_vol_name } },
+            ] })
+            if len(resp['responses'][0]['kvs']) == 0:
+                raise exception.VolumeBackendAPIException(data = 'Volume '+vol_name+' does not exist')
+            if len(resp['responses'][1]['kvs']) > 0:
+                if allow_existing:
+                    snap_idx = resp['responses'][1]['kvs'][0]['value']
+                    resp = self._etcd_txn({ 'success': [
+                        { 'request_range': { 'key': 'config/inode/'+str(snap_idx['pool_id'])+'/'+str(snap_idx['id']) } },
+                    ] })
+                    if len(resp['responses'][0]['kvs']) == 0:
+                        raise exception.VolumeBackendAPIException(data =
+                            'Volume '+snap_vol_name+' is already indexed, but does not exist'
+                        )
+                    return resp['responses'][0]['kvs'][0]['value']
+                raise exception.VolumeBackendAPIException(
+                    data = 'Volume '+snap_vol_name+' already exists'
+                )
+            vol_idx = resp['responses'][0]['kvs'][0]['value']
+            vol_idx_mod = resp['responses'][0]['kvs'][0]['mod_revision']
+            # get image inode config and find a new ID
+            resp = self._etcd_txn({ 'success': [
+                { 'request_range': { 'key': 'config/inode/'+str(vol_idx['pool_id'])+'/'+str(vol_idx['id']) } },
+                { 'request_range': { 'key': 'index/maxid/'+str(self.cfg['pool_id']) } },
+            ] })
+            if len(resp['responses'][0]['kvs']) == 0:
+                raise exception.VolumeBackendAPIException(data = 'Volume '+vol_name+' does not exist')
+            vol_cfg = resp['responses'][0]['kvs'][0]['value']
+            vol_mod = resp['responses'][0]['kvs'][0]['mod_revision']
+            new_id, id_mod = self._next_id(resp['responses'][1])
+            # try to redirect image to the new inode
+            new_cfg = {
+                **vol_cfg, 'name': vol_name, 'parent_id': vol_idx['id'], 'parent_pool_id': vol_idx['pool_id']
+            }
+            resp = self._etcd_txn({ 'compare': [
+                { 'target': 'MOD', 'mod_revision': vol_idx_mod, 'key': 'index/image/'+vol_name },
+                { 'target': 'MOD', 'mod_revision': vol_mod, 'key': 'config/inode/'+str(vol_idx['pool_id'])+'/'+str(vol_idx['id']) },
+                { 'target': 'MOD', 'mod_revision': id_mod, 'key': 'index/maxid/'+str(self.cfg['pool_id']) },
+                { 'target': 'VERSION', 'version': 0, 'key': 'index/image/'+snap_vol_name },
+                { 'target': 'VERSION', 'version': 0, 'key': 'config/inode/'+str(self.cfg['pool_id'])+'/'+str(new_id) },
+            ], 'success': [
+                { 'request_put': { 'key': 'index/maxid/'+str(self.cfg['pool_id']), 'value': new_id } },
+                { 'request_put': { 'key': 'index/image/'+vol_name, 'value': json.dumps({
+                    'id': new_id, 'pool_id': self.cfg['pool_id']
+                }) } },
+                { 'request_put': { 'key': 'config/inode/'+str(self.cfg['pool_id'])+'/'+str(new_id), 'value': json.dumps(new_cfg) } },
+                { 'request_put': { 'key': 'index/image/'+snap_vol_name, 'value': json.dumps({
+                    'id': vol_idx['id'], 'pool_id': vol_idx['pool_id']
+                }) } },
+                { 'request_put': { 'key': 'config/inode/'+str(vol_idx['pool_id'])+'/'+str(vol_idx['id']), 'value': json.dumps({
+                    **vol_cfg, 'name': snap_vol_name, 'readonly': True
+                }) } }
+            ] })
+            if resp.get('succeeded'):
+                return new_cfg
+
+    def initialize_connection(self, volume, connector):
+        data = {
+            'driver_volume_type': 'vitastor',
+            'data': {
+                'config_path': self.configuration.vitastor_config_path,
+                'etcd_address': self.configuration.vitastor_etcd_address,
+                'etcd_prefix': self.configuration.vitastor_etcd_prefix,
+                'name': volume.name,
+                'logical_block_size': 512,
+                'physical_block_size': 4096,
+            }
+        }
+        LOG.debug('connection data: %s', data)
+        return data
+
+    def terminate_connection(self, volume, connector, **kwargs):
+        pass
+
+    def clone_image(self, context, volume, image_location, image_meta, image_service):
+        if image_location:
+            # Note: image_location[0] is glance image direct_url.
+            # image_location[1] contains the list of all locations (including
+            # direct_url) or None if show_multiple_locations is False in
+            # glance configuration.
+            if image_location[1]:
+                url_locations = [location['url'] for location in image_location[1]]
+            else:
+                url_locations = [image_location[0]]
+            # iterate all locations to look for a cloneable one.
+            for url_location in url_locations:
+                if url_location and url_location.startswith('cinder://'):
+                    # The idea is to use cinder://<volume-id> Glance volumes as base images
+                    base_vol = self.db.volume_get(context, url_location[len('cinder://') : ])
+                    if not base_vol or base_vol.volume_type_id != volume.volume_type_id:
+                        continue
+                    size = int(volume.size) * units.Gi
+                    dest_name = utils.convert_str(volume.name)
+                    # Find or create the base snapshot
+                    snap_cfg = self._create_snapshot(base_vol.name, base_vol.name+'@.clone_snap', True)
+                    # Then create a clone from it
+                    new_cfg = self._create_image(dest_name, {
+                        'size': size,
+                        'parent_id': snap_cfg['parent_id'],
+                        'parent_pool_id': snap_cfg['parent_pool_id'],
+                    })
+                    return ({}, True)
+        return ({}, False)
+
+    def copy_image_to_encrypted_volume(self, context, volume, image_service, image_id):
+        self.copy_image_to_volume(context, volume, image_service, image_id, encrypted = True)
+
+    def copy_image_to_volume(self, context, volume, image_service, image_id, encrypted = False):
+        tmp_dir = volume_utils.image_conversion_dir()
+        with tempfile.NamedTemporaryFile(dir = tmp_dir) as tmp:
+            image_utils.fetch_to_raw(
+                context, image_service, image_id, tmp.name,
+                self.configuration.volume_dd_blocksize, size = volume.size
+            )
+            out_format = [ '-O', 'raw' ]
+            if encrypted:
+                key_file, opts = self._encrypt_opts(volume, context)
+                out_format = [ '-O', 'luks', *opts ]
+            dest_name = utils.convert_str(volume.name)
+            self._try_execute(
+                'qemu-img', 'convert', '-f', 'raw', tmp.name, *out_format,
+                'vitastor:image='+dest_name.replace(':', '\\:')+self._qemu_args()
+            )
+            if encrypted:
+                key_file.close()
+
+    def copy_volume_to_image(self, context, volume, image_service, image_meta):
+        tmp_dir = volume_utils.image_conversion_dir()
+        tmp_file = os.path.join(tmp_dir, volume.name + '-' + image_meta['id'])
+        with fileutils.remove_path_on_error(tmp_file):
+            vol_name = utils.convert_str(volume.name)
+            self._try_execute(
+                'qemu-img', 'convert', '-f', 'raw',
+                'vitastor:image='+vol_name.replace(':', '\\:')+self._qemu_args(),
+                '-O', 'raw', tmp_file
+            )
+            # FIXME: Copy directly if the destination image is also in Vitastor
+            volume_utils.upload_volume(context, image_service, image_meta, tmp_file, volume)
+        os.unlink(tmp_file)
+
+    def _get_image(self, vol_name):
+        # find the image
+        resp = self._etcd_txn({ 'success': [
+            { 'request_range': { 'key': 'index/image/'+vol_name } },
+        ] })
+        if len(resp['responses'][0]['kvs']) == 0:
+            return None
+        vol_idx = resp['responses'][0]['kvs'][0]['value']
+        vol_idx_mod = resp['responses'][0]['kvs'][0]['mod_revision']
+        # get image inode config
+        resp = self._etcd_txn({ 'success': [
+            { 'request_range': { 'key': 'config/inode/'+str(vol_idx['pool_id'])+'/'+str(vol_idx['id']) } },
+        ] })
+        if len(resp['responses'][0]['kvs']) == 0:
+            return None
+        vol_cfg = resp['responses'][0]['kvs'][0]['value']
+        vol_cfg_mod = resp['responses'][0]['kvs'][0]['mod_revision']
+        return {
+            'cfg': vol_cfg,
+            'cfg_mod': vol_cfg_mod,
+            'idx': vol_idx,
+            'idx_mod': vol_idx_mod,
+        }
+
+    def extend_volume(self, volume, new_size):
+        """Extend an existing volume."""
+        vol_name = utils.convert_str(volume.name)
+        while True:
+            vol = self._get_image(vol_name)
+            if not vol:
+                raise exception.VolumeBackendAPIException(data = 'Volume '+vol_name+' does not exist')
+            # change size
+            size = int(new_size) * units.Gi
+            if size == vol['cfg']['size']:
+                break
+            resp = self._etcd_txn({ 'compare': [ {
+                'target': 'MOD',
+                'mod_revision': vol['cfg_mod'],
+                'key': 'config/inode/'+str(vol['idx']['pool_id'])+'/'+str(vol['idx']['id']),
+            } ], 'success': [
+                { 'request_put': {
+                    'key': 'config/inode/'+str(vol['idx']['pool_id'])+'/'+str(vol['idx']['id']),
+                    'value': json.dumps({ **vol['cfg'], 'size': size }),
+                } },
+            ] })
+            if resp.get('succeeded'):
+                break
+        LOG.debug(
+            "Extend volume from %(old_size)s GB to %(new_size)s GB.",
+            {'old_size': volume.size, 'new_size': new_size}
+        )
+
+    def _add_manageable_volume(self, kv, manageable_volumes, cinder_ids):
+        cfg = kv['value']
+        if kv['key'].find('@') >= 0:
+            # snapshot
+            return
+        image_id = volume_utils.extract_id_from_volume_name(cfg['name'])
+        image_info = {
+            'reference': {'source-name': image_name},
+            'size': int(math.ceil(float(cfg['size']) / units.Gi)),
+            'cinder_id': None,
+            'extra_info': None,
+        }
+        if image_id in cinder_ids:
+            image_info['cinder_id'] = image_id
+            image_info['safe_to_manage'] = False
+            image_info['reason_not_safe'] = 'already managed'
+        else:
+            image_info['safe_to_manage'] = True
+            image_info['reason_not_safe'] = None
+        manageable_volumes.append(image_info)
+
+    def get_manageable_volumes(self, cinder_volumes, marker, limit, offset, sort_keys, sort_dirs):
+        manageable_volumes = []
+        cinder_ids = [resource['id'] for resource in cinder_volumes]
+
+        # List all volumes
+        # FIXME: It's possible to use pagination in our case, but.. do we want it?
+        self._etcd_foreach('config/inode/'+str(self.cfg['pool_id']),
+            lambda kv: self._add_manageable_volume(kv, manageable_volumes, cinder_ids))
+
+        return volume_utils.paginate_entries_list(
+            manageable_volumes, marker, limit, offset, sort_keys, sort_dirs)
+
+    def _get_existing_name(existing_ref):
+        if not isinstance(existing_ref, dict):
+            existing_ref = {"source-name": existing_ref}
+        if 'source-name' not in existing_ref:
+            reason = _('Reference must contain source-name element.')
+            raise exception.ManageExistingInvalidReference(existing_ref=existing_ref, reason=reason)
+        src_name = utils.convert_str(existing_ref['source-name'])
+        if not src_name:
+            reason = _('Reference must contain source-name element.')
+            raise exception.ManageExistingInvalidReference(existing_ref=existing_ref, reason=reason)
+        return src_name
+
+    def manage_existing_get_size(self, volume, existing_ref):
+        """Return size of an existing image for manage_existing.
+
+        :param volume: volume ref info to be set
+        :param existing_ref: {'source-name': <image name>}
+        """
+        src_name = self._get_existing_name(existing_ref)
+        vol = self._get_image(src_name)
+        if not vol:
+            raise exception.VolumeBackendAPIException(data = 'Volume '+src_name+' does not exist')
+        return int(math.ceil(float(vol['cfg']['size']) / units.Gi))
+
+    def manage_existing(self, volume, existing_ref):
+        """Manages an existing image.
+
+        Renames the image name to match the expected name for the volume.
+
+        :param volume: volume ref info to be set
+        :param existing_ref: {'source-name': <image name>}
+        """
+        from_name = self._get_existing_name(existing_ref)
+        to_name = utils.convert_str(volume.name)
+        self._rename(from_name, to_name)
+
+    def _rename(self, from_name, to_name):
+        while True:
+            vol = self._get_image(from_name)
+            if not vol:
+                raise exception.VolumeBackendAPIException(data = 'Volume '+from_name+' does not exist')
+            to = self._get_image(to_name)
+            if to:
+                raise exception.VolumeBackendAPIException(data = 'Volume '+to_name+' already exists')
+            resp = self._etcd_txn({ 'compare': [
+                { 'target': 'MOD', 'mod_revision': vol['idx_mod'], 'key': 'index/image/'+vol['cfg']['name'] },
+                { 'target': 'MOD', 'mod_revision': vol['cfg_mod'], 'key': 'config/inode/'+str(vol['idx']['pool_id'])+'/'+str(vol['idx']['id']) },
+                { 'target': 'VERSION', 'version': 0, 'key': 'index/image/'+to_name },
+            ], 'success': [
+                { 'request_delete_range': { 'key': 'index/image/'+vol['cfg']['name'] } },
+                { 'request_put': { 'key': 'index/image/'+to_name, 'value': json.dumps(vol['idx']) } },
+                { 'request_put': { 'key': 'config/inode/'+str(vol['idx']['pool_id'])+'/'+str(vol['idx']['id']),
+                    'value': json.dumps({ **vol['cfg'], 'name': to_name }) } },
+            ] })
+            if resp.get('succeeded'):
+                break
+
+    def unmanage(self, volume):
+        pass
+
+    def _add_manageable_snapshot(self, kv, manageable_snapshots, cinder_ids):
+        cfg = kv['value']
+        dog = kv['key'].find('@')
+        if dog < 0:
+            # snapshot
+            return
+        image_name = kv['key'][0 : dog]
+        snap_name = kv['key'][dog+1 : ]
+        snapshot_id = volume_utils.extract_id_from_snapshot_name(snap_name)
+        snapshot_info = {
+            'reference': {'source-name': snap_name},
+            'size': int(math.ceil(float(cfg['size']) / units.Gi)),
+            'cinder_id': None,
+            'extra_info': None,
+            'safe_to_manage': False,
+            'reason_not_safe': None,
+            'source_reference': {'source-name': image_name}
+        }
+        if snapshot_id in cinder_ids:
+            # Exclude snapshots already managed.
+            snapshot_info['reason_not_safe'] = ('already managed')
+            snapshot_info['cinder_id'] = snapshot_id
+        elif snap_name.endswith('.clone_snap'):
+            # Exclude clone snapshot.
+            snapshot_info['reason_not_safe'] = ('used for clone snap')
+        else:
+            snapshot_info['safe_to_manage'] = True
+        manageable_snapshots.append(snapshot_info)
+
+    def get_manageable_snapshots(self, cinder_snapshots, marker, limit, offset, sort_keys, sort_dirs):
+        """List manageable snapshots in Vitastor."""
+        manageable_snapshots = []
+        cinder_snapshot_ids = [resource['id'] for resource in cinder_snapshots]
+        # List all volumes
+        # FIXME: It's possible to use pagination in our case, but.. do we want it?
+        self._etcd_foreach('config/inode/'+str(self.cfg['pool_id']),
+            lambda kv: self._add_manageable_volume(kv, manageable_snapshots, cinder_snapshot_ids))
+        return volume_utils.paginate_entries_list(
+            manageable_snapshots, marker, limit, offset, sort_keys, sort_dirs)
+
+    def manage_existing_snapshot_get_size(self, snapshot, existing_ref):
+        """Return size of an existing image for manage_existing.
+
+        :param snapshot: snapshot ref info to be set
+        :param existing_ref: {'source-name': <name of snapshot>}
+        """
+        vol_name = utils.convert_str(snapshot.volume_name)
+        snap_name = self._get_existing_name(existing_ref)
+        vol = self._get_image(vol_name+'@'+snap_name)
+        if not vol:
+            raise exception.ManageExistingInvalidReference(
+                existing_ref=snapshot_name, reason='Specified snapshot does not exist.'
+            )
+        return int(math.ceil(float(vol['cfg']['size']) / units.Gi))
+
+    def manage_existing_snapshot(self, snapshot, existing_ref):
+        """Manages an existing snapshot.
+
+        Renames the snapshot name to match the expected name for the snapshot.
+        Error checking done by manage_existing_get_size is not repeated.
+
+        :param snapshot: snapshot ref info to be set
+        :param existing_ref: {'source-name': <name of snapshot>}
+        """
+        vol_name = utils.convert_str(snapshot.volume_name)
+        snap_name = self._get_existing_name(existing_ref)
+        from_name = vol_name+'@'+snap_name
+        to_name = vol_name+'@'+utils.convert_str(snapshot.name)
+        self._rename(from_name, to_name)
+
+    def unmanage_snapshot(self, snapshot):
+        """Removes the specified snapshot from Cinder management."""
+        pass
+
+    def _dumps(self, obj):
+        return json.dumps(obj, separators=(',', ':'), sort_keys=True)

patches/devstack-local.conf

@@ -0,0 +1,23 @@
+# Devstack configuration for bridged networking
+
+[[local|localrc]]
+ADMIN_PASSWORD=secret
+DATABASE_PASSWORD=$ADMIN_PASSWORD
+RABBIT_PASSWORD=$ADMIN_PASSWORD
+SERVICE_PASSWORD=$ADMIN_PASSWORD
+HOST_IP=10.0.2.15
+Q_USE_SECGROUP=True
+FLOATING_RANGE="10.0.2.0/24"
+IPV4_ADDRS_SAFE_TO_USE="10.0.5.0/24"
+Q_FLOATING_ALLOCATION_POOL=start=10.0.2.50,end=10.0.2.100
+PUBLIC_NETWORK_GATEWAY=10.0.2.2
+PUBLIC_INTERFACE=ens3
+Q_USE_PROVIDERNET_FOR_PUBLIC=True
+Q_AGENT=linuxbridge
+Q_ML2_PLUGIN_MECHANISM_DRIVERS=linuxbridge
+LB_PHYSICAL_INTERFACE=ens3
+PUBLIC_PHYSICAL_NETWORK=default
+LB_INTERFACE_MAPPINGS=default:ens3
+Q_SERVICE_PLUGIN_CLASSES=
+Q_ML2_PLUGIN_TYPE_DRIVERS=flat
+Q_ML2_PLUGIN_EXT_DRIVERS=

patches/libvirt-5.0-vitastor.diff

@@ -0,0 +1,609 @@
+commit bd283191b3e7a4c6d1c100d3d96e348a1ebffe55
+Author: Vitaliy Filippov <vitalif@yourcmc.ru>
+Date:   Sun Jun 27 12:52:40 2021 +0300
+
+    Add Vitastor support
+
+diff --git a/docs/schemas/domaincommon.rng b/docs/schemas/domaincommon.rng
+index aa50eac..082b4f8 100644
+--- a/docs/schemas/domaincommon.rng
++++ b/docs/schemas/domaincommon.rng
+@@ -1728,6 +1728,35 @@
+     </element>
+   </define>
+ 
++  <define name="diskSourceNetworkProtocolVitastor">
++    <element name="source">
++      <interleave>
++        <attribute name="protocol">
++          <value>vitastor</value>
++        </attribute>
++        <ref name="diskSourceCommon"/>
++        <optional>
++          <attribute name="name"/>
++        </optional>
++        <optional>
++          <attribute name="query"/>
++        </optional>
++        <zeroOrMore>
++          <ref name="diskSourceNetworkHost"/>
++        </zeroOrMore>
++        <optional>
++          <element name="config">
++            <attribute name="file">
++              <ref name="absFilePath"/>
++            </attribute>
++            <empty/>
++          </element>
++        </optional>
++        <empty/>
++      </interleave>
++    </element>
++  </define>
++
+   <define name="diskSourceNetworkProtocolISCSI">
+     <element name="source">
+       <attribute name="protocol">
+@@ -1851,6 +1880,7 @@
+       <ref name="diskSourceNetworkProtocolHTTP"/>
+       <ref name="diskSourceNetworkProtocolSimple"/>
+       <ref name="diskSourceNetworkProtocolVxHS"/>
++      <ref name="diskSourceNetworkProtocolVitastor"/>
+     </choice>
+   </define>
+ 
+diff --git a/include/libvirt/libvirt-storage.h b/include/libvirt/libvirt-storage.h
+index 4bf2b5f..dbc011b 100644
+--- a/include/libvirt/libvirt-storage.h
++++ b/include/libvirt/libvirt-storage.h
+@@ -240,6 +240,7 @@ typedef enum {
+     VIR_CONNECT_LIST_STORAGE_POOLS_GLUSTER       = 1 << 16,
+     VIR_CONNECT_LIST_STORAGE_POOLS_ZFS           = 1 << 17,
+     VIR_CONNECT_LIST_STORAGE_POOLS_VSTORAGE      = 1 << 18,
++    VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR      = 1 << 20,
+ } virConnectListAllStoragePoolsFlags;
+ 
+ int                     virConnectListAllStoragePools(virConnectPtr conn,
+diff --git a/src/conf/domain_conf.c b/src/conf/domain_conf.c
+index 222bb8c..685d255 100644
+--- a/src/conf/domain_conf.c
++++ b/src/conf/domain_conf.c
+@@ -8653,6 +8653,10 @@ virDomainDiskSourceNetworkParse(xmlNodePtr node,
+         goto cleanup;
+     }
+ 
++    if (src->protocol == VIR_STORAGE_NET_PROTOCOL_VITASTOR) {
++        src->relPath = virXMLPropString(node, "query");
++    }
++
+     if ((haveTLS = virXMLPropString(node, "tls")) &&
+         (src->haveTLS = virTristateBoolTypeFromString(haveTLS)) <= 0) {
+         virReportError(VIR_ERR_XML_ERROR,
+@@ -23849,6 +23853,10 @@ virDomainDiskSourceFormatNetwork(virBufferPtr attrBuf,
+ 
+     virBufferEscapeString(attrBuf, " name='%s'", path ? path : src->path);
+ 
++    if (src->protocol == VIR_STORAGE_NET_PROTOCOL_VITASTOR && src->relPath != NULL) {
++        virBufferEscapeString(attrBuf, " query='%s'", src->relPath);
++    }
++
+     VIR_FREE(path);
+ 
+     if (src->haveTLS != VIR_TRISTATE_BOOL_ABSENT &&
+@@ -30930,6 +30938,7 @@ virDomainDiskTranslateSourcePool(virDomainDiskDefPtr def)
+ 
+     case VIR_STORAGE_POOL_MPATH:
+     case VIR_STORAGE_POOL_RBD:
++    case VIR_STORAGE_POOL_VITASTOR:
+     case VIR_STORAGE_POOL_SHEEPDOG:
+     case VIR_STORAGE_POOL_GLUSTER:
+     case VIR_STORAGE_POOL_LAST:
+diff --git a/src/conf/storage_conf.c b/src/conf/storage_conf.c
+index 55db7a9..7cbe937 100644
+--- a/src/conf/storage_conf.c
++++ b/src/conf/storage_conf.c
+@@ -58,7 +58,7 @@ VIR_ENUM_IMPL(virStoragePool,
+               "logical", "disk", "iscsi",
+               "iscsi-direct", "scsi", "mpath",
+               "rbd", "sheepdog", "gluster",
+-              "zfs", "vstorage")
++              "zfs", "vstorage", "vitastor")
+ 
+ VIR_ENUM_IMPL(virStoragePoolFormatFileSystem,
+               VIR_STORAGE_POOL_FS_LAST,
+@@ -232,6 +232,18 @@ static virStoragePoolTypeInfo poolTypeInfo[] = {
+           .formatToString = virStorageFileFormatTypeToString,
+       }
+     },
++    {.poolType = VIR_STORAGE_POOL_VITASTOR,
++     .poolOptions = {
++         .flags = (VIR_STORAGE_POOL_SOURCE_HOST |
++                   VIR_STORAGE_POOL_SOURCE_NETWORK |
++                   VIR_STORAGE_POOL_SOURCE_NAME),
++      },
++      .volOptions = {
++          .defaultFormat = VIR_STORAGE_FILE_RAW,
++          .formatFromString = virStorageVolumeFormatFromString,
++          .formatToString = virStorageFileFormatTypeToString,
++      }
++    },
+     {.poolType = VIR_STORAGE_POOL_SHEEPDOG,
+      .poolOptions = {
+          .flags = (VIR_STORAGE_POOL_SOURCE_HOST |
+@@ -434,6 +446,11 @@ virStoragePoolDefParseSource(xmlXPathContextPtr ctxt,
+                        _("element 'name' is mandatory for RBD pool"));
+         goto cleanup;
+     }
++    if (pool_type == VIR_STORAGE_POOL_VITASTOR && source->name == NULL) {
++        virReportError(VIR_ERR_XML_ERROR, "%s",
++                       _("element 'name' is mandatory for Vitastor pool"));
++        return -1;
++    }
+ 
+     if (options->formatFromString) {
+         char *format = virXPathString("string(./format/@type)", ctxt);
+@@ -1009,6 +1026,7 @@ virStoragePoolDefFormatBuf(virBufferPtr buf,
+     /* RBD, Sheepdog, Gluster and Iscsi-direct devices are not local block devs nor
+      * files, so they don't have a target */
+     if (def->type != VIR_STORAGE_POOL_RBD &&
++        def->type != VIR_STORAGE_POOL_VITASTOR &&
+         def->type != VIR_STORAGE_POOL_SHEEPDOG &&
+         def->type != VIR_STORAGE_POOL_GLUSTER &&
+         def->type != VIR_STORAGE_POOL_ISCSI_DIRECT) {
+diff --git a/src/conf/storage_conf.h b/src/conf/storage_conf.h
+index dc0aa2a..ed4983d 100644
+--- a/src/conf/storage_conf.h
++++ b/src/conf/storage_conf.h
+@@ -91,6 +91,7 @@ typedef enum {
+     VIR_STORAGE_POOL_GLUSTER,  /* Gluster device */
+     VIR_STORAGE_POOL_ZFS,      /* ZFS */
+     VIR_STORAGE_POOL_VSTORAGE, /* Virtuozzo Storage */
++    VIR_STORAGE_POOL_VITASTOR, /* Vitastor */
+ 
+     VIR_STORAGE_POOL_LAST,
+ } virStoragePoolType;
+@@ -422,6 +423,7 @@ VIR_ENUM_DECL(virStoragePartedFs)
+                  VIR_CONNECT_LIST_STORAGE_POOLS_SCSI     | \
+                  VIR_CONNECT_LIST_STORAGE_POOLS_MPATH    | \
+                  VIR_CONNECT_LIST_STORAGE_POOLS_RBD      | \
++                 VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR | \
+                  VIR_CONNECT_LIST_STORAGE_POOLS_SHEEPDOG | \
+                  VIR_CONNECT_LIST_STORAGE_POOLS_GLUSTER  | \
+                  VIR_CONNECT_LIST_STORAGE_POOLS_ZFS      | \
+diff --git a/src/conf/virstorageobj.c b/src/conf/virstorageobj.c
+index 6ea6a97..3ba45b9 100644
+--- a/src/conf/virstorageobj.c
++++ b/src/conf/virstorageobj.c
+@@ -1478,6 +1478,7 @@ virStoragePoolObjSourceFindDuplicateCb(const void *payload,
+             return 1;
+         break;
+ 
++    case VIR_STORAGE_POOL_VITASTOR:
+     case VIR_STORAGE_POOL_RBD:
+     case VIR_STORAGE_POOL_LAST:
+         break;
+@@ -1971,6 +1972,8 @@ virStoragePoolObjMatch(virStoragePoolObjPtr obj,
+                (obj->def->type == VIR_STORAGE_POOL_MPATH))   ||
+               (MATCH(VIR_CONNECT_LIST_STORAGE_POOLS_RBD) &&
+                (obj->def->type == VIR_STORAGE_POOL_RBD))     ||
++              (MATCH(VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR) &&
++               (obj->def->type == VIR_STORAGE_POOL_VITASTOR)) ||
+               (MATCH(VIR_CONNECT_LIST_STORAGE_POOLS_SHEEPDOG) &&
+                (obj->def->type == VIR_STORAGE_POOL_SHEEPDOG)) ||
+               (MATCH(VIR_CONNECT_LIST_STORAGE_POOLS_GLUSTER) &&
+diff --git a/src/libvirt-storage.c b/src/libvirt-storage.c
+index 2ea3e94..d5d2273 100644
+--- a/src/libvirt-storage.c
++++ b/src/libvirt-storage.c
+@@ -92,6 +92,7 @@ virStoragePoolGetConnect(virStoragePoolPtr pool)
+  * VIR_CONNECT_LIST_STORAGE_POOLS_SCSI
+  * VIR_CONNECT_LIST_STORAGE_POOLS_MPATH
+  * VIR_CONNECT_LIST_STORAGE_POOLS_RBD
++ * VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR
+  * VIR_CONNECT_LIST_STORAGE_POOLS_SHEEPDOG
+  *
+  * Returns the number of storage pools found or -1 and sets @pools to
+diff --git a/src/libxl/libxl_conf.c b/src/libxl/libxl_conf.c
+index 73e988a..ab7bb81 100644
+--- a/src/libxl/libxl_conf.c
++++ b/src/libxl/libxl_conf.c
+@@ -905,6 +905,7 @@ libxlMakeNetworkDiskSrcStr(virStorageSourcePtr src,
+     case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+     case VIR_STORAGE_NET_PROTOCOL_SSH:
+     case VIR_STORAGE_NET_PROTOCOL_VXHS:
++    case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+     case VIR_STORAGE_NET_PROTOCOL_LAST:
+     case VIR_STORAGE_NET_PROTOCOL_NONE:
+         virReportError(VIR_ERR_NO_SUPPORT,
+diff --git a/src/qemu/qemu_block.c b/src/qemu/qemu_block.c
+index cbf0aa4..096700d 100644
+--- a/src/qemu/qemu_block.c
++++ b/src/qemu/qemu_block.c
+@@ -959,6 +959,42 @@ qemuBlockStorageSourceGetRBDProps(virStorageSourcePtr src)
+ }
+ 
+ 
++static virJSONValuePtr
++qemuBlockStorageSourceGetVitastorProps(virStorageSource *src)
++{
++    virJSONValuePtr ret = NULL;
++    virStorageNetHostDefPtr host;
++    size_t i;
++    virBuffer buf = VIR_BUFFER_INITIALIZER;
++    char *etcd = NULL;
++
++    for (i = 0; i < src->nhosts; i++) {
++        host = src->hosts + i;
++        if ((virStorageNetHostTransport)host->transport != VIR_STORAGE_NET_HOST_TRANS_TCP) {
++            goto cleanup;
++        }
++        virBufferAsprintf(&buf, i > 0 ? ",%s:%u" : "%s:%u", host->name, host->port);
++    }
++    if (src->nhosts > 0) {
++        etcd = virBufferContentAndReset(&buf);
++    }
++
++    if (virJSONValueObjectCreate(&ret,
++                                 "s:driver", "vitastor",
++                                 "S:etcd_host", etcd,
++                                 "S:etcd_prefix", src->relPath,
++                                 "S:config_path", src->configFile,
++                                 "s:image", src->path,
++                                 NULL) < 0)
++        goto cleanup;
++
++cleanup:
++    VIR_FREE(etcd);
++    virBufferFreeAndReset(&buf);
++    return ret;
++}
++
++
+ static virJSONValuePtr
+ qemuBlockStorageSourceGetSheepdogProps(virStorageSourcePtr src)
+ {
+@@ -1174,6 +1210,11 @@ qemuBlockStorageSourceGetBackendProps(virStorageSourcePtr src,
+                 return NULL;
+             break;
+ 
++        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
++            if (!(fileprops = qemuBlockStorageSourceGetVitastorProps(src)))
++                return NULL;
++            break;
++
+         case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+             if (!(fileprops = qemuBlockStorageSourceGetSheepdogProps(src)))
+                 return NULL;
+diff --git a/src/qemu/qemu_command.c b/src/qemu/qemu_command.c
+index 822d5f8..e375cef 100644
+--- a/src/qemu/qemu_command.c
++++ b/src/qemu/qemu_command.c
+@@ -975,6 +975,43 @@ qemuBuildNetworkDriveStr(virStorageSourcePtr src,
+             ret = virBufferContentAndReset(&buf);
+             break;
+ 
++        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
++            if (strchr(src->path, ':')) {
++                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
++                               _("':' not allowed in Vitastor source volume name '%s'"),
++                               src->path);
++                return NULL;
++            }
++
++            virBufferStrcat(&buf, "vitastor:image=", src->path, NULL);
++
++            if (src->nhosts > 0) {
++                virBufferAddLit(&buf, ":etcd_host=");
++                for (i = 0; i < src->nhosts; i++) {
++                    if (i)
++                        virBufferAddLit(&buf, ",");
++
++                    /* assume host containing : is ipv6 */
++                    if (strchr(src->hosts[i].name, ':'))
++                        virBufferEscape(&buf, '\\', ":", "[%s]",
++                                        src->hosts[i].name);
++                    else
++                        virBufferAsprintf(&buf, "%s", src->hosts[i].name);
++
++                    if (src->hosts[i].port)
++                        virBufferAsprintf(&buf, "\\:%u", src->hosts[i].port);
++                }
++            }
++
++            if (src->configFile)
++                virBufferEscape(&buf, '\\', ":", ":config_path=%s", src->configFile);
++
++            if (src->relPath)
++                virBufferEscape(&buf, '\\', ":", ":etcd_prefix=%s", src->relPath);
++
++            ret = virBufferContentAndReset(&buf);
++            break;
++
+         case VIR_STORAGE_NET_PROTOCOL_VXHS:
+             virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
+                            _("VxHS protocol does not support URI syntax"));
+diff --git a/src/qemu/qemu_domain.c b/src/qemu/qemu_domain.c
+index ec6b340..f399efa 100644
+--- a/src/qemu/qemu_domain.c
++++ b/src/qemu/qemu_domain.c
+@@ -10881,6 +10881,7 @@ qemuDomainPrepareStorageSourceTLS(virStorageSourcePtr src,
+         break;
+ 
+     case VIR_STORAGE_NET_PROTOCOL_RBD:
++    case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+     case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+     case VIR_STORAGE_NET_PROTOCOL_GLUSTER:
+     case VIR_STORAGE_NET_PROTOCOL_ISCSI:
+diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c
+index 1d96170..2d24396 100644
+--- a/src/qemu/qemu_driver.c
++++ b/src/qemu/qemu_driver.c
+@@ -14687,6 +14687,7 @@ qemuDomainSnapshotPrepareDiskExternalInactive(virDomainSnapshotDiskDefPtr snapdi
+         case VIR_STORAGE_NET_PROTOCOL_TFTP:
+         case VIR_STORAGE_NET_PROTOCOL_SSH:
+         case VIR_STORAGE_NET_PROTOCOL_VXHS:
++        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+         case VIR_STORAGE_NET_PROTOCOL_LAST:
+             virReportError(VIR_ERR_INTERNAL_ERROR,
+                            _("external inactive snapshots are not supported on "
+@@ -14764,6 +14765,7 @@ qemuDomainSnapshotPrepareDiskExternalActive(virDomainSnapshotDiskDefPtr snapdisk
+         case VIR_STORAGE_NET_PROTOCOL_TFTP:
+         case VIR_STORAGE_NET_PROTOCOL_SSH:
+         case VIR_STORAGE_NET_PROTOCOL_VXHS:
++        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+         case VIR_STORAGE_NET_PROTOCOL_LAST:
+             virReportError(VIR_ERR_INTERNAL_ERROR,
+                            _("external active snapshots are not supported on "
+@@ -14887,6 +14889,7 @@ qemuDomainSnapshotPrepareDiskInternal(virDomainDiskDefPtr disk,
+         case VIR_STORAGE_NET_PROTOCOL_TFTP:
+         case VIR_STORAGE_NET_PROTOCOL_SSH:
+         case VIR_STORAGE_NET_PROTOCOL_VXHS:
++        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+         case VIR_STORAGE_NET_PROTOCOL_LAST:
+             virReportError(VIR_ERR_INTERNAL_ERROR,
+                            _("internal inactive snapshots are not supported on "
+diff --git a/src/qemu/qemu_parse_command.c b/src/qemu/qemu_parse_command.c
+index c4650f0..551da41 100644
+--- a/src/qemu/qemu_parse_command.c
++++ b/src/qemu/qemu_parse_command.c
+@@ -2184,6 +2184,7 @@ qemuParseCommandLine(virFileCachePtr capsCache,
+                 case VIR_STORAGE_NET_PROTOCOL_TFTP:
+                 case VIR_STORAGE_NET_PROTOCOL_SSH:
+                 case VIR_STORAGE_NET_PROTOCOL_LAST:
++                case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+                 case VIR_STORAGE_NET_PROTOCOL_NONE:
+                     /* ignored for now */
+                     break;
+diff --git a/src/storage/storage_driver.c b/src/storage/storage_driver.c
+index 4a13e90..33301c7 100644
+--- a/src/storage/storage_driver.c
++++ b/src/storage/storage_driver.c
+@@ -1568,6 +1568,7 @@ storageVolLookupByPathCallback(virStoragePoolObjPtr obj,
+         case VIR_STORAGE_POOL_RBD:
+         case VIR_STORAGE_POOL_SHEEPDOG:
+         case VIR_STORAGE_POOL_ZFS:
++        case VIR_STORAGE_POOL_VITASTOR:
+         case VIR_STORAGE_POOL_LAST:
+             ignore_value(VIR_STRDUP(stable_path, data->path));
+             break;
+diff --git a/src/util/virstoragefile.c b/src/util/virstoragefile.c
+index bd4b027..b323cd6 100644
+--- a/src/util/virstoragefile.c
++++ b/src/util/virstoragefile.c
+@@ -84,7 +84,8 @@ VIR_ENUM_IMPL(virStorageNetProtocol, VIR_STORAGE_NET_PROTOCOL_LAST,
+               "ftps",
+               "tftp",
+               "ssh",
+-              "vxhs")
++              "vxhs",
++              "vitastor")
+ 
+ VIR_ENUM_IMPL(virStorageNetHostTransport, VIR_STORAGE_NET_HOST_TRANS_LAST,
+               "tcp",
+@@ -2839,6 +2840,83 @@ virStorageSourceParseRBDColonString(const char *rbdstr,
+ }
+ 
+ 
++static int
++virStorageSourceParseVitastorColonString(const char *colonstr,
++                                         virStorageSourcePtr src)
++{
++    char *p, *e, *next;
++    char *options = NULL;
++
++    /* optionally skip the "vitastor:" prefix if provided */
++    if (STRPREFIX(colonstr, "vitastor:"))
++        colonstr += strlen("vitastor:");
++
++    if (VIR_STRDUP(options, colonstr) < 0)
++        return -1;
++
++    p = options;
++    while (*p) {
++        /* find : delimiter or end of string */
++        for (e = p; *e && *e != ':'; ++e) {
++            if (*e == '\\') {
++                e++;
++                if (*e == '\0')
++                    break;
++            }
++        }
++        if (*e == '\0') {
++            next = e;    /* last kv pair */
++        } else {
++            next = e + 1;
++            *e = '\0';
++        }
++
++        if (STRPREFIX(p, "image=")) {
++            if (VIR_STRDUP(src->path, p + strlen("image=")) < 0)
++                return -1;
++        } else if (STRPREFIX(p, "etcd_prefix=")) {
++            if (VIR_STRDUP(src->relPath, p + strlen("etcd_prefix=")) < 0)
++                return -1;
++        } else if (STRPREFIX(p, "config_file=")) {
++            if (VIR_STRDUP(src->configFile, p + strlen("config_file=")) < 0)
++                return -1;
++        } else if (STRPREFIX(p, "etcd_host=")) {
++            char *h, *sep;
++
++            h = p + strlen("etcd_host=");
++            while (h < e) {
++                for (sep = h; sep < e; ++sep) {
++                    if (*sep == '\\' && (sep[1] == ',' ||
++                                         sep[1] == ';' ||
++                                         sep[1] == ' ')) {
++                        *sep = '\0';
++                        sep += 2;
++                        break;
++                    }
++                }
++
++                if (virStorageSourceRBDAddHost(src, h) < 0)
++                    goto error;
++
++                h = sep;
++            }
++        }
++
++        p = next;
++    }
++
++    if (!src->path) {
++        goto error;
++    }
++
++    return 0;
++
++error:
++    VIR_FREE(options);
++    return -1;
++}
++
++
+ static int
+ virStorageSourceParseNBDColonString(const char *nbdstr,
+                                     virStorageSourcePtr src)
+@@ -2942,6 +3020,11 @@ virStorageSourceParseBackingColon(virStorageSourcePtr src,
+             goto cleanup;
+         break;
+ 
++    case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
++        if (virStorageSourceParseVitastorColonString(path, src) < 0)
++            return -1;
++        break;
++
+     case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+     case VIR_STORAGE_NET_PROTOCOL_LAST:
+     case VIR_STORAGE_NET_PROTOCOL_NONE:
+@@ -3441,6 +3524,56 @@ virStorageSourceParseBackingJSONRBD(virStorageSourcePtr src,
+     return ret;
+ }
+ 
++static int
++virStorageSourceParseBackingJSONVitastor(virStorageSourcePtr src,
++                                         virJSONValuePtr json,
++                                         int opaque ATTRIBUTE_UNUSED)
++{
++    const char *filename;
++    const char *image = virJSONValueObjectGetString(json, "image");
++    const char *conf = virJSONValueObjectGetString(json, "config_path");
++    const char *etcd_prefix = virJSONValueObjectGetString(json, "etcd_prefix");
++    virJSONValuePtr servers = virJSONValueObjectGetArray(json, "server");
++    size_t nservers;
++    size_t i;
++
++    src->type = VIR_STORAGE_TYPE_NETWORK;
++    src->protocol = VIR_STORAGE_NET_PROTOCOL_VITASTOR;
++
++    /* legacy syntax passed via 'filename' option */
++    if ((filename = virJSONValueObjectGetString(json, "filename")))
++        return virStorageSourceParseVitastorColonString(filename, src);
++
++    if (!image) {
++        virReportError(VIR_ERR_INVALID_ARG, "%s",
++                       _("missing image name in Vitastor backing volume "
++                         "JSON specification"));
++        return -1;
++    }
++
++    if (VIR_STRDUP(src->path, image) < 0 ||
++        VIR_STRDUP(src->configFile, conf) < 0 ||
++        VIR_STRDUP(src->relPath, etcd_prefix) < 0)
++        return -1;
++
++    if (servers) {
++        nservers = virJSONValueArraySize(servers);
++
++        if (VIR_ALLOC_N(src->hosts, nservers) < 0)
++            return -1;
++
++        src->nhosts = nservers;
++
++        for (i = 0; i < nservers; i++) {
++            if (virStorageSourceParseBackingJSONInetSocketAddress(src->hosts + i,
++                                                                  virJSONValueArrayGet(servers, i)) < 0)
++                return -1;
++        }
++    }
++
++    return 0;
++}
++
+ static int
+ virStorageSourceParseBackingJSONRaw(virStorageSourcePtr src,
+                                     virJSONValuePtr json,
+@@ -3507,6 +3640,7 @@ static const struct virStorageSourceJSONDriverParser jsonParsers[] = {
+     {"sheepdog", virStorageSourceParseBackingJSONSheepdog, 0},
+     {"ssh", virStorageSourceParseBackingJSONSSH, 0},
+     {"rbd", virStorageSourceParseBackingJSONRBD, 0},
++    {"vitastor", virStorageSourceParseBackingJSONVitastor, 0},
+     {"raw", virStorageSourceParseBackingJSONRaw, 0},
+     {"vxhs", virStorageSourceParseBackingJSONVxHS, 0},
+ };
+@@ -4276,6 +4410,7 @@ virStorageSourceNetworkDefaultPort(virStorageNetProtocol protocol)
+         case VIR_STORAGE_NET_PROTOCOL_GLUSTER:
+             return 24007;
+ 
++        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+         case VIR_STORAGE_NET_PROTOCOL_RBD:
+             /* we don't provide a default for RBD */
+             return 0;
+diff --git a/src/util/virstoragefile.h b/src/util/virstoragefile.h
+index 1d6161a..8d83bf3 100644
+--- a/src/util/virstoragefile.h
++++ b/src/util/virstoragefile.h
+@@ -134,6 +134,7 @@ typedef enum {
+     VIR_STORAGE_NET_PROTOCOL_TFTP,
+     VIR_STORAGE_NET_PROTOCOL_SSH,
+     VIR_STORAGE_NET_PROTOCOL_VXHS,
++    VIR_STORAGE_NET_PROTOCOL_VITASTOR,
+ 
+     VIR_STORAGE_NET_PROTOCOL_LAST
+ } virStorageNetProtocol;
+diff --git a/src/xenconfig/xen_xl.c b/src/xenconfig/xen_xl.c
+index accfc3a..a18f9c3 100644
+--- a/src/xenconfig/xen_xl.c
++++ b/src/xenconfig/xen_xl.c
+@@ -1535,6 +1535,7 @@ xenFormatXLDiskSrcNet(virStorageSourcePtr src)
+     case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+     case VIR_STORAGE_NET_PROTOCOL_SSH:
+     case VIR_STORAGE_NET_PROTOCOL_VXHS:
++    case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+     case VIR_STORAGE_NET_PROTOCOL_LAST:
+     case VIR_STORAGE_NET_PROTOCOL_NONE:
+         virReportError(VIR_ERR_NO_SUPPORT,
+diff --git a/tools/virsh-pool.c b/tools/virsh-pool.c
+index 70ca39b..9caef51 100644
+--- a/tools/virsh-pool.c
++++ b/tools/virsh-pool.c
+@@ -1219,6 +1219,9 @@ cmdPoolList(vshControl *ctl, const vshCmd *cmd ATTRIBUTE_UNUSED)
+             case VIR_STORAGE_POOL_VSTORAGE:
+                 flags |= VIR_CONNECT_LIST_STORAGE_POOLS_VSTORAGE;
+                 break;
++            case VIR_STORAGE_POOL_VITASTOR:
++                flags |= VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR;
++                break;
+             case VIR_STORAGE_POOL_LAST:
+                 break;
+             }

patches/libvirt-7.0-vitastor.diff

@@ -0,0 +1,657 @@
+commit 41cdfe8317d98f70aadedfdbb381effed2641bdd
+Author: Vitaliy Filippov <vitalif@yourcmc.ru>
+Date:   Fri Jul 9 01:31:57 2021 +0300
+
+    Add Vitastor support
+
+diff --git a/docs/schemas/domaincommon.rng b/docs/schemas/domaincommon.rng
+index 7dc419b..875433b 100644
+--- a/docs/schemas/domaincommon.rng
++++ b/docs/schemas/domaincommon.rng
+@@ -1827,6 +1827,35 @@
+     </element>
+   </define>
+ 
++  <define name="diskSourceNetworkProtocolVitastor">
++    <element name="source">
++      <interleave>
++        <attribute name="protocol">
++          <value>vitastor</value>
++        </attribute>
++        <ref name="diskSourceCommon"/>
++        <optional>
++          <attribute name="name"/>
++        </optional>
++        <optional>
++          <attribute name="query"/>
++        </optional>
++        <zeroOrMore>
++          <ref name="diskSourceNetworkHost"/>
++        </zeroOrMore>
++        <optional>
++          <element name="config">
++            <attribute name="file">
++              <ref name="absFilePath"/>
++            </attribute>
++            <empty/>
++          </element>
++        </optional>
++        <empty/>
++      </interleave>
++    </element>
++  </define>
++
+   <define name="diskSourceNetworkProtocolISCSI">
+     <element name="source">
+       <attribute name="protocol">
+@@ -2083,6 +2112,7 @@
+       <ref name="diskSourceNetworkProtocolSimple"/>
+       <ref name="diskSourceNetworkProtocolVxHS"/>
+       <ref name="diskSourceNetworkProtocolNFS"/>
++      <ref name="diskSourceNetworkProtocolVitastor"/>
+     </choice>
+   </define>
+ 
+diff --git a/include/libvirt/libvirt-storage.h b/include/libvirt/libvirt-storage.h
+index 089e1e0..d7e7ef4 100644
+--- a/include/libvirt/libvirt-storage.h
++++ b/include/libvirt/libvirt-storage.h
+@@ -245,6 +245,7 @@ typedef enum {
+     VIR_CONNECT_LIST_STORAGE_POOLS_ZFS           = 1 << 17,
+     VIR_CONNECT_LIST_STORAGE_POOLS_VSTORAGE      = 1 << 18,
+     VIR_CONNECT_LIST_STORAGE_POOLS_ISCSI_DIRECT  = 1 << 19,
++    VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR      = 1 << 20,
+ } virConnectListAllStoragePoolsFlags;
+ 
+ int                     virConnectListAllStoragePools(virConnectPtr conn,
+diff --git a/src/conf/domain_conf.c b/src/conf/domain_conf.c
+index 01b7187..c6e9702 100644
+--- a/src/conf/domain_conf.c
++++ b/src/conf/domain_conf.c
+@@ -8261,7 +8261,8 @@ virDomainDiskSourceNetworkParse(xmlNodePtr node,
+     src->configFile = virXPathString("string(./config/@file)", ctxt);
+ 
+     if (src->protocol == VIR_STORAGE_NET_PROTOCOL_HTTP ||
+-        src->protocol == VIR_STORAGE_NET_PROTOCOL_HTTPS)
++        src->protocol == VIR_STORAGE_NET_PROTOCOL_HTTPS ||
++        src->protocol == VIR_STORAGE_NET_PROTOCOL_VITASTOR)
+         src->query = virXMLPropString(node, "query");
+ 
+     if (virDomainStorageNetworkParseHosts(node, ctxt, &src->hosts, &src->nhosts) < 0)
+@@ -31392,6 +31393,7 @@ virDomainStorageSourceTranslateSourcePool(virStorageSourcePtr src,
+ 
+     case VIR_STORAGE_POOL_MPATH:
+     case VIR_STORAGE_POOL_RBD:
++    case VIR_STORAGE_POOL_VITASTOR:
+     case VIR_STORAGE_POOL_SHEEPDOG:
+     case VIR_STORAGE_POOL_GLUSTER:
+     case VIR_STORAGE_POOL_LAST:
+diff --git a/src/conf/storage_conf.c b/src/conf/storage_conf.c
+index 0c50529..fe97574 100644
+--- a/src/conf/storage_conf.c
++++ b/src/conf/storage_conf.c
+@@ -60,7 +60,7 @@ VIR_ENUM_IMPL(virStoragePool,
+               "logical", "disk", "iscsi",
+               "iscsi-direct", "scsi", "mpath",
+               "rbd", "sheepdog", "gluster",
+-              "zfs", "vstorage",
++              "zfs", "vstorage", "vitastor",
+ );
+ 
+ VIR_ENUM_IMPL(virStoragePoolFormatFileSystem,
+@@ -249,6 +249,18 @@ static virStoragePoolTypeInfo poolTypeInfo[] = {
+           .formatToString = virStorageFileFormatTypeToString,
+       }
+     },
++    {.poolType = VIR_STORAGE_POOL_VITASTOR,
++     .poolOptions = {
++         .flags = (VIR_STORAGE_POOL_SOURCE_HOST |
++                   VIR_STORAGE_POOL_SOURCE_NETWORK |
++                   VIR_STORAGE_POOL_SOURCE_NAME),
++      },
++      .volOptions = {
++          .defaultFormat = VIR_STORAGE_FILE_RAW,
++          .formatFromString = virStorageVolumeFormatFromString,
++          .formatToString = virStorageFileFormatTypeToString,
++      }
++    },
+     {.poolType = VIR_STORAGE_POOL_SHEEPDOG,
+      .poolOptions = {
+          .flags = (VIR_STORAGE_POOL_SOURCE_HOST |
+@@ -551,6 +563,11 @@ virStoragePoolDefParseSource(xmlXPathContextPtr ctxt,
+                        _("element 'name' is mandatory for RBD pool"));
+         goto cleanup;
+     }
++    if (pool_type == VIR_STORAGE_POOL_VITASTOR && source->name == NULL) {
++        virReportError(VIR_ERR_XML_ERROR, "%s",
++                       _("element 'name' is mandatory for Vitastor pool"));
++        return -1;
++    }
+ 
+     if (options->formatFromString) {
+         g_autofree char *format = NULL;
+@@ -1217,6 +1234,7 @@ virStoragePoolDefFormatBuf(virBufferPtr buf,
+     /* RBD, Sheepdog, Gluster and Iscsi-direct devices are not local block devs nor
+      * files, so they don't have a target */
+     if (def->type != VIR_STORAGE_POOL_RBD &&
++        def->type != VIR_STORAGE_POOL_VITASTOR &&
+         def->type != VIR_STORAGE_POOL_SHEEPDOG &&
+         def->type != VIR_STORAGE_POOL_GLUSTER &&
+         def->type != VIR_STORAGE_POOL_ISCSI_DIRECT) {
+diff --git a/src/conf/storage_conf.h b/src/conf/storage_conf.h
+index ffd406e..8868a05 100644
+--- a/src/conf/storage_conf.h
++++ b/src/conf/storage_conf.h
+@@ -110,6 +110,7 @@ typedef enum {
+     VIR_STORAGE_POOL_GLUSTER,  /* Gluster device */
+     VIR_STORAGE_POOL_ZFS,      /* ZFS */
+     VIR_STORAGE_POOL_VSTORAGE, /* Virtuozzo Storage */
++    VIR_STORAGE_POOL_VITASTOR, /* Vitastor */
+ 
+     VIR_STORAGE_POOL_LAST,
+ } virStoragePoolType;
+@@ -474,6 +475,7 @@ VIR_ENUM_DECL(virStoragePartedFs);
+                  VIR_CONNECT_LIST_STORAGE_POOLS_SCSI     | \
+                  VIR_CONNECT_LIST_STORAGE_POOLS_MPATH    | \
+                  VIR_CONNECT_LIST_STORAGE_POOLS_RBD      | \
++                 VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR | \
+                  VIR_CONNECT_LIST_STORAGE_POOLS_SHEEPDOG | \
+                  VIR_CONNECT_LIST_STORAGE_POOLS_GLUSTER  | \
+                  VIR_CONNECT_LIST_STORAGE_POOLS_ZFS      | \
+diff --git a/src/conf/virstorageobj.c b/src/conf/virstorageobj.c
+index 9fe8b3f..bf595b0 100644
+--- a/src/conf/virstorageobj.c
++++ b/src/conf/virstorageobj.c
+@@ -1491,6 +1491,7 @@ virStoragePoolObjSourceFindDuplicateCb(const void *payload,
+             return 1;
+         break;
+ 
++    case VIR_STORAGE_POOL_VITASTOR:
+     case VIR_STORAGE_POOL_RBD:
+     case VIR_STORAGE_POOL_LAST:
+         break;
+@@ -1990,6 +1991,8 @@ virStoragePoolObjMatch(virStoragePoolObjPtr obj,
+                (obj->def->type == VIR_STORAGE_POOL_MPATH))   ||
+               (MATCH(VIR_CONNECT_LIST_STORAGE_POOLS_RBD) &&
+                (obj->def->type == VIR_STORAGE_POOL_RBD))     ||
++              (MATCH(VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR) &&
++               (obj->def->type == VIR_STORAGE_POOL_VITASTOR)) ||
+               (MATCH(VIR_CONNECT_LIST_STORAGE_POOLS_SHEEPDOG) &&
+                (obj->def->type == VIR_STORAGE_POOL_SHEEPDOG)) ||
+               (MATCH(VIR_CONNECT_LIST_STORAGE_POOLS_GLUSTER) &&
+diff --git a/src/libvirt-storage.c b/src/libvirt-storage.c
+index 2a7cdca..f756be1 100644
+--- a/src/libvirt-storage.c
++++ b/src/libvirt-storage.c
+@@ -92,6 +92,7 @@ virStoragePoolGetConnect(virStoragePoolPtr pool)
+  * VIR_CONNECT_LIST_STORAGE_POOLS_SCSI
+  * VIR_CONNECT_LIST_STORAGE_POOLS_MPATH
+  * VIR_CONNECT_LIST_STORAGE_POOLS_RBD
++ * VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR
+  * VIR_CONNECT_LIST_STORAGE_POOLS_SHEEPDOG
+  * VIR_CONNECT_LIST_STORAGE_POOLS_GLUSTER
+  * VIR_CONNECT_LIST_STORAGE_POOLS_ZFS
+diff --git a/src/libxl/libxl_conf.c b/src/libxl/libxl_conf.c
+index 6a8ae27..a735bc6 100644
+--- a/src/libxl/libxl_conf.c
++++ b/src/libxl/libxl_conf.c
+@@ -942,6 +942,7 @@ libxlMakeNetworkDiskSrcStr(virStorageSourcePtr src,
+     case VIR_STORAGE_NET_PROTOCOL_SSH:
+     case VIR_STORAGE_NET_PROTOCOL_VXHS:
+     case VIR_STORAGE_NET_PROTOCOL_NFS:
++    case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+     case VIR_STORAGE_NET_PROTOCOL_LAST:
+     case VIR_STORAGE_NET_PROTOCOL_NONE:
+         virReportError(VIR_ERR_NO_SUPPORT,
+diff --git a/src/libxl/xen_xl.c b/src/libxl/xen_xl.c
+index 17b93d0..c5a0084 100644
+--- a/src/libxl/xen_xl.c
++++ b/src/libxl/xen_xl.c
+@@ -1601,6 +1601,7 @@ xenFormatXLDiskSrcNet(virStorageSourcePtr src)
+     case VIR_STORAGE_NET_PROTOCOL_SSH:
+     case VIR_STORAGE_NET_PROTOCOL_VXHS:
+     case VIR_STORAGE_NET_PROTOCOL_NFS:
++    case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+     case VIR_STORAGE_NET_PROTOCOL_LAST:
+     case VIR_STORAGE_NET_PROTOCOL_NONE:
+         virReportError(VIR_ERR_NO_SUPPORT,
+diff --git a/src/qemu/qemu_block.c b/src/qemu/qemu_block.c
+index f9c6da2..922dde5 100644
+--- a/src/qemu/qemu_block.c
++++ b/src/qemu/qemu_block.c
+@@ -938,6 +938,38 @@ qemuBlockStorageSourceGetRBDProps(virStorageSourcePtr src,
+ }
+ 
+ 
++static virJSONValuePtr
++qemuBlockStorageSourceGetVitastorProps(virStorageSource *src)
++{
++    virJSONValuePtr ret = NULL;
++    virStorageNetHostDefPtr host;
++    size_t i;
++    g_auto(virBuffer) buf = VIR_BUFFER_INITIALIZER;
++    g_autofree char *etcd = NULL;
++
++    for (i = 0; i < src->nhosts; i++) {
++        host = src->hosts + i;
++        if ((virStorageNetHostTransport)host->transport != VIR_STORAGE_NET_HOST_TRANS_TCP) {
++            return NULL;
++        }
++        virBufferAsprintf(&buf, i > 0 ? ",%s:%u" : "%s:%u", host->name, host->port);
++    }
++    if (src->nhosts > 0) {
++        etcd = virBufferContentAndReset(&buf);
++    }
++
++    if (virJSONValueObjectCreate(&ret,
++                                 "S:etcd_host", etcd,
++                                 "S:etcd_prefix", src->query,
++                                 "S:config_path", src->configFile,
++                                 "s:image", src->path,
++                                 NULL) < 0)
++        return NULL;
++
++    return ret;
++}
++
++
+ static virJSONValuePtr
+ qemuBlockStorageSourceGetSheepdogProps(virStorageSourcePtr src)
+ {
+@@ -1224,6 +1256,12 @@ qemuBlockStorageSourceGetBackendProps(virStorageSourcePtr src,
+                 return NULL;
+             break;
+ 
++        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
++            driver = "vitastor";
++            if (!(fileprops = qemuBlockStorageSourceGetVitastorProps(src)))
++                return NULL;
++            break;
++
+         case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+             driver = "sheepdog";
+             if (!(fileprops = qemuBlockStorageSourceGetSheepdogProps(src)))
+@@ -2183,6 +2221,7 @@ qemuBlockGetBackingStoreString(virStorageSourcePtr src,
+ 
+             case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+             case VIR_STORAGE_NET_PROTOCOL_RBD:
++            case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+             case VIR_STORAGE_NET_PROTOCOL_VXHS:
+             case VIR_STORAGE_NET_PROTOCOL_NFS:
+             case VIR_STORAGE_NET_PROTOCOL_SSH:
+@@ -2560,6 +2599,12 @@ qemuBlockStorageSourceCreateGetStorageProps(virStorageSourcePtr src,
+                 return -1;
+             break;
+ 
++        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
++            driver = "vitastor";
++            if (!(location = qemuBlockStorageSourceGetVitastorProps(src)))
++                return -1;
++            break;
++
+         case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+             driver = "sheepdog";
+             if (!(location = qemuBlockStorageSourceGetSheepdogProps(src)))
+diff --git a/src/qemu/qemu_command.c b/src/qemu/qemu_command.c
+index 6f970a3..10b39ca 100644
+--- a/src/qemu/qemu_command.c
++++ b/src/qemu/qemu_command.c
+@@ -1034,6 +1034,43 @@ qemuBuildNetworkDriveStr(virStorageSourcePtr src,
+             ret = virBufferContentAndReset(&buf);
+             break;
+ 
++        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
++            if (strchr(src->path, ':')) {
++                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
++                               _("':' not allowed in Vitastor source volume name '%s'"),
++                               src->path);
++                return NULL;
++            }
++
++            virBufferStrcat(&buf, "vitastor:image=", src->path, NULL);
++
++            if (src->nhosts > 0) {
++                virBufferAddLit(&buf, ":etcd_host=");
++                for (i = 0; i < src->nhosts; i++) {
++                    if (i)
++                        virBufferAddLit(&buf, ",");
++
++                    /* assume host containing : is ipv6 */
++                    if (strchr(src->hosts[i].name, ':'))
++                        virBufferEscape(&buf, '\\', ":", "[%s]",
++                                        src->hosts[i].name);
++                    else
++                        virBufferAsprintf(&buf, "%s", src->hosts[i].name);
++
++                    if (src->hosts[i].port)
++                        virBufferAsprintf(&buf, "\\:%u", src->hosts[i].port);
++                }
++            }
++
++            if (src->configFile)
++                virBufferEscape(&buf, '\\', ":", ":config_path=%s", src->configFile);
++
++            if (src->query)
++                virBufferEscape(&buf, '\\', ":", ":etcd_prefix=%s", src->query);
++
++            ret = virBufferContentAndReset(&buf);
++            break;
++
+         case VIR_STORAGE_NET_PROTOCOL_VXHS:
+             virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
+                            _("VxHS protocol does not support URI syntax"));
+diff --git a/src/qemu/qemu_domain.c b/src/qemu/qemu_domain.c
+index 0765dc7..4cff344 100644
+--- a/src/qemu/qemu_domain.c
++++ b/src/qemu/qemu_domain.c
+@@ -4610,7 +4610,8 @@ qemuDomainValidateStorageSource(virStorageSourcePtr src,
+     if (src->query &&
+         (actualType != VIR_STORAGE_TYPE_NETWORK ||
+          (src->protocol != VIR_STORAGE_NET_PROTOCOL_HTTPS &&
+-          src->protocol != VIR_STORAGE_NET_PROTOCOL_HTTP))) {
++          src->protocol != VIR_STORAGE_NET_PROTOCOL_HTTP &&
++          src->protocol != VIR_STORAGE_NET_PROTOCOL_VITASTOR))) {
+         virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
+                        _("query is supported only with HTTP(S) protocols"));
+         return -1;
+@@ -9704,6 +9705,7 @@ qemuDomainPrepareStorageSourceTLS(virStorageSourcePtr src,
+         break;
+ 
+     case VIR_STORAGE_NET_PROTOCOL_RBD:
++    case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+     case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+     case VIR_STORAGE_NET_PROTOCOL_GLUSTER:
+     case VIR_STORAGE_NET_PROTOCOL_ISCSI:
+diff --git a/src/qemu/qemu_snapshot.c b/src/qemu/qemu_snapshot.c
+index ee333c3..674aa58 100644
+--- a/src/qemu/qemu_snapshot.c
++++ b/src/qemu/qemu_snapshot.c
+@@ -403,6 +403,7 @@ qemuSnapshotPrepareDiskExternalInactive(virDomainSnapshotDiskDefPtr snapdisk,
+         case VIR_STORAGE_NET_PROTOCOL_NONE:
+         case VIR_STORAGE_NET_PROTOCOL_NBD:
+         case VIR_STORAGE_NET_PROTOCOL_RBD:
++        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+         case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+         case VIR_STORAGE_NET_PROTOCOL_GLUSTER:
+         case VIR_STORAGE_NET_PROTOCOL_ISCSI:
+@@ -493,6 +494,7 @@ qemuSnapshotPrepareDiskExternalActive(virDomainObjPtr vm,
+         case VIR_STORAGE_NET_PROTOCOL_NONE:
+         case VIR_STORAGE_NET_PROTOCOL_NBD:
+         case VIR_STORAGE_NET_PROTOCOL_RBD:
++        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+         case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+         case VIR_STORAGE_NET_PROTOCOL_ISCSI:
+         case VIR_STORAGE_NET_PROTOCOL_HTTP:
+@@ -623,6 +625,7 @@ qemuSnapshotPrepareDiskInternal(virDomainDiskDefPtr disk,
+         case VIR_STORAGE_NET_PROTOCOL_NONE:
+         case VIR_STORAGE_NET_PROTOCOL_NBD:
+         case VIR_STORAGE_NET_PROTOCOL_RBD:
++        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+         case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+         case VIR_STORAGE_NET_PROTOCOL_GLUSTER:
+         case VIR_STORAGE_NET_PROTOCOL_ISCSI:
+diff --git a/src/storage/storage_driver.c b/src/storage/storage_driver.c
+index 16bc53a..1e5d820 100644
+--- a/src/storage/storage_driver.c
++++ b/src/storage/storage_driver.c
+@@ -1645,6 +1645,7 @@ storageVolLookupByPathCallback(virStoragePoolObjPtr obj,
+ 
+         case VIR_STORAGE_POOL_GLUSTER:
+         case VIR_STORAGE_POOL_RBD:
++        case VIR_STORAGE_POOL_VITASTOR:
+         case VIR_STORAGE_POOL_SHEEPDOG:
+         case VIR_STORAGE_POOL_ZFS:
+         case VIR_STORAGE_POOL_LAST:
+diff --git a/src/test/test_driver.c b/src/test/test_driver.c
+index 29c4c86..a27ad94 100644
+--- a/src/test/test_driver.c
++++ b/src/test/test_driver.c
+@@ -7096,6 +7096,7 @@ testStorageVolumeTypeForPool(int pooltype)
+     case VIR_STORAGE_POOL_ISCSI_DIRECT:
+     case VIR_STORAGE_POOL_GLUSTER:
+     case VIR_STORAGE_POOL_RBD:
++    case VIR_STORAGE_POOL_VITASTOR:
+         return VIR_STORAGE_VOL_NETWORK;
+     case VIR_STORAGE_POOL_LOGICAL:
+     case VIR_STORAGE_POOL_DISK:
+diff --git a/src/util/virstoragefile.c b/src/util/virstoragefile.c
+index 0d3c2af..36e3afc 100644
+--- a/src/util/virstoragefile.c
++++ b/src/util/virstoragefile.c
+@@ -91,6 +91,7 @@ VIR_ENUM_IMPL(virStorageNetProtocol,
+               "ssh",
+               "vxhs",
+               "nfs",
++              "vitastor",
+ );
+ 
+ VIR_ENUM_IMPL(virStorageNetHostTransport,
+@@ -2880,6 +2881,75 @@ virStorageSourceParseRBDColonString(const char *rbdstr,
+ }
+ 
+ 
++static int
++virStorageSourceParseVitastorColonString(const char *colonstr,
++                                         virStorageSourcePtr src)
++{
++    char *p, *e, *next;
++    g_autofree char *options = NULL;
++
++    /* optionally skip the "vitastor:" prefix if provided */
++    if (STRPREFIX(colonstr, "vitastor:"))
++        colonstr += strlen("vitastor:");
++
++    options = g_strdup(colonstr);
++
++    p = options;
++    while (*p) {
++        /* find : delimiter or end of string */
++        for (e = p; *e && *e != ':'; ++e) {
++            if (*e == '\\') {
++                e++;
++                if (*e == '\0')
++                    break;
++            }
++        }
++        if (*e == '\0') {
++            next = e;    /* last kv pair */
++        } else {
++            next = e + 1;
++            *e = '\0';
++        }
++
++        if (STRPREFIX(p, "image=")) {
++            src->path = g_strdup(p + strlen("image="));
++        } else if (STRPREFIX(p, "etcd_prefix=")) {
++            src->query = g_strdup(p + strlen("etcd_prefix="));
++        } else if (STRPREFIX(p, "config_file=")) {
++            src->configFile = g_strdup(p + strlen("config_file="));
++        } else if (STRPREFIX(p, "etcd_host=")) {
++            char *h, *sep;
++
++            h = p + strlen("etcd_host=");
++            while (h < e) {
++                for (sep = h; sep < e; ++sep) {
++                    if (*sep == '\\' && (sep[1] == ',' ||
++                                         sep[1] == ';' ||
++                                         sep[1] == ' ')) {
++                        *sep = '\0';
++                        sep += 2;
++                        break;
++                    }
++                }
++
++                if (virStorageSourceRBDAddHost(src, h) < 0)
++                    return -1;
++
++                h = sep;
++            }
++        }
++
++        p = next;
++    }
++
++    if (!src->path) {
++        return -1;
++    }
++
++    return 0;
++}
++
++
+ static int
+ virStorageSourceParseNBDColonString(const char *nbdstr,
+                                     virStorageSourcePtr src)
+@@ -2992,6 +3062,11 @@ virStorageSourceParseBackingColon(virStorageSourcePtr src,
+             return -1;
+         break;
+ 
++    case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
++        if (virStorageSourceParseVitastorColonString(path, src) < 0)
++            return -1;
++        break;
++
+     case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+     case VIR_STORAGE_NET_PROTOCOL_LAST:
+     case VIR_STORAGE_NET_PROTOCOL_NONE:
+@@ -3581,6 +3656,54 @@ virStorageSourceParseBackingJSONRBD(virStorageSourcePtr src,
+     return 0;
+ }
+ 
++static int
++virStorageSourceParseBackingJSONVitastor(virStorageSourcePtr src,
++                                         virJSONValuePtr json,
++                                         const char *jsonstr G_GNUC_UNUSED,
++                                         int opaque G_GNUC_UNUSED)
++{
++    const char *filename;
++    const char *image = virJSONValueObjectGetString(json, "image");
++    const char *conf = virJSONValueObjectGetString(json, "config_path");
++    const char *etcd_prefix = virJSONValueObjectGetString(json, "etcd_prefix");
++    virJSONValuePtr servers = virJSONValueObjectGetArray(json, "server");
++    size_t nservers;
++    size_t i;
++
++    src->type = VIR_STORAGE_TYPE_NETWORK;
++    src->protocol = VIR_STORAGE_NET_PROTOCOL_VITASTOR;
++
++    /* legacy syntax passed via 'filename' option */
++    if ((filename = virJSONValueObjectGetString(json, "filename")))
++        return virStorageSourceParseVitastorColonString(filename, src);
++
++    if (!image) {
++        virReportError(VIR_ERR_INVALID_ARG, "%s",
++                       _("missing image name in Vitastor backing volume "
++                         "JSON specification"));
++        return -1;
++    }
++
++    src->path = g_strdup(image);
++    src->configFile = g_strdup(conf);
++    src->query = g_strdup(etcd_prefix);
++
++    if (servers) {
++        nservers = virJSONValueArraySize(servers);
++
++        src->hosts = g_new0(virStorageNetHostDef, nservers);
++        src->nhosts = nservers;
++
++        for (i = 0; i < nservers; i++) {
++            if (virStorageSourceParseBackingJSONInetSocketAddress(src->hosts + i,
++                                                                  virJSONValueArrayGet(servers, i)) < 0)
++                return -1;
++        }
++    }
++
++    return 0;
++}
++
+ static int
+ virStorageSourceParseBackingJSONRaw(virStorageSourcePtr src,
+                                     virJSONValuePtr json,
+@@ -3759,6 +3882,7 @@ static const struct virStorageSourceJSONDriverParser jsonParsers[] = {
+     {"sheepdog", false, virStorageSourceParseBackingJSONSheepdog, 0},
+     {"ssh", false, virStorageSourceParseBackingJSONSSH, 0},
+     {"rbd", false, virStorageSourceParseBackingJSONRBD, 0},
++    {"vitastor", false, virStorageSourceParseBackingJSONVitastor, 0},
+     {"raw", true, virStorageSourceParseBackingJSONRaw, 0},
+     {"nfs", false, virStorageSourceParseBackingJSONNFS, 0},
+     {"vxhs", false, virStorageSourceParseBackingJSONVxHS, 0},
+@@ -4503,6 +4627,7 @@ virStorageSourceNetworkDefaultPort(virStorageNetProtocol protocol)
+         case VIR_STORAGE_NET_PROTOCOL_GLUSTER:
+             return 24007;
+ 
++        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+         case VIR_STORAGE_NET_PROTOCOL_RBD:
+             /* we don't provide a default for RBD */
+             return 0;
+diff --git a/src/util/virstoragefile.h b/src/util/virstoragefile.h
+index 5689c39..3eb4e3c 100644
+--- a/src/util/virstoragefile.h
++++ b/src/util/virstoragefile.h
+@@ -136,6 +136,7 @@ typedef enum {
+     VIR_STORAGE_NET_PROTOCOL_SSH,
+     VIR_STORAGE_NET_PROTOCOL_VXHS,
+     VIR_STORAGE_NET_PROTOCOL_NFS,
++    VIR_STORAGE_NET_PROTOCOL_VITASTOR,
+ 
+     VIR_STORAGE_NET_PROTOCOL_LAST
+ } virStorageNetProtocol;
+diff --git a/tests/storagepoolcapsschemadata/poolcaps-fs.xml b/tests/storagepoolcapsschemadata/poolcaps-fs.xml
+index eee75af..8bd0a57 100644
+--- a/tests/storagepoolcapsschemadata/poolcaps-fs.xml
++++ b/tests/storagepoolcapsschemadata/poolcaps-fs.xml
+@@ -204,4 +204,11 @@
+       </enum>
+     </volOptions>
+   </pool>
++  <pool type='vitastor' supported='no'>
++    <volOptions>
++      <defaultFormat type='raw'/>
++      <enum name='targetFormatType'>
++      </enum>
++    </volOptions>
++  </pool>
+ </storagepoolCapabilities>
+diff --git a/tests/storagepoolcapsschemadata/poolcaps-full.xml b/tests/storagepoolcapsschemadata/poolcaps-full.xml
+index 805950a..852df0d 100644
+--- a/tests/storagepoolcapsschemadata/poolcaps-full.xml
++++ b/tests/storagepoolcapsschemadata/poolcaps-full.xml
+@@ -204,4 +204,11 @@
+       </enum>
+     </volOptions>
+   </pool>
++  <pool type='vitastor' supported='yes'>
++    <volOptions>
++      <defaultFormat type='raw'/>
++      <enum name='targetFormatType'>
++      </enum>
++    </volOptions>
++  </pool>
+ </storagepoolCapabilities>
+diff --git a/tests/storagepoolxml2argvtest.c b/tests/storagepoolxml2argvtest.c
+index 967d1f2..1e8ff7a 100644
+--- a/tests/storagepoolxml2argvtest.c
++++ b/tests/storagepoolxml2argvtest.c
+@@ -68,6 +68,7 @@ testCompareXMLToArgvFiles(bool shouldFail,
+     case VIR_STORAGE_POOL_GLUSTER:
+     case VIR_STORAGE_POOL_ZFS:
+     case VIR_STORAGE_POOL_VSTORAGE:
++    case VIR_STORAGE_POOL_VITASTOR:
+     case VIR_STORAGE_POOL_LAST:
+     default:
+         VIR_TEST_DEBUG("pool type '%s' has no xml2argv test", defTypeStr);
+diff --git a/tools/virsh-pool.c b/tools/virsh-pool.c
+index 7835fa6..8841fcf 100644
+--- a/tools/virsh-pool.c
++++ b/tools/virsh-pool.c
+@@ -1237,6 +1237,9 @@ cmdPoolList(vshControl *ctl, const vshCmd *cmd G_GNUC_UNUSED)
+             case VIR_STORAGE_POOL_VSTORAGE:
+                 flags |= VIR_CONNECT_LIST_STORAGE_POOLS_VSTORAGE;
+                 break;
++            case VIR_STORAGE_POOL_VITASTOR:
++                flags |= VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR;
++                break;
+             case VIR_STORAGE_POOL_LAST:
+                 break;
+             }

patches/libvirt-7.5-vitastor.diff

@@ -0,0 +1,661 @@
+commit c6e1958a1b4974828e8e5852beb252ce6594e670
+Author: Vitaliy Filippov <vitalif@yourcmc.ru>
+Date:   Mon Jun 28 01:20:19 2021 +0300
+
+    Add Vitastor support
+
+diff --git a/docs/schemas/domaincommon.rng b/docs/schemas/domaincommon.rng
+index 5ea14b6..a9df168 100644
+--- a/docs/schemas/domaincommon.rng
++++ b/docs/schemas/domaincommon.rng
+@@ -1859,6 +1859,35 @@
+     </element>
+   </define>
+ 
++  <define name="diskSourceNetworkProtocolVitastor">
++    <element name="source">
++      <interleave>
++        <attribute name="protocol">
++          <value>vitastor</value>
++        </attribute>
++        <ref name="diskSourceCommon"/>
++        <optional>
++          <attribute name="name"/>
++        </optional>
++        <optional>
++          <attribute name="query"/>
++        </optional>
++        <zeroOrMore>
++          <ref name="diskSourceNetworkHost"/>
++        </zeroOrMore>
++        <optional>
++          <element name="config">
++            <attribute name="file">
++              <ref name="absFilePath"/>
++            </attribute>
++            <empty/>
++          </element>
++        </optional>
++        <empty/>
++      </interleave>
++    </element>
++  </define>
++
+   <define name="diskSourceNetworkProtocolISCSI">
+     <element name="source">
+       <attribute name="protocol">
+@@ -2115,6 +2144,7 @@
+       <ref name="diskSourceNetworkProtocolSimple"/>
+       <ref name="diskSourceNetworkProtocolVxHS"/>
+       <ref name="diskSourceNetworkProtocolNFS"/>
++      <ref name="diskSourceNetworkProtocolVitastor"/>
+     </choice>
+   </define>
+ 
+diff --git a/include/libvirt/libvirt-storage.h b/include/libvirt/libvirt-storage.h
+index 089e1e0..d7e7ef4 100644
+--- a/include/libvirt/libvirt-storage.h
++++ b/include/libvirt/libvirt-storage.h
+@@ -245,6 +245,7 @@ typedef enum {
+     VIR_CONNECT_LIST_STORAGE_POOLS_ZFS           = 1 << 17,
+     VIR_CONNECT_LIST_STORAGE_POOLS_VSTORAGE      = 1 << 18,
+     VIR_CONNECT_LIST_STORAGE_POOLS_ISCSI_DIRECT  = 1 << 19,
++    VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR      = 1 << 20,
+ } virConnectListAllStoragePoolsFlags;
+ 
+ int                     virConnectListAllStoragePools(virConnectPtr conn,
+diff --git a/src/conf/domain_conf.c b/src/conf/domain_conf.c
+index d78f846..f7222e3 100644
+--- a/src/conf/domain_conf.c
++++ b/src/conf/domain_conf.c
+@@ -8251,7 +8251,8 @@ virDomainDiskSourceNetworkParse(xmlNodePtr node,
+     src->configFile = virXPathString("string(./config/@file)", ctxt);
+ 
+     if (src->protocol == VIR_STORAGE_NET_PROTOCOL_HTTP ||
+-        src->protocol == VIR_STORAGE_NET_PROTOCOL_HTTPS)
++        src->protocol == VIR_STORAGE_NET_PROTOCOL_HTTPS ||
++        src->protocol == VIR_STORAGE_NET_PROTOCOL_VITASTOR)
+         src->query = virXMLPropString(node, "query");
+ 
+     if (virDomainStorageNetworkParseHosts(node, ctxt, &src->hosts, &src->nhosts) < 0)
+@@ -30775,6 +30776,7 @@ virDomainStorageSourceTranslateSourcePool(virStorageSource *src,
+ 
+     case VIR_STORAGE_POOL_MPATH:
+     case VIR_STORAGE_POOL_RBD:
++    case VIR_STORAGE_POOL_VITASTOR:
+     case VIR_STORAGE_POOL_SHEEPDOG:
+     case VIR_STORAGE_POOL_GLUSTER:
+     case VIR_STORAGE_POOL_LAST:
+diff --git a/src/conf/storage_conf.c b/src/conf/storage_conf.c
+index 2aa9a3d..166ca1f 100644
+--- a/src/conf/storage_conf.c
++++ b/src/conf/storage_conf.c
+@@ -60,7 +60,7 @@ VIR_ENUM_IMPL(virStoragePool,
+               "logical", "disk", "iscsi",
+               "iscsi-direct", "scsi", "mpath",
+               "rbd", "sheepdog", "gluster",
+-              "zfs", "vstorage",
++              "zfs", "vstorage", "vitastor",
+ );
+ 
+ VIR_ENUM_IMPL(virStoragePoolFormatFileSystem,
+@@ -246,6 +246,18 @@ static virStoragePoolTypeInfo poolTypeInfo[] = {
+           .formatToString = virStorageFileFormatTypeToString,
+       }
+     },
++    {.poolType = VIR_STORAGE_POOL_VITASTOR,
++     .poolOptions = {
++         .flags = (VIR_STORAGE_POOL_SOURCE_HOST |
++                   VIR_STORAGE_POOL_SOURCE_NETWORK |
++                   VIR_STORAGE_POOL_SOURCE_NAME),
++      },
++      .volOptions = {
++          .defaultFormat = VIR_STORAGE_FILE_RAW,
++          .formatFromString = virStorageVolumeFormatFromString,
++          .formatToString = virStorageFileFormatTypeToString,
++      }
++    },
+     {.poolType = VIR_STORAGE_POOL_SHEEPDOG,
+      .poolOptions = {
+          .flags = (VIR_STORAGE_POOL_SOURCE_HOST |
+@@ -546,6 +558,11 @@ virStoragePoolDefParseSource(xmlXPathContextPtr ctxt,
+                        _("element 'name' is mandatory for RBD pool"));
+         return -1;
+     }
++    if (pool_type == VIR_STORAGE_POOL_VITASTOR && source->name == NULL) {
++        virReportError(VIR_ERR_XML_ERROR, "%s",
++                       _("element 'name' is mandatory for Vitastor pool"));
++        return -1;
++    }
+ 
+     if (options->formatFromString) {
+         g_autofree char *format = NULL;
+@@ -1182,6 +1199,7 @@ virStoragePoolDefFormatBuf(virBuffer *buf,
+     /* RBD, Sheepdog, Gluster and Iscsi-direct devices are not local block devs nor
+      * files, so they don't have a target */
+     if (def->type != VIR_STORAGE_POOL_RBD &&
++        def->type != VIR_STORAGE_POOL_VITASTOR &&
+         def->type != VIR_STORAGE_POOL_SHEEPDOG &&
+         def->type != VIR_STORAGE_POOL_GLUSTER &&
+         def->type != VIR_STORAGE_POOL_ISCSI_DIRECT) {
+diff --git a/src/conf/storage_conf.h b/src/conf/storage_conf.h
+index 76efaac..928149a 100644
+--- a/src/conf/storage_conf.h
++++ b/src/conf/storage_conf.h
+@@ -106,6 +106,7 @@ typedef enum {
+     VIR_STORAGE_POOL_GLUSTER,  /* Gluster device */
+     VIR_STORAGE_POOL_ZFS,      /* ZFS */
+     VIR_STORAGE_POOL_VSTORAGE, /* Virtuozzo Storage */
++    VIR_STORAGE_POOL_VITASTOR, /* Vitastor */
+ 
+     VIR_STORAGE_POOL_LAST,
+ } virStoragePoolType;
+@@ -465,6 +466,7 @@ VIR_ENUM_DECL(virStoragePartedFs);
+                  VIR_CONNECT_LIST_STORAGE_POOLS_SCSI     | \
+                  VIR_CONNECT_LIST_STORAGE_POOLS_MPATH    | \
+                  VIR_CONNECT_LIST_STORAGE_POOLS_RBD      | \
++                 VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR | \
+                  VIR_CONNECT_LIST_STORAGE_POOLS_SHEEPDOG | \
+                  VIR_CONNECT_LIST_STORAGE_POOLS_GLUSTER  | \
+                  VIR_CONNECT_LIST_STORAGE_POOLS_ZFS      | \
+diff --git a/src/conf/storage_source_conf.c b/src/conf/storage_source_conf.c
+index 5ca06fa..05ded49 100644
+--- a/src/conf/storage_source_conf.c
++++ b/src/conf/storage_source_conf.c
+@@ -85,6 +85,7 @@ VIR_ENUM_IMPL(virStorageNetProtocol,
+               "ssh",
+               "vxhs",
+               "nfs",
++              "vitastor",
+ );
+ 
+ 
+@@ -1262,6 +1263,7 @@ virStorageSourceNetworkDefaultPort(virStorageNetProtocol protocol)
+         case VIR_STORAGE_NET_PROTOCOL_GLUSTER:
+             return 24007;
+ 
++        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+         case VIR_STORAGE_NET_PROTOCOL_RBD:
+             /* we don't provide a default for RBD */
+             return 0;
+diff --git a/src/conf/storage_source_conf.h b/src/conf/storage_source_conf.h
+index 389c7b5..dbf02e3 100644
+--- a/src/conf/storage_source_conf.h
++++ b/src/conf/storage_source_conf.h
+@@ -127,6 +127,7 @@ typedef enum {
+     VIR_STORAGE_NET_PROTOCOL_SSH,
+     VIR_STORAGE_NET_PROTOCOL_VXHS,
+     VIR_STORAGE_NET_PROTOCOL_NFS,
++    VIR_STORAGE_NET_PROTOCOL_VITASTOR,
+ 
+     VIR_STORAGE_NET_PROTOCOL_LAST
+ } virStorageNetProtocol;
+diff --git a/src/conf/virstorageobj.c b/src/conf/virstorageobj.c
+index 24957d6..4520a73 100644
+--- a/src/conf/virstorageobj.c
++++ b/src/conf/virstorageobj.c
+@@ -1487,6 +1487,7 @@ virStoragePoolObjSourceFindDuplicateCb(const void *payload,
+             return 1;
+         break;
+ 
++    case VIR_STORAGE_POOL_VITASTOR:
+     case VIR_STORAGE_POOL_RBD:
+     case VIR_STORAGE_POOL_LAST:
+         break;
+@@ -1986,6 +1987,8 @@ virStoragePoolObjMatch(virStoragePoolObj *obj,
+                (obj->def->type == VIR_STORAGE_POOL_MPATH))   ||
+               (MATCH(VIR_CONNECT_LIST_STORAGE_POOLS_RBD) &&
+                (obj->def->type == VIR_STORAGE_POOL_RBD))     ||
++              (MATCH(VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR) &&
++               (obj->def->type == VIR_STORAGE_POOL_VITASTOR)) ||
+               (MATCH(VIR_CONNECT_LIST_STORAGE_POOLS_SHEEPDOG) &&
+                (obj->def->type == VIR_STORAGE_POOL_SHEEPDOG)) ||
+               (MATCH(VIR_CONNECT_LIST_STORAGE_POOLS_GLUSTER) &&
+diff --git a/src/libvirt-storage.c b/src/libvirt-storage.c
+index 2a7cdca..f756be1 100644
+--- a/src/libvirt-storage.c
++++ b/src/libvirt-storage.c
+@@ -92,6 +92,7 @@ virStoragePoolGetConnect(virStoragePoolPtr pool)
+  * VIR_CONNECT_LIST_STORAGE_POOLS_SCSI
+  * VIR_CONNECT_LIST_STORAGE_POOLS_MPATH
+  * VIR_CONNECT_LIST_STORAGE_POOLS_RBD
++ * VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR
+  * VIR_CONNECT_LIST_STORAGE_POOLS_SHEEPDOG
+  * VIR_CONNECT_LIST_STORAGE_POOLS_GLUSTER
+  * VIR_CONNECT_LIST_STORAGE_POOLS_ZFS
+diff --git a/src/libxl/libxl_conf.c b/src/libxl/libxl_conf.c
+index 56cb9ab..dfb31b9 100644
+--- a/src/libxl/libxl_conf.c
++++ b/src/libxl/libxl_conf.c
+@@ -972,6 +972,7 @@ libxlMakeNetworkDiskSrcStr(virStorageSource *src,
+     case VIR_STORAGE_NET_PROTOCOL_SSH:
+     case VIR_STORAGE_NET_PROTOCOL_VXHS:
+     case VIR_STORAGE_NET_PROTOCOL_NFS:
++    case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+     case VIR_STORAGE_NET_PROTOCOL_LAST:
+     case VIR_STORAGE_NET_PROTOCOL_NONE:
+         virReportError(VIR_ERR_NO_SUPPORT,
+diff --git a/src/libxl/xen_xl.c b/src/libxl/xen_xl.c
+index c0905b0..c172378 100644
+--- a/src/libxl/xen_xl.c
++++ b/src/libxl/xen_xl.c
+@@ -1540,6 +1540,7 @@ xenFormatXLDiskSrcNet(virStorageSource *src)
+     case VIR_STORAGE_NET_PROTOCOL_SSH:
+     case VIR_STORAGE_NET_PROTOCOL_VXHS:
+     case VIR_STORAGE_NET_PROTOCOL_NFS:
++    case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+     case VIR_STORAGE_NET_PROTOCOL_LAST:
+     case VIR_STORAGE_NET_PROTOCOL_NONE:
+         virReportError(VIR_ERR_NO_SUPPORT,
+diff --git a/src/qemu/qemu_block.c b/src/qemu/qemu_block.c
+index 6627d04..c33f428 100644
+--- a/src/qemu/qemu_block.c
++++ b/src/qemu/qemu_block.c
+@@ -928,6 +928,38 @@ qemuBlockStorageSourceGetRBDProps(virStorageSource *src,
+ }
+ 
+ 
++static virJSONValue *
++qemuBlockStorageSourceGetVitastorProps(virStorageSource *src)
++{
++    virJSONValuePtr ret = NULL;
++    virStorageNetHostDefPtr host;
++    size_t i;
++    g_auto(virBuffer) buf = VIR_BUFFER_INITIALIZER;
++    g_autofree char *etcd = NULL;
++
++    for (i = 0; i < src->nhosts; i++) {
++        host = src->hosts + i;
++        if ((virStorageNetHostTransport)host->transport != VIR_STORAGE_NET_HOST_TRANS_TCP) {
++            return NULL;
++        }
++        virBufferAsprintf(&buf, i > 0 ? ",%s:%u" : "%s:%u", host->name, host->port);
++    }
++    if (src->nhosts > 0) {
++        etcd = virBufferContentAndReset(&buf);
++    }
++
++    if (virJSONValueObjectCreate(&ret,
++                                 "S:etcd_host", etcd,
++                                 "S:etcd_prefix", src->query,
++                                 "S:config_path", src->configFile,
++                                 "s:image", src->path,
++                                 NULL) < 0)
++        return NULL;
++
++    return ret;
++}
++
++
+ static virJSONValue *
+ qemuBlockStorageSourceGetSheepdogProps(virStorageSource *src)
+ {
+@@ -1218,6 +1250,12 @@ qemuBlockStorageSourceGetBackendProps(virStorageSource *src,
+                 return NULL;
+             break;
+ 
++        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
++            driver = "vitastor";
++            if (!(fileprops = qemuBlockStorageSourceGetVitastorProps(src)))
++                return NULL;
++            break;
++
+         case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+             driver = "sheepdog";
+             if (!(fileprops = qemuBlockStorageSourceGetSheepdogProps(src)))
+@@ -2231,6 +2269,7 @@ qemuBlockGetBackingStoreString(virStorageSource *src,
+ 
+             case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+             case VIR_STORAGE_NET_PROTOCOL_RBD:
++            case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+             case VIR_STORAGE_NET_PROTOCOL_VXHS:
+             case VIR_STORAGE_NET_PROTOCOL_NFS:
+             case VIR_STORAGE_NET_PROTOCOL_SSH:
+@@ -2608,6 +2647,12 @@ qemuBlockStorageSourceCreateGetStorageProps(virStorageSource *src,
+                 return -1;
+             break;
+ 
++        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
++            driver = "vitastor";
++            if (!(location = qemuBlockStorageSourceGetVitastorProps(src)))
++                return -1;
++            break;
++
+         case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+             driver = "sheepdog";
+             if (!(location = qemuBlockStorageSourceGetSheepdogProps(src)))
+diff --git a/src/qemu/qemu_command.c b/src/qemu/qemu_command.c
+index ea51369..8258632 100644
+--- a/src/qemu/qemu_command.c
++++ b/src/qemu/qemu_command.c
+@@ -1074,6 +1074,43 @@ qemuBuildNetworkDriveStr(virStorageSource *src,
+             ret = virBufferContentAndReset(&buf);
+             break;
+ 
++        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
++            if (strchr(src->path, ':')) {
++                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
++                               _("':' not allowed in Vitastor source volume name '%s'"),
++                               src->path);
++                return NULL;
++            }
++
++            virBufferStrcat(&buf, "vitastor:image=", src->path, NULL);
++
++            if (src->nhosts > 0) {
++                virBufferAddLit(&buf, ":etcd_host=");
++                for (i = 0; i < src->nhosts; i++) {
++                    if (i)
++                        virBufferAddLit(&buf, ",");
++
++                    /* assume host containing : is ipv6 */
++                    if (strchr(src->hosts[i].name, ':'))
++                        virBufferEscape(&buf, '\\', ":", "[%s]",
++                                        src->hosts[i].name);
++                    else
++                        virBufferAsprintf(&buf, "%s", src->hosts[i].name);
++
++                    if (src->hosts[i].port)
++                        virBufferAsprintf(&buf, "\\:%u", src->hosts[i].port);
++                }
++            }
++
++            if (src->configFile)
++                virBufferEscape(&buf, '\\', ":", ":config_path=%s", src->configFile);
++
++            if (src->query)
++                virBufferEscape(&buf, '\\', ":", ":etcd_prefix=%s", src->query);
++
++            ret = virBufferContentAndReset(&buf);
++            break;
++
+         case VIR_STORAGE_NET_PROTOCOL_VXHS:
+             virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
+                            _("VxHS protocol does not support URI syntax"));
+diff --git a/src/qemu/qemu_domain.c b/src/qemu/qemu_domain.c
+index fc60e15..5ab410d 100644
+--- a/src/qemu/qemu_domain.c
++++ b/src/qemu/qemu_domain.c
+@@ -4829,7 +4829,8 @@ qemuDomainValidateStorageSource(virStorageSource *src,
+     if (src->query &&
+         (actualType != VIR_STORAGE_TYPE_NETWORK ||
+          (src->protocol != VIR_STORAGE_NET_PROTOCOL_HTTPS &&
+-          src->protocol != VIR_STORAGE_NET_PROTOCOL_HTTP))) {
++          src->protocol != VIR_STORAGE_NET_PROTOCOL_HTTP &&
++          src->protocol != VIR_STORAGE_NET_PROTOCOL_VITASTOR))) {
+         virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
+                        _("query is supported only with HTTP(S) protocols"));
+         return -1;
+@@ -10027,6 +10028,7 @@ qemuDomainPrepareStorageSourceTLS(virStorageSource *src,
+         break;
+ 
+     case VIR_STORAGE_NET_PROTOCOL_RBD:
++    case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+     case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+     case VIR_STORAGE_NET_PROTOCOL_GLUSTER:
+     case VIR_STORAGE_NET_PROTOCOL_ISCSI:
+diff --git a/src/qemu/qemu_snapshot.c b/src/qemu/qemu_snapshot.c
+index 4e74ddd..14e5f2e 100644
+--- a/src/qemu/qemu_snapshot.c
++++ b/src/qemu/qemu_snapshot.c
+@@ -402,6 +402,7 @@ qemuSnapshotPrepareDiskExternalInactive(virDomainSnapshotDiskDef *snapdisk,
+         case VIR_STORAGE_NET_PROTOCOL_NONE:
+         case VIR_STORAGE_NET_PROTOCOL_NBD:
+         case VIR_STORAGE_NET_PROTOCOL_RBD:
++        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+         case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+         case VIR_STORAGE_NET_PROTOCOL_GLUSTER:
+         case VIR_STORAGE_NET_PROTOCOL_ISCSI:
+@@ -494,6 +495,7 @@ qemuSnapshotPrepareDiskExternalActive(virDomainObj *vm,
+         case VIR_STORAGE_NET_PROTOCOL_NONE:
+         case VIR_STORAGE_NET_PROTOCOL_NBD:
+         case VIR_STORAGE_NET_PROTOCOL_RBD:
++        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+         case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+         case VIR_STORAGE_NET_PROTOCOL_ISCSI:
+         case VIR_STORAGE_NET_PROTOCOL_HTTP:
+@@ -647,6 +649,7 @@ qemuSnapshotPrepareDiskInternal(virDomainDiskDef *disk,
+         case VIR_STORAGE_NET_PROTOCOL_NONE:
+         case VIR_STORAGE_NET_PROTOCOL_NBD:
+         case VIR_STORAGE_NET_PROTOCOL_RBD:
++        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+         case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+         case VIR_STORAGE_NET_PROTOCOL_GLUSTER:
+         case VIR_STORAGE_NET_PROTOCOL_ISCSI:
+diff --git a/src/storage/storage_driver.c b/src/storage/storage_driver.c
+index c2ff4b8..70d0689 100644
+--- a/src/storage/storage_driver.c
++++ b/src/storage/storage_driver.c
+@@ -1644,6 +1644,7 @@ storageVolLookupByPathCallback(virStoragePoolObj *obj,
+ 
+         case VIR_STORAGE_POOL_GLUSTER:
+         case VIR_STORAGE_POOL_RBD:
++        case VIR_STORAGE_POOL_VITASTOR:
+         case VIR_STORAGE_POOL_SHEEPDOG:
+         case VIR_STORAGE_POOL_ZFS:
+         case VIR_STORAGE_POOL_LAST:
+diff --git a/src/storage_file/storage_source_backingstore.c b/src/storage_file/storage_source_backingstore.c
+index e48ae72..d7a9b72 100644
+--- a/src/storage_file/storage_source_backingstore.c
++++ b/src/storage_file/storage_source_backingstore.c
+@@ -284,6 +284,75 @@ virStorageSourceParseRBDColonString(const char *rbdstr,
+ }
+ 
+ 
++static int
++virStorageSourceParseVitastorColonString(const char *colonstr,
++                                         virStorageSource *src)
++{
++    char *p, *e, *next;
++    g_autofree char *options = NULL;
++
++    /* optionally skip the "vitastor:" prefix if provided */
++    if (STRPREFIX(colonstr, "vitastor:"))
++        colonstr += strlen("vitastor:");
++
++    options = g_strdup(colonstr);
++
++    p = options;
++    while (*p) {
++        /* find : delimiter or end of string */
++        for (e = p; *e && *e != ':'; ++e) {
++            if (*e == '\\') {
++                e++;
++                if (*e == '\0')
++                    break;
++            }
++        }
++        if (*e == '\0') {
++            next = e;    /* last kv pair */
++        } else {
++            next = e + 1;
++            *e = '\0';
++        }
++
++        if (STRPREFIX(p, "image=")) {
++            src->path = g_strdup(p + strlen("image="));
++        } else if (STRPREFIX(p, "etcd_prefix=")) {
++            src->query = g_strdup(p + strlen("etcd_prefix="));
++        } else if (STRPREFIX(p, "config_file=")) {
++            src->configFile = g_strdup(p + strlen("config_file="));
++        } else if (STRPREFIX(p, "etcd_host=")) {
++            char *h, *sep;
++
++            h = p + strlen("etcd_host=");
++            while (h < e) {
++                for (sep = h; sep < e; ++sep) {
++                    if (*sep == '\\' && (sep[1] == ',' ||
++                                         sep[1] == ';' ||
++                                         sep[1] == ' ')) {
++                        *sep = '\0';
++                        sep += 2;
++                        break;
++                    }
++                }
++
++                if (virStorageSourceRBDAddHost(src, h) < 0)
++                    return -1;
++
++                h = sep;
++            }
++        }
++
++        p = next;
++    }
++
++    if (!src->path) {
++        return -1;
++    }
++
++    return 0;
++}
++
++
+ static int
+ virStorageSourceParseNBDColonString(const char *nbdstr,
+                                     virStorageSource *src)
+@@ -396,6 +465,11 @@ virStorageSourceParseBackingColon(virStorageSource *src,
+             return -1;
+         break;
+ 
++    case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
++        if (virStorageSourceParseVitastorColonString(path, src) < 0)
++            return -1;
++        break;
++
+     case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+     case VIR_STORAGE_NET_PROTOCOL_LAST:
+     case VIR_STORAGE_NET_PROTOCOL_NONE:
+@@ -984,6 +1058,54 @@ virStorageSourceParseBackingJSONRBD(virStorageSource *src,
+     return 0;
+ }
+ 
++static int
++virStorageSourceParseBackingJSONVitastor(virStorageSource *src,
++                                         virJSONValue *json,
++                                         const char *jsonstr G_GNUC_UNUSED,
++                                         int opaque G_GNUC_UNUSED)
++{
++    const char *filename;
++    const char *image = virJSONValueObjectGetString(json, "image");
++    const char *conf = virJSONValueObjectGetString(json, "config_path");
++    const char *etcd_prefix = virJSONValueObjectGetString(json, "etcd_prefix");
++    virJSONValue *servers = virJSONValueObjectGetArray(json, "server");
++    size_t nservers;
++    size_t i;
++
++    src->type = VIR_STORAGE_TYPE_NETWORK;
++    src->protocol = VIR_STORAGE_NET_PROTOCOL_VITASTOR;
++
++    /* legacy syntax passed via 'filename' option */
++    if ((filename = virJSONValueObjectGetString(json, "filename")))
++        return virStorageSourceParseVitastorColonString(filename, src);
++
++    if (!image) {
++        virReportError(VIR_ERR_INVALID_ARG, "%s",
++                       _("missing image name in Vitastor backing volume "
++                         "JSON specification"));
++        return -1;
++    }
++
++    src->path = g_strdup(image);
++    src->configFile = g_strdup(conf);
++    src->query = g_strdup(etcd_prefix);
++
++    if (servers) {
++        nservers = virJSONValueArraySize(servers);
++
++        src->hosts = g_new0(virStorageNetHostDef, nservers);
++        src->nhosts = nservers;
++
++        for (i = 0; i < nservers; i++) {
++            if (virStorageSourceParseBackingJSONInetSocketAddress(src->hosts + i,
++                                                                  virJSONValueArrayGet(servers, i)) < 0)
++                return -1;
++        }
++    }
++
++    return 0;
++}
++
+ static int
+ virStorageSourceParseBackingJSONRaw(virStorageSource *src,
+                                     virJSONValue *json,
+@@ -1162,6 +1284,7 @@ static const struct virStorageSourceJSONDriverParser jsonParsers[] = {
+     {"sheepdog", false, virStorageSourceParseBackingJSONSheepdog, 0},
+     {"ssh", false, virStorageSourceParseBackingJSONSSH, 0},
+     {"rbd", false, virStorageSourceParseBackingJSONRBD, 0},
++    {"vitastor", false, virStorageSourceParseBackingJSONVitastor, 0},
+     {"raw", true, virStorageSourceParseBackingJSONRaw, 0},
+     {"nfs", false, virStorageSourceParseBackingJSONNFS, 0},
+     {"vxhs", false, virStorageSourceParseBackingJSONVxHS, 0},
+diff --git a/src/test/test_driver.c b/src/test/test_driver.c
+index ef0ddab..2173dc3 100644
+--- a/src/test/test_driver.c
++++ b/src/test/test_driver.c
+@@ -7131,6 +7131,7 @@ testStorageVolumeTypeForPool(int pooltype)
+     case VIR_STORAGE_POOL_ISCSI_DIRECT:
+     case VIR_STORAGE_POOL_GLUSTER:
+     case VIR_STORAGE_POOL_RBD:
++    case VIR_STORAGE_POOL_VITASTOR:
+         return VIR_STORAGE_VOL_NETWORK;
+     case VIR_STORAGE_POOL_LOGICAL:
+     case VIR_STORAGE_POOL_DISK:
+diff --git a/tests/storagepoolcapsschemadata/poolcaps-fs.xml b/tests/storagepoolcapsschemadata/poolcaps-fs.xml
+index eee75af..8bd0a57 100644
+--- a/tests/storagepoolcapsschemadata/poolcaps-fs.xml
++++ b/tests/storagepoolcapsschemadata/poolcaps-fs.xml
+@@ -204,4 +204,11 @@
+       </enum>
+     </volOptions>
+   </pool>
++  <pool type='vitastor' supported='no'>
++    <volOptions>
++      <defaultFormat type='raw'/>
++      <enum name='targetFormatType'>
++      </enum>
++    </volOptions>
++  </pool>
+ </storagepoolCapabilities>
+diff --git a/tests/storagepoolcapsschemadata/poolcaps-full.xml b/tests/storagepoolcapsschemadata/poolcaps-full.xml
+index 805950a..852df0d 100644
+--- a/tests/storagepoolcapsschemadata/poolcaps-full.xml
++++ b/tests/storagepoolcapsschemadata/poolcaps-full.xml
+@@ -204,4 +204,11 @@
+       </enum>
+     </volOptions>
+   </pool>
++  <pool type='vitastor' supported='yes'>
++    <volOptions>
++      <defaultFormat type='raw'/>
++      <enum name='targetFormatType'>
++      </enum>
++    </volOptions>
++  </pool>
+ </storagepoolCapabilities>
+diff --git a/tests/storagepoolxml2argvtest.c b/tests/storagepoolxml2argvtest.c
+index 449b745..7f95cc8 100644
+--- a/tests/storagepoolxml2argvtest.c
++++ b/tests/storagepoolxml2argvtest.c
+@@ -68,6 +68,7 @@ testCompareXMLToArgvFiles(bool shouldFail,
+     case VIR_STORAGE_POOL_GLUSTER:
+     case VIR_STORAGE_POOL_ZFS:
+     case VIR_STORAGE_POOL_VSTORAGE:
++    case VIR_STORAGE_POOL_VITASTOR:
+     case VIR_STORAGE_POOL_LAST:
+     default:
+         VIR_TEST_DEBUG("pool type '%s' has no xml2argv test", defTypeStr);
+diff --git a/tools/virsh-pool.c b/tools/virsh-pool.c
+index 18f3839..c8e1436 100644
+--- a/tools/virsh-pool.c
++++ b/tools/virsh-pool.c
+@@ -1231,6 +1231,9 @@ cmdPoolList(vshControl *ctl, const vshCmd *cmd G_GNUC_UNUSED)
+             case VIR_STORAGE_POOL_VSTORAGE:
+                 flags |= VIR_CONNECT_LIST_STORAGE_POOLS_VSTORAGE;
+                 break;
++            case VIR_STORAGE_POOL_VITASTOR:
++                flags |= VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR;
++                break;
+             case VIR_STORAGE_POOL_LAST:
+                 break;
+             }

patches/libvirt-example.xml

@@ -0,0 +1,32 @@
+<!-- Example libvirt VM configuration with Vitastor disk -->
+<domain type='kvm'>
+  <name>debian9</name>
+  <uuid>96f277fb-fd9c-49da-bf21-a5cfd54eb162</uuid>
+  <memory unit="KiB">524288</memory>
+  <currentMemory>524288</currentMemory>
+  <vcpu>1</vcpu>
+  <os>
+    <type arch='x86_64'>hvm</type>
+    <boot dev='hd' />
+  </os>
+  <devices>
+    <emulator>/usr/bin/qemu-system-x86_64</emulator>
+    <disk type='network' device='disk'>
+      <target dev='vda' bus='virtio' />
+      <driver name='qemu' type='raw' />
+      <!-- name is Vitastor image name -->
+      <!-- config (optional) is the path to Vitastor's configuration file -->
+      <!-- query (optional) is Vitastor's etcd_prefix -->
+      <source protocol='vitastor' name='debian9' query='/vitastor' config='/etc/vitastor/vitastor.conf'>
+        <!-- hosts = etcd addresses -->
+        <host name='192.168.7.2' port='2379' />
+      </source>
+      <!-- required because Vitastor only supports 4k physical sectors -->
+      <blockio physical_block_size="4096" logical_block_size="512" />
+    </disk>
+    <interface type='network'>
+      <source network='default' />
+    </interface>
+    <graphics type='vnc' port='-1' />
+  </devices>
+</domain>

patches/nova-20.diff

@@ -0,0 +1,287 @@
+diff --git a/nova/virt/image/model.py b/nova/virt/image/model.py
+index 971f7e9c07..70ed70d5e2 100644
+--- a/nova/virt/image/model.py
++++ b/nova/virt/image/model.py
+@@ -129,3 +129,22 @@ class RBDImage(Image):
+         self.user = user
+         self.password = password
+         self.servers = servers
++
++
++class VitastorImage(Image):
++    """Class for images in a remote Vitastor cluster"""
++
++    def __init__(self, name, etcd_address = None, etcd_prefix = None, config_path = None):
++        """Create a new Vitastor image object
++
++        :param name: name of the image
++        :param etcd_address: etcd URL(s) (optional)
++        :param etcd_prefix: etcd prefix (optional)
++        :param config_path: path to the configuration (optional)
++        """
++        super(RBDImage, self).__init__(FORMAT_RAW)
++
++        self.name = name
++        self.etcd_address = etcd_address
++        self.etcd_prefix = etcd_prefix
++        self.config_path = config_path
+diff --git a/nova/virt/images.py b/nova/virt/images.py
+index 5358f3766a..ebe3d6effb 100644
+--- a/nova/virt/images.py
++++ b/nova/virt/images.py
+@@ -41,7 +41,7 @@ IMAGE_API = glance.API()
+ 
+ def qemu_img_info(path, format=None):
+     """Return an object containing the parsed output from qemu-img info."""
+-    if not os.path.exists(path) and not path.startswith('rbd:'):
++    if not os.path.exists(path) and not path.startswith('rbd:') and not path.startswith('vitastor:'):
+         raise exception.DiskNotFound(location=path)
+ 
+     info = nova.privsep.qemu.unprivileged_qemu_img_info(path, format=format)
+@@ -50,7 +50,7 @@ def qemu_img_info(path, format=None):
+ 
+ def privileged_qemu_img_info(path, format=None, output_format='json'):
+     """Return an object containing the parsed output from qemu-img info."""
+-    if not os.path.exists(path) and not path.startswith('rbd:'):
++    if not os.path.exists(path) and not path.startswith('rbd:') and not path.startswith('vitastor:'):
+         raise exception.DiskNotFound(location=path)
+ 
+     info = nova.privsep.qemu.privileged_qemu_img_info(path, format=format)
+diff --git a/nova/virt/libvirt/config.py b/nova/virt/libvirt/config.py
+index f9475776b3..51573fe41d 100644
+--- a/nova/virt/libvirt/config.py
++++ b/nova/virt/libvirt/config.py
+@@ -1060,6 +1060,8 @@ class LibvirtConfigGuestDisk(LibvirtConfigGuestDevice):
+         self.driver_iommu = False
+         self.source_path = None
+         self.source_protocol = None
++        self.source_query = None
++        self.source_config = None
+         self.source_name = None
+         self.source_hosts = []
+         self.source_ports = []
+@@ -1186,7 +1188,8 @@ class LibvirtConfigGuestDisk(LibvirtConfigGuestDevice):
+         elif self.source_type == "mount":
+             dev.append(etree.Element("source", dir=self.source_path))
+         elif self.source_type == "network" and self.source_protocol:
+-            source = etree.Element("source", protocol=self.source_protocol)
++            source = etree.Element("source", protocol=self.source_protocol,
++                query=self.source_query, config=self.source_config)
+             if self.source_name is not None:
+                 source.set('name', self.source_name)
+             hosts_info = zip(self.source_hosts, self.source_ports)
+diff --git a/nova/virt/libvirt/driver.py b/nova/virt/libvirt/driver.py
+index 391231c527..34dc60dcdd 100644
+--- a/nova/virt/libvirt/driver.py
++++ b/nova/virt/libvirt/driver.py
+@@ -179,6 +179,7 @@ VOLUME_DRIVERS = {
+     'local': 'nova.virt.libvirt.volume.volume.LibvirtVolumeDriver',
+     'fake': 'nova.virt.libvirt.volume.volume.LibvirtFakeVolumeDriver',
+     'rbd': 'nova.virt.libvirt.volume.net.LibvirtNetVolumeDriver',
++    'vitastor': 'nova.virt.libvirt.volume.vitastor.LibvirtVitastorVolumeDriver',
+     'nfs': 'nova.virt.libvirt.volume.nfs.LibvirtNFSVolumeDriver',
+     'smbfs': 'nova.virt.libvirt.volume.smbfs.LibvirtSMBFSVolumeDriver',
+     'fibre_channel': 'nova.virt.libvirt.volume.fibrechannel.LibvirtFibreChannelVolumeDriver',  # noqa:E501
+@@ -385,10 +386,10 @@ class LibvirtDriver(driver.ComputeDriver):
+         # This prevents the risk of one test setting a capability
+         # which bleeds over into other tests.
+ 
+-        # LVM and RBD require raw images. If we are not configured to
++        # LVM, RBD, Vitastor require raw images. If we are not configured to
+         # force convert images into raw format, then we _require_ raw
+         # images only.
+-        raw_only = ('rbd', 'lvm')
++        raw_only = ('rbd', 'lvm', 'vitastor')
+         requires_raw_image = (CONF.libvirt.images_type in raw_only and
+                               not CONF.force_raw_images)
+         requires_ploop_image = CONF.libvirt.virt_type == 'parallels'
+@@ -775,12 +776,12 @@ class LibvirtDriver(driver.ComputeDriver):
+         # Some imagebackends are only able to import raw disk images,
+         # and will fail if given any other format. See the bug
+         # https://bugs.launchpad.net/nova/+bug/1816686 for more details.
+-        if CONF.libvirt.images_type in ('rbd',):
++        if CONF.libvirt.images_type in ('rbd', 'vitastor'):
+             if not CONF.force_raw_images:
+                 msg = _("'[DEFAULT]/force_raw_images = False' is not "
+-                        "allowed with '[libvirt]/images_type = rbd'. "
++                        "allowed with '[libvirt]/images_type = rbd' or 'vitastor'. "
+                         "Please check the two configs and if you really "
+-                        "do want to use rbd as images_type, set "
++                        "do want to use rbd or vitastor as images_type, set "
+                         "force_raw_images to True.")
+                 raise exception.InvalidConfiguration(msg)
+ 
+@@ -2603,6 +2604,16 @@ class LibvirtDriver(driver.ComputeDriver):
+                     if connection_info['data'].get('auth_enabled'):
+                         username = connection_info['data']['auth_username']
+                         path = f"rbd:{volume_name}:id={username}"
++                elif connection_info['driver_volume_type'] == 'vitastor':
++                    volume_name = connection_info['data']['name']
++                    path = 'vitastor:image='+volume_name.replace(':', '\\:')
++                    for k in [ 'config_path', 'etcd_address', 'etcd_prefix' ]:
++                        if k in connection_info['data']:
++                            kk = k
++                            if kk == 'etcd_address':
++                                # FIXME use etcd_address in qemu driver
++                                kk = 'etcd_host'
++                            path += ":"+kk+"="+connection_info['data'][k].replace(':', '\\:')
+                 else:
+                     path = 'unknown'
+                     raise exception.DiskNotFound(location='unknown')
+@@ -2827,8 +2838,8 @@ class LibvirtDriver(driver.ComputeDriver):
+ 
+         image_format = CONF.libvirt.snapshot_image_format or source_type
+ 
+-        # NOTE(bfilippov): save lvm and rbd as raw
+-        if image_format == 'lvm' or image_format == 'rbd':
++        # NOTE(bfilippov): save lvm and rbd and vitastor as raw
++        if image_format == 'lvm' or image_format == 'rbd' or image_format == 'vitastor':
+             image_format = 'raw'
+ 
+         metadata = self._create_snapshot_metadata(instance.image_meta,
+@@ -2899,7 +2910,7 @@ class LibvirtDriver(driver.ComputeDriver):
+                               expected_state=task_states.IMAGE_UPLOADING)
+ 
+             # TODO(nic): possibly abstract this out to the root_disk
+-            if source_type == 'rbd' and live_snapshot:
++            if (source_type == 'rbd' or source_type == 'vitastor') and live_snapshot:
+                 # Standard snapshot uses qemu-img convert from RBD which is
+                 # not safe to run with live_snapshot.
+                 live_snapshot = False
+@@ -4099,7 +4110,7 @@ class LibvirtDriver(driver.ComputeDriver):
+         # cleanup rescue volume
+         lvm.remove_volumes([lvmdisk for lvmdisk in self._lvm_disks(instance)
+                                 if lvmdisk.endswith('.rescue')])
+-        if CONF.libvirt.images_type == 'rbd':
++        if CONF.libvirt.images_type == 'rbd' or CONF.libvirt.images_type == 'vitastor':
+             filter_fn = lambda disk: (disk.startswith(instance.uuid) and
+                                       disk.endswith('.rescue'))
+             rbd_utils.RBDDriver().cleanup_volumes(filter_fn)
+@@ -4356,6 +4367,8 @@ class LibvirtDriver(driver.ComputeDriver):
+         # TODO(mikal): there is a bug here if images_type has
+         # changed since creation of the instance, but I am pretty
+         # sure that this bug already exists.
++        if CONF.libvirt.images_type == 'vitastor':
++            return 'vitastor'
+         return 'rbd' if CONF.libvirt.images_type == 'rbd' else 'raw'
+ 
+     @staticmethod
+@@ -4764,10 +4777,10 @@ class LibvirtDriver(driver.ComputeDriver):
+                 finally:
+                     # NOTE(mikal): if the config drive was imported into RBD,
+                     # then we no longer need the local copy
+-                    if CONF.libvirt.images_type == 'rbd':
++                    if CONF.libvirt.images_type == 'rbd' or CONF.libvirt.images_type == 'vitastor':
+                         LOG.info('Deleting local config drive %(path)s '
+-                                 'because it was imported into RBD.',
+-                                 {'path': config_disk_local_path},
++                                 'because it was imported into %(type).',
++                                 {'path': config_disk_local_path, 'type': CONF.libvirt.images_type},
+                                  instance=instance)
+                         os.unlink(config_disk_local_path)
+ 
+diff --git a/nova/virt/libvirt/utils.py b/nova/virt/libvirt/utils.py
+index da2a6e8b8a..52c02e72f1 100644
+--- a/nova/virt/libvirt/utils.py
++++ b/nova/virt/libvirt/utils.py
+@@ -340,6 +340,10 @@ def find_disk(guest: libvirt_guest.Guest) -> ty.Tuple[str, ty.Optional[str]]:
+             disk_path = disk.source_name
+             if disk_path:
+                 disk_path = 'rbd:' + disk_path
++        elif not disk_path and disk.source_protocol == 'vitastor':
++            disk_path = disk.source_name
++            if disk_path:
++                disk_path = 'vitastor:' + disk_path
+ 
+     if not disk_path:
+         raise RuntimeError(_("Can't retrieve root device path "
+@@ -354,6 +358,8 @@ def get_disk_type_from_path(path: str) -> ty.Optional[str]:
+         return 'lvm'
+     elif path.startswith('rbd:'):
+         return 'rbd'
++    elif path.startswith('vitastor:'):
++        return 'vitastor'
+     elif (os.path.isdir(path) and
+           os.path.exists(os.path.join(path, "DiskDescriptor.xml"))):
+         return 'ploop'
+diff --git a/nova/virt/libvirt/volume/vitastor.py b/nova/virt/libvirt/volume/vitastor.py
+new file mode 100644
+index 0000000000..0256df62c1
+--- /dev/null
++++ b/nova/virt/libvirt/volume/vitastor.py
+@@ -0,0 +1,75 @@
++# Copyright (c) 2021+, Vitaliy Filippov <vitalif@yourcmc.ru>
++#
++#    Licensed under the Apache License, Version 2.0 (the "License"); you may
++#    not use this file except in compliance with the License. You may obtain
++#    a copy of the License at
++#
++#         http://www.apache.org/licenses/LICENSE-2.0
++#
++#    Unless required by applicable law or agreed to in writing, software
++#    distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
++#    WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
++#    License for the specific language governing permissions and limitations
++#    under the License.
++
++from os_brick import exception as os_brick_exception
++from os_brick import initiator
++from os_brick.initiator import connector
++from oslo_log import log as logging
++
++import nova.conf
++from nova import utils
++from nova.virt.libvirt.volume import volume as libvirt_volume
++
++
++CONF = nova.conf.CONF
++LOG = logging.getLogger(__name__)
++
++
++class LibvirtVitastorVolumeDriver(libvirt_volume.LibvirtBaseVolumeDriver):
++    """Driver to attach Vitastor volumes to libvirt."""
++    def __init__(self, host):
++        super(LibvirtVitastorVolumeDriver, self).__init__(host, is_block_dev=False)
++
++    def connect_volume(self, connection_info, instance):
++        pass
++
++    def disconnect_volume(self, connection_info, instance):
++        pass
++
++    def get_config(self, connection_info, disk_info):
++        """Returns xml for libvirt."""
++        conf = super(LibvirtVitastorVolumeDriver, self).get_config(connection_info, disk_info)
++        conf.source_type = 'network'
++        conf.source_protocol = 'vitastor'
++        conf.source_name = connection_info['data'].get('name')
++        conf.source_query = connection_info['data'].get('etcd_prefix') or None
++        conf.source_config = connection_info['data'].get('config_path') or None
++        conf.source_hosts = []
++        conf.source_ports = []
++        addresses = connection_info['data'].get('etcd_address', '')
++        if addresses:
++            if not isinstance(addresses, list):
++                addresses = addresses.split(',')
++            for addr in addresses:
++                if addr.startswith('https://'):
++                    raise NotImplementedError('Vitastor block driver does not support SSL for etcd communication yet')
++                if addr.startswith('http://'):
++                    addr = addr[7:]
++                addr = addr.rstrip('/')
++                if addr.endswith('/v3'):
++                    addr = addr[0:-3]
++                p = addr.find('/')
++                if p > 0:
++                    raise NotImplementedError('libvirt does not support custom URL paths for Vitastor etcd yet. Use /etc/vitastor/vitastor.conf')
++                p = addr.find(':')
++                port = '2379'
++                if p > 0:
++                    port = addr[p+1:]
++                    addr = addr[0:p]
++                conf.source_hosts.append(addr)
++                conf.source_ports.append(port)
++        return conf
++
++    def extend_volume(self, connection_info, instance, requested_size):
++        raise NotImplementedError

patches/qemu-3.1-vitastor.patch

@@ -11,7 +11,7 @@ Index: qemu-3.1+dfsg/qapi/block-core.json
              'host_cdrom', 'host_device', 'http', 'https', 'iscsi', 'luks',
              'nbd', 'nfs', 'null-aio', 'null-co', 'nvme', 'parallels', 'qcow',
              'qcow2', 'qed', 'quorum', 'raw', 'rbd', 'replication', 'sheepdog',
-@@ -3367,6 +3367,24 @@
+@@ -3367,6 +3367,28 @@
              '*tag': 'str' } }
  
  ##
@@ -19,17 +19,21 @@ Index: qemu-3.1+dfsg/qapi/block-core.json
 +#
 +# Driver specific block device options for vitastor
 +#
++# @image:       Image name
 +# @inode:       Inode number
 +# @pool:        Pool ID
 +# @size:        Desired image size in bytes
-+# @etcd_host:   etcd connection address
++# @config_path: Path to Vitastor configuration
++# @etcd_host:   etcd connection address(es)
 +# @etcd_prefix: etcd key/value prefix
 +##
 +{ 'struct': 'BlockdevOptionsVitastor',
-+  'data': { 'inode': 'uint64',
-+            'pool': 'uint64',
-+            'size': 'uint64',
-+            'etcd_host': 'str',
++  'data': { '*inode': 'uint64',
++            '*pool': 'uint64',
++            '*size': 'uint64',
++            '*image': 'str',
++            '*config_path': 'str',
++            '*etcd_host': 'str',
 +            '*etcd_prefix': 'str' } }
 +
 +##

patches/qemu-4.2-vitastor.patch

@@ -11,7 +11,7 @@ Index: qemu/qapi/block-core.json
              'ssh', 'throttle', 'vdi', 'vhdx', 'vmdk', 'vpc', 'vvfat', 'vxhs' ] }
  
  ##
-@@ -3725,6 +3725,24 @@
+@@ -3725,6 +3725,28 @@
              '*tag': 'str' } }
  
  ##
@@ -19,17 +19,21 @@ Index: qemu/qapi/block-core.json
 +#
 +# Driver specific block device options for vitastor
 +#
++# @image:       Image name
 +# @inode:       Inode number
 +# @pool:        Pool ID
 +# @size:        Desired image size in bytes
-+# @etcd_host:   etcd connection address
++# @config_path: Path to Vitastor configuration
++# @etcd_host:   etcd connection address(es)
 +# @etcd_prefix: etcd key/value prefix
 +##
 +{ 'struct': 'BlockdevOptionsVitastor',
-+  'data': { 'inode': 'uint64',
-+            'pool': 'uint64',
-+            'size': 'uint64',
-+            'etcd_host': 'str',
++  'data': { '*inode': 'uint64',
++            '*pool': 'uint64',
++            '*size': 'uint64',
++            '*image': 'str',
++            '*config_path': 'str',
++            '*etcd_host': 'str',
 +            '*etcd_prefix': 'str' } }
 +
 +##

patches/qemu-5.0-vitastor.patch

@@ -11,7 +11,7 @@ Index: qemu/qapi/block-core.json
              'ssh', 'throttle', 'vdi', 'vhdx', 'vmdk', 'vpc', 'vvfat', 'vxhs' ] }
  
  ##
-@@ -3635,6 +3635,24 @@
+@@ -3635,6 +3635,28 @@
              '*tag': 'str' } }
  
  ##
@@ -19,17 +19,21 @@ Index: qemu/qapi/block-core.json
 +#
 +# Driver specific block device options for vitastor
 +#
++# @image:       Image name
 +# @inode:       Inode number
 +# @pool:        Pool ID
 +# @size:        Desired image size in bytes
-+# @etcd_host:   etcd connection address
++# @config_path: Path to Vitastor configuration
++# @etcd_host:   etcd connection address(es)
 +# @etcd_prefix: etcd key/value prefix
 +##
 +{ 'struct': 'BlockdevOptionsVitastor',
-+  'data': { 'inode': 'uint64',
-+            'pool': 'uint64',
-+            'size': 'uint64',
-+            'etcd_host': 'str',
++  'data': { '*inode': 'uint64',
++            '*pool': 'uint64',
++            '*size': 'uint64',
++            '*image': 'str',
++            '*config_path': 'str',
++            '*etcd_host': 'str',
 +            '*etcd_prefix': 'str' } }
 +
 +##

patches/qemu-5.1-vitastor.patch

@@ -11,7 +11,7 @@ Index: qemu-5.1+dfsg/qapi/block-core.json
              'ssh', 'throttle', 'vdi', 'vhdx', 'vmdk', 'vpc', 'vvfat' ] }
  
  ##
-@@ -3644,6 +3644,24 @@
+@@ -3644,6 +3644,28 @@
              '*tag': 'str' } }
  
  ##
@@ -19,17 +19,21 @@ Index: qemu-5.1+dfsg/qapi/block-core.json
 +#
 +# Driver specific block device options for vitastor
 +#
++# @image:       Image name
 +# @inode:       Inode number
 +# @pool:        Pool ID
 +# @size:        Desired image size in bytes
-+# @etcd_host:   etcd connection address
++# @config_path: Path to Vitastor configuration
++# @etcd_host:   etcd connection address(es)
 +# @etcd_prefix: etcd key/value prefix
 +##
 +{ 'struct': 'BlockdevOptionsVitastor',
-+  'data': { 'inode': 'uint64',
-+            'pool': 'uint64',
-+            'size': 'uint64',
-+            'etcd_host': 'str',
++  'data': { '*inode': 'uint64',
++            '*pool': 'uint64',
++            '*size': 'uint64',
++            '*image': 'str',
++            '*config_path': 'str',
++            '*etcd_host': 'str',
 +            '*etcd_prefix': 'str' } }
 +
 +##

rpm/build-tarball.sh

@@ -48,4 +48,4 @@ FIO=`rpm -qi fio | perl -e 'while(<>) { /^Epoch[\s:]+(\S+)/ && print "$1:"; /^Ve
 QEMU=`rpm -qi qemu qemu-kvm | perl -e 'while(<>) { /^Epoch[\s:]+(\S+)/ && print "$1:"; /^Version[\s:]+(\S+)/ && print $1; /^Release[\s:]+(\S+)/ && print "-$1"; }'`
 perl -i -pe 's/(Requires:\s*fio)([^\n]+)?/$1 = '$FIO'/' $VITASTOR/rpm/vitastor-el$EL.spec
 perl -i -pe 's/(Requires:\s*qemu(?:-kvm)?)([^\n]+)?/$1 = '$QEMU'/' $VITASTOR/rpm/vitastor-el$EL.spec
-tar --transform 's#^#vitastor-0.6.3/#' --exclude 'rpm/*.rpm' -czf $VITASTOR/../vitastor-0.6.3$(rpm --eval '%dist').tar.gz *
+tar --transform 's#^#vitastor-0.6.6/#' --exclude 'rpm/*.rpm' -czf $VITASTOR/../vitastor-0.6.6$(rpm --eval '%dist').tar.gz *

rpm/qemu-el8.Dockerfile

@@ -11,7 +11,7 @@ RUN rm -rf /var/lib/dnf/*; dnf download --disablerepo='*' --enablerepo='centos-a
 RUN rpm --nomd5 -i qemu*.src.rpm
 RUN cd ~/rpmbuild/SPECS && dnf builddep -y --enablerepo=PowerTools --spec qemu-kvm.spec
 
-ADD qemu-*-vitastor.patch /root/vitastor/
+ADD patches/qemu-*-vitastor.patch /root/vitastor/patches/
 
 RUN set -e; \
     mkdir -p /root/packages/qemu-el8; \
@@ -25,7 +25,7 @@ RUN set -e; \
     echo "Patch$((PN+1)): qemu-4.2-vitastor.patch" >> qemu-kvm.spec; \
     tail -n +2 xx01 >> qemu-kvm.spec; \
     perl -i -pe 's/(^Release:\s*\d+)/$1.vitastor/' qemu-kvm.spec; \
-    cp /root/vitastor/qemu-4.2-vitastor.patch ~/rpmbuild/SOURCES; \
+    cp /root/vitastor/patches/qemu-4.2-vitastor.patch ~/rpmbuild/SOURCES; \
     rpmbuild --nocheck -ba qemu-kvm.spec; \
     cp ~/rpmbuild/RPMS/*/*qemu* /root/packages/qemu-el8/; \
     cp ~/rpmbuild/SRPMS/*qemu* /root/packages/qemu-el8/

rpm/vitastor-el7.Dockerfile

@@ -15,8 +15,8 @@ RUN yumdownloader --disablerepo=centos-sclo-rh --source fio
 RUN rpm --nomd5 -i qemu*.src.rpm
 RUN rpm --nomd5 -i fio*.src.rpm
 RUN rm -f /etc/yum.repos.d/CentOS-Media.repo
-RUN cd ~/rpmbuild/SPECS && yum-builddep -y --enablerepo='*' --disablerepo=centos-sclo-rh --disablerepo=centos-sclo-rh-source --disablerepo=centos-sclo-sclo-testing qemu-kvm.spec
-RUN cd ~/rpmbuild/SPECS && yum-builddep -y --enablerepo='*' --disablerepo=centos-sclo-rh --disablerepo=centos-sclo-rh-source --disablerepo=centos-sclo-sclo-testing fio.spec
+RUN cd ~/rpmbuild/SPECS && yum-builddep -y qemu-kvm.spec
+RUN cd ~/rpmbuild/SPECS && yum-builddep -y fio.spec
 RUN yum -y install rdma-core-devel
 
 ADD https://vitastor.io/rpms/liburing-el7/liburing-0.7-2.el7.src.rpm /root
@@ -38,7 +38,7 @@ ADD . /root/vitastor
 RUN set -e; \
     cd /root/vitastor/rpm; \
     sh build-tarball.sh; \
-    cp /root/vitastor-0.6.3.el7.tar.gz ~/rpmbuild/SOURCES; \
+    cp /root/vitastor-0.6.6.el7.tar.gz ~/rpmbuild/SOURCES; \
     cp vitastor-el7.spec ~/rpmbuild/SPECS/vitastor.spec; \
     cd ~/rpmbuild/SPECS/; \
     rpmbuild -ba vitastor.spec; \

rpm/vitastor-el7.spec

@@ -1,11 +1,11 @@
 Name:           vitastor
-Version:        0.6.3
+Version:        0.6.6
 Release:        1%{?dist}
 Summary:        Vitastor, a fast software-defined clustered block storage
 
 License:        Vitastor Network Public License 1.1
 URL:            https://vitastor.io/
-Source0:        vitastor-0.6.3.el7.tar.gz
+Source0:        vitastor-0.6.6.el7.tar.gz
 
 BuildRequires:  liburing-devel >= 0.6
 BuildRequires:  gperftools-devel
@@ -57,6 +57,7 @@ cp -r mon %buildroot/usr/lib/vitastor/mon
 %_bindir/vitastor-dump-journal
 %_bindir/vitastor-nbd
 %_bindir/vitastor-osd
+%_bindir/vitastor-cli
 %_bindir/vitastor-rm
 %_libdir/qemu-kvm/block-vitastor.so
 %_libdir/libfio_vitastor.so
@@ -64,6 +65,7 @@ cp -r mon %buildroot/usr/lib/vitastor/mon
 %_libdir/libfio_vitastor_sec.so
 %_libdir/libvitastor_blk.so*
 %_libdir/libvitastor_client.so*
+%_includedir/vitastor_c.h
 /usr/lib/vitastor
 
 

rpm/vitastor-el8.Dockerfile

@@ -15,7 +15,7 @@ RUN rpm --nomd5 -i qemu*.src.rpm
 RUN rpm --nomd5 -i fio*.src.rpm
 RUN cd ~/rpmbuild/SPECS && dnf builddep -y --enablerepo=powertools --spec qemu-kvm.spec
 RUN cd ~/rpmbuild/SPECS && dnf builddep -y --enablerepo=powertools --spec fio.spec && dnf install -y cmake
-RUN yum -y install libibverbs-devel
+RUN yum -y install libibverbs-devel libarchive
 
 ADD https://vitastor.io/rpms/liburing-el7/liburing-0.7-2.el7.src.rpm /root
 
@@ -36,7 +36,7 @@ ADD . /root/vitastor
 RUN set -e; \
     cd /root/vitastor/rpm; \
     sh build-tarball.sh; \
-    cp /root/vitastor-0.6.3.el8.tar.gz ~/rpmbuild/SOURCES; \
+    cp /root/vitastor-0.6.6.el8.tar.gz ~/rpmbuild/SOURCES; \
     cp vitastor-el8.spec ~/rpmbuild/SPECS/vitastor.spec; \
     cd ~/rpmbuild/SPECS/; \
     rpmbuild -ba vitastor.spec; \

rpm/vitastor-el8.spec

@@ -1,11 +1,11 @@
 Name:           vitastor
-Version:        0.6.3
+Version:        0.6.6
 Release:        1%{?dist}
 Summary:        Vitastor, a fast software-defined clustered block storage
 
 License:        Vitastor Network Public License 1.1
 URL:            https://vitastor.io/
-Source0:        vitastor-0.6.3.el8.tar.gz
+Source0:        vitastor-0.6.6.el8.tar.gz
 
 BuildRequires:  liburing-devel >= 0.6
 BuildRequires:  gperftools-devel
@@ -54,6 +54,7 @@ cp -r mon %buildroot/usr/lib/vitastor
 %_bindir/vitastor-dump-journal
 %_bindir/vitastor-nbd
 %_bindir/vitastor-osd
+%_bindir/vitastor-cli
 %_bindir/vitastor-rm
 %_libdir/qemu-kvm/block-vitastor.so
 %_libdir/libfio_vitastor.so
@@ -61,6 +62,7 @@ cp -r mon %buildroot/usr/lib/vitastor
 %_libdir/libfio_vitastor_sec.so
 %_libdir/libvitastor_blk.so*
 %_libdir/libvitastor_client.so*
+%_includedir/vitastor_c.h
 /usr/lib/vitastor
 
 

src/CMakeLists.txt

@@ -4,6 +4,8 @@ project(vitastor)
 
 include(GNUInstallDirs)
 
+set(WITH_QEMU true CACHE BOOL "Build QEMU driver")
+set(WITH_FIO true CACHE BOOL "Build FIO driver")
 set(QEMU_PLUGINDIR qemu CACHE STRING "QEMU plugin directory suffix (qemu-kvm on RHEL)")
 set(WITH_ASAN false CACHE BOOL "Build with AddressSanitizer")
 if("${CMAKE_INSTALL_PREFIX}" MATCHES "^/usr/local/?$")
@@ -13,8 +15,8 @@ if("${CMAKE_INSTALL_PREFIX}" MATCHES "^/usr/local/?$")
 	set(CMAKE_INSTALL_RPATH "${CMAKE_INSTALL_PREFIX}/${CMAKE_INSTALL_LIBDIR}")
 endif()
 
-add_definitions(-DVERSION="0.6.3")
-add_definitions(-Wall -Wno-sign-compare -Wno-comment -Wno-parentheses -Wno-pointer-arith -I ${CMAKE_SOURCE_DIR}/src)
+add_definitions(-DVERSION="0.6.6")
+add_definitions(-Wall -Wno-sign-compare -Wno-comment -Wno-parentheses -Wno-pointer-arith -fdiagnostics-color=always -I ${CMAKE_SOURCE_DIR}/src)
 if (${WITH_ASAN})
 	add_definitions(-fsanitize=address -fno-omit-frame-pointer)
 	add_link_options(-fsanitize=address -fno-omit-frame-pointer)
@@ -34,9 +36,16 @@ string(REGEX REPLACE "([\\/\\-]D) *NDEBUG" "" CMAKE_C_FLAGS_RELEASE "${CMAKE_C_F
 string(REGEX REPLACE "([\\/\\-]D) *NDEBUG" "" CMAKE_C_FLAGS_MINSIZEREL "${CMAKE_C_FLAGS_MINSIZEREL}")
 string(REGEX REPLACE "([\\/\\-]D) *NDEBUG" "" CMAKE_C_FLAGS_RELWITHDEBINFO "${CMAKE_C_FLAGS_RELWITHDEBINFO}")
 
+macro(install_symlink filepath sympath)
+	install(CODE "execute_process(COMMAND ${CMAKE_COMMAND} -E create_symlink ${filepath} ${sympath})")
+	install(CODE "message(\"-- Created symlink: ${sympath} -> ${filepath}\")")
+endmacro(install_symlink)
+
 find_package(PkgConfig)
 pkg_check_modules(LIBURING REQUIRED liburing)
-pkg_check_modules(GLIB REQUIRED glib-2.0)
+if (${WITH_QEMU})
+	pkg_check_modules(GLIB REQUIRED glib-2.0)
+endif (${WITH_QEMU})
 pkg_check_modules(IBVERBS libibverbs)
 if (IBVERBS_LIBRARIES)
 	add_definitions(-DWITH_RDMA)
@@ -62,14 +71,16 @@ target_link_libraries(vitastor_blk
 )
 set_target_properties(vitastor_blk PROPERTIES VERSION ${VERSION} SOVERSION 0)
 
-# libfio_vitastor_blk.so
-add_library(fio_vitastor_blk SHARED
-	fio_engine.cpp
-	../json11/json11.cpp
-)
-target_link_libraries(fio_vitastor_blk
-	vitastor_blk
-)
+if (${WITH_FIO})
+	# libfio_vitastor_blk.so
+	add_library(fio_vitastor_blk SHARED
+		fio_engine.cpp
+		../json11/json11.cpp
+	)
+	target_link_libraries(fio_vitastor_blk
+		vitastor_blk
+	)
+endif (${WITH_FIO})
 
 # libvitastor_common.a
 set(MSGR_RDMA "")
@@ -96,19 +107,24 @@ target_link_libraries(vitastor-osd
 	${IBVERBS_LIBRARIES}
 )
 
-# libfio_vitastor_sec.so
-add_library(fio_vitastor_sec SHARED
-	fio_sec_osd.cpp
-	rw_blocking.cpp
-)
-target_link_libraries(fio_vitastor_sec
-	tcmalloc_minimal
-)
+if (${WITH_FIO})
+	# libfio_vitastor_sec.so
+	add_library(fio_vitastor_sec SHARED
+		fio_sec_osd.cpp
+		rw_blocking.cpp
+	)
+	target_link_libraries(fio_vitastor_sec
+		tcmalloc_minimal
+	)
+endif (${WITH_FIO})
 
 # libvitastor_client.so
 add_library(vitastor_client SHARED
 	cluster_client.cpp
+	cluster_client_list.cpp
+	vitastor_c.cpp
 )
+set_target_properties(vitastor_client PROPERTIES PUBLIC_HEADER "vitastor_c.h")
 target_link_libraries(vitastor_client
 	vitastor_common
 	tcmalloc_minimal
@@ -117,13 +133,15 @@ target_link_libraries(vitastor_client
 )
 set_target_properties(vitastor_client PROPERTIES VERSION ${VERSION} SOVERSION 0)
 
-# libfio_vitastor.so
-add_library(fio_vitastor SHARED
-	fio_cluster.cpp
-)
-target_link_libraries(fio_vitastor
-	vitastor_client
-)
+if (${WITH_FIO})
+	# libfio_vitastor.so
+	add_library(fio_vitastor SHARED
+		fio_cluster.cpp
+	)
+	target_link_libraries(fio_vitastor
+		vitastor_client
+	)
+endif (${WITH_FIO})
 
 # vitastor-nbd
 add_executable(vitastor-nbd
@@ -133,11 +151,11 @@ target_link_libraries(vitastor-nbd
 	vitastor_client
 )
 
-# vitastor-rm
-add_executable(vitastor-rm
-	rm_inode.cpp
+# vitastor-cli
+add_executable(vitastor-cli
+	cli.cpp cli_flatten.cpp cli_merge.cpp cli_rm.cpp cli_snap_rm.cpp
 )
-target_link_libraries(vitastor-rm
+target_link_libraries(vitastor-cli
 	vitastor_client
 )
 
@@ -146,27 +164,24 @@ add_executable(vitastor-dump-journal
 	dump_journal.cpp crc32c.c
 )
 
-# qemu_driver.so
-add_library(qemu_proxy STATIC qemu_proxy.cpp)
-target_compile_options(qemu_proxy PUBLIC -fPIC)
-target_include_directories(qemu_proxy PUBLIC
-	../qemu/b/qemu
-	../qemu/include
-	${GLIB_INCLUDE_DIRS}
-)
-target_link_libraries(qemu_proxy
-	vitastor_client
-)
-add_library(qemu_vitastor SHARED
-	qemu_driver.c
-)
-target_link_libraries(qemu_vitastor
-	qemu_proxy
-)
-set_target_properties(qemu_vitastor PROPERTIES
-	PREFIX ""
-	OUTPUT_NAME "block-vitastor"
-)
+if (${WITH_QEMU})
+	# qemu_driver.so
+	add_library(qemu_vitastor SHARED
+		qemu_driver.c
+	)
+	target_include_directories(qemu_vitastor PUBLIC
+		../qemu/b/qemu
+		../qemu/include
+		${GLIB_INCLUDE_DIRS}
+	)
+	target_link_libraries(qemu_vitastor
+		vitastor_client
+	)
+	set_target_properties(qemu_vitastor PROPERTIES
+		PREFIX ""
+		OUTPUT_NAME "block-vitastor"
+	)
+endif (${WITH_QEMU})
 
 ### Test stubs
 
@@ -200,10 +215,18 @@ target_link_libraries(osd_peering_pg_test tcmalloc_minimal)
 # test_allocator
 add_executable(test_allocator test_allocator.cpp allocator.cpp)
 
+# test_cas
+add_executable(test_cas
+	test_cas.cpp
+)
+target_link_libraries(test_cas
+	vitastor_client
+)
+
 # test_cluster_client
 add_executable(test_cluster_client
 	test_cluster_client.cpp
-	pg_states.cpp osd_ops.cpp cluster_client.cpp msgr_op.cpp mock/messenger.cpp msgr_stop.cpp
+	pg_states.cpp osd_ops.cpp cluster_client.cpp cluster_client_list.cpp msgr_op.cpp mock/messenger.cpp msgr_stop.cpp
 	etcd_state_client.cpp timerfd_manager.cpp ../json11/json11.cpp
 )
 target_compile_definitions(test_cluster_client PUBLIC -D__MOCK__)
@@ -217,6 +240,16 @@ target_include_directories(test_cluster_client PUBLIC ${CMAKE_SOURCE_DIR}/src/mo
 
 ### Install
 
-install(TARGETS vitastor-osd vitastor-dump-journal vitastor-nbd vitastor-rm RUNTIME DESTINATION ${CMAKE_INSTALL_BINDIR})
-install(TARGETS fio_vitastor fio_vitastor_blk fio_vitastor_sec vitastor_blk vitastor_client LIBRARY DESTINATION ${CMAKE_INSTALL_LIBDIR})
-install(TARGETS qemu_vitastor LIBRARY DESTINATION /usr/${CMAKE_INSTALL_LIBDIR}/${QEMU_PLUGINDIR})
+install(TARGETS vitastor-osd vitastor-dump-journal vitastor-nbd vitastor-cli RUNTIME DESTINATION ${CMAKE_INSTALL_BINDIR})
+install_symlink(${CMAKE_INSTALL_BINDIR}/vitastor-rm vitastor-cli)
+install(
+	TARGETS vitastor_blk vitastor_client
+	LIBRARY DESTINATION ${CMAKE_INSTALL_LIBDIR}
+	PUBLIC_HEADER DESTINATION ${CMAKE_INSTALL_INCLUDEDIR}
+)
+if (${WITH_FIO})
+	install(TARGETS fio_vitastor fio_vitastor_blk fio_vitastor_sec LIBRARY DESTINATION ${CMAKE_INSTALL_LIBDIR})
+endif (${WITH_FIO})
+if (${WITH_QEMU})
+	install(TARGETS qemu_vitastor LIBRARY DESTINATION /usr/${CMAKE_INSTALL_LIBDIR}/${QEMU_PLUGINDIR})
+endif (${WITH_QEMU})

src/blockstore.cpp

@@ -48,6 +48,11 @@ std::map<uint64_t, uint64_t> & blockstore_t::get_inode_space_stats()
     return impl->inode_space_stats;
 }
 
+void blockstore_t::dump_diagnostics()
+{
+    return impl->dump_diagnostics();
+}
+
 uint32_t blockstore_t::get_block_size()
 {
     return impl->get_block_size();

src/blockstore.h

@@ -186,6 +186,9 @@ public:
     // Get per-inode space usage statistics
     std::map<uint64_t, uint64_t> & get_inode_space_stats();
 
+    // Print diagnostics to stdout
+    void dump_diagnostics();
+
     // FIXME rename to object_size
     uint32_t get_block_size();
     uint64_t get_block_count();

src/blockstore_flush.cpp

@@ -182,6 +182,75 @@ void journal_flusher_t::release_trim()
     trim_wanted--;
 }
 
+void journal_flusher_t::dump_diagnostics()
+{
+    const char *unflushable_type = "";
+    obj_ver_id unflushable = { 0 };
+    // Try to find out if there is a flushable object for information
+    for (object_id cur_oid: flush_queue)
+    {
+        obj_ver_id cur = { .oid = cur_oid, .version = flush_versions[cur_oid] };
+        auto dirty_end = bs->dirty_db.find(cur);
+        if (dirty_end == bs->dirty_db.end())
+        {
+            // Already flushed
+            continue;
+        }
+        auto repeat_it = sync_to_repeat.find(cur.oid);
+        if (repeat_it != sync_to_repeat.end())
+        {
+            // Someone is already flushing it
+            unflushable_type = "locked,";
+            unflushable = cur;
+            break;
+        }
+        if (dirty_end->second.journal_sector >= bs->journal.dirty_start &&
+            (bs->journal.dirty_start >= bs->journal.used_start ||
+            dirty_end->second.journal_sector < bs->journal.used_start))
+        {
+            // Object is more recent than possible to flush
+            bool found = try_find_older(dirty_end, cur);
+            if (!found)
+            {
+                unflushable_type = "dirty,";
+                unflushable = cur;
+                break;
+            }
+        }
+        unflushable_type = "ok,";
+        unflushable = cur;
+        break;
+    }
+    printf(
+        "Flusher: queued=%ld first=%s%lx:%lx trim_wanted=%d dequeuing=%d trimming=%d cur=%d target=%d active=%d syncing=%d\n",
+        flush_queue.size(), unflushable_type, unflushable.oid.inode, unflushable.oid.stripe,
+        trim_wanted, dequeuing, trimming, cur_flusher_count, target_flusher_count,
+        active_flushers, syncing_flushers
+    );
+}
+
+bool journal_flusher_t::try_find_older(std::map<obj_ver_id, dirty_entry>::iterator & dirty_end, obj_ver_id & cur)
+{
+    bool found = false;
+    while (dirty_end != bs->dirty_db.begin())
+    {
+        dirty_end--;
+        if (dirty_end->first.oid != cur.oid)
+        {
+            break;
+        }
+        if (!(dirty_end->second.journal_sector >= bs->journal.dirty_start &&
+            (bs->journal.dirty_start >= bs->journal.used_start ||
+            dirty_end->second.journal_sector < bs->journal.used_start)))
+        {
+            found = true;
+            cur.version = dirty_end->first.version;
+            break;
+        }
+    }
+    return found;
+}
+
 #define await_sqe(label) \
     resume_##label:\
         sqe = bs->get_sqe();\
@@ -286,30 +355,15 @@ stop_flusher:
             // And it may even block writes if we don't flush the older version
             // (if it's in the beginning of the journal)...
             // So first try to find an older version of the same object to flush.
-            bool found = false;
-            while (dirty_end != bs->dirty_db.begin())
-            {
-                dirty_end--;
-                if (dirty_end->first.oid != cur.oid)
-                {
-                    break;
-                }
-                if (!(dirty_end->second.journal_sector >= bs->journal.dirty_start &&
-                    (bs->journal.dirty_start >= bs->journal.used_start ||
-                    dirty_end->second.journal_sector < bs->journal.used_start)))
-                {
-                    found = true;
-                    cur.version = dirty_end->first.version;
-                    break;
-                }
-            }
+            bool found = flusher->try_find_older(dirty_end, cur);
             if (!found)
             {
                 // Try other objects
                 flusher->sync_to_repeat.erase(cur.oid);
                 int search_left = flusher->flush_queue.size() - 1;
 #ifdef BLOCKSTORE_DEBUG
-                printf("Flusher overran writers (dirty_start=%08lx) - searching for older flushes (%d left)\n", bs->journal.dirty_start, search_left);
+                printf("Flusher overran writers (%lx:%lx v%lu, dirty_start=%08lx) - searching for older flushes (%d left)\n",
+                    cur.oid.inode, cur.oid.stripe, cur.version, bs->journal.dirty_start, search_left);
 #endif
                 while (search_left > 0)
                 {
@@ -332,7 +386,12 @@ stop_flusher:
                         else
                         {
                             repeat_it = flusher->sync_to_repeat.find(cur.oid);
-                            if (repeat_it == flusher->sync_to_repeat.end())
+                            if (repeat_it != flusher->sync_to_repeat.end())
+                            {
+                                if (repeat_it->second < cur.version)
+                                    repeat_it->second = cur.version;
+                            }
+                            else
                             {
                                 flusher->sync_to_repeat[cur.oid] = 0;
                                 break;

src/blockstore_flush.h

@@ -97,6 +97,9 @@ class journal_flusher_t
     std::map<uint64_t, meta_sector_t> meta_sectors;
     std::deque<object_id> flush_queue;
     std::map<object_id, uint64_t> flush_versions;
+
+    bool try_find_older(std::map<obj_ver_id, dirty_entry>::iterator & dirty_end, obj_ver_id & cur);
+
 public:
     journal_flusher_t(blockstore_impl_t *bs);
     ~journal_flusher_t();
@@ -108,4 +111,5 @@ public:
     void enqueue_flush(obj_ver_id oid);
     void unshift_flush(obj_ver_id oid, bool force);
     void remove_flush(object_id oid);
+    void dump_diagnostics();
 };

src/blockstore_impl.cpp

@@ -595,3 +595,9 @@ void blockstore_impl_t::process_list(blockstore_op_t *op)
     op->buf = stable;
     FINISH_OP(op);
 }
+
+void blockstore_impl_t::dump_diagnostics()
+{
+    journal.dump_diagnostics();
+    flusher->dump_diagnostics();
+}

src/blockstore_impl.h

@@ -361,6 +361,9 @@ public:
     // Space usage statistics
     std::map<uint64_t, uint64_t> inode_space_stats;
 
+    // Print diagnostics to stdout
+    void dump_diagnostics();
+
     inline uint32_t get_block_size() { return block_size; }
     inline uint64_t get_block_count() { return block_count; }
     inline uint64_t get_free_block_count() { return data_alloc->get_free_count(); }

src/blockstore_journal.cpp

@@ -218,3 +218,19 @@ uint64_t journal_t::get_trim_pos()
     // Can't trim journal
     return used_start;
 }
+
+void journal_t::dump_diagnostics()
+{
+    auto journal_used_it = used_sectors.lower_bound(used_start);
+    if (journal_used_it == used_sectors.end())
+    {
+        // Journal is cleared to its end, restart from the beginning
+        journal_used_it = used_sectors.begin();
+    }
+    printf(
+        "Journal: used_start=%08lx next_free=%08lx dirty_start=%08lx trim_to=%08lx trim_to_refs=%ld\n",
+        used_start, next_free, dirty_start,
+        journal_used_it == used_sectors.end() ? 0 : journal_used_it->first,
+        journal_used_it == used_sectors.end() ? 0 : journal_used_it->second
+    );
+}

src/blockstore_journal.h

@@ -180,6 +180,7 @@ struct journal_t
     ~journal_t();
     bool trim();
     uint64_t get_trim_pos();
+    void dump_diagnostics();
     inline bool entry_fits(int size)
     {
         return !(block_size - in_sector_pos < size ||

src/blockstore_write.cpp

@@ -478,15 +478,15 @@ resume_2:
     }
 resume_4:
     // Switch object state
-#ifdef BLOCKSTORE_DEBUG
-    printf("Ack write %lx:%lx v%lu = state 0x%x\n", op->oid.inode, op->oid.stripe, op->version, dirty_it->second.state);
-#endif
     {
         auto dirty_it = dirty_db.find((obj_ver_id){
             .oid = op->oid,
             .version = op->version,
         });
         assert(dirty_it != dirty_db.end());
+#ifdef BLOCKSTORE_DEBUG
+        printf("Ack write %lx:%lx v%lu = state 0x%x\n", op->oid.inode, op->oid.stripe, op->version, dirty_it->second.state);
+#endif
         bool is_big = (dirty_it->second.state & BS_ST_TYPE_MASK) == BS_ST_BIG_WRITE;
         bool imm = is_big ? (immediate_commit == IMMEDIATE_ALL) : (immediate_commit != IMMEDIATE_NONE);
         if (imm)

src/cli.cpp

@@ -0,0 +1,251 @@
+// Copyright (c) Vitaliy Filippov, 2019+
+// License: VNPL-1.1 (see README.md for details)
+
+/**
+ * CLI tool
+ * Currently can (a) remove inodes and (b) merge snapshot/clone layers
+ */
+
+#include <vector>
+#include <algorithm>
+
+#include "cli.h"
+#include "epoll_manager.h"
+#include "cluster_client.h"
+#include "pg_states.h"
+#include "base64.h"
+
+static const char *exe_name = NULL;
+
+json11::Json::object cli_tool_t::parse_args(int narg, const char *args[])
+{
+    json11::Json::object cfg;
+    json11::Json::array cmd;
+    cfg["progress"] = "1";
+    for (int i = 1; i < narg; i++)
+    {
+        if (!strcmp(args[i], "-h") || !strcmp(args[i], "--help"))
+        {
+            help();
+        }
+        else if (args[i][0] == '-' && args[i][1] == '-')
+        {
+            const char *opt = args[i]+2;
+            cfg[opt] = !strcmp(opt, "json") || !strcmp(opt, "wait-list") || i == narg-1 ? "1" : args[++i];
+        }
+        else
+        {
+            cmd.push_back(std::string(args[i]));
+        }
+    }
+    if (!cmd.size())
+    {
+        std::string exe(exe_name);
+        if (exe.substr(exe.size()-11) == "vitastor-rm")
+        {
+            cmd.push_back("rm-data");
+        }
+    }
+    cfg["command"] = cmd;
+    return cfg;
+}
+
+void cli_tool_t::help()
+{
+    printf(
+        "Vitastor command-line tool\n"
+        "(c) Vitaliy Filippov, 2019+ (VNPL-1.1)\n\n"
+        "USAGE:\n"
+        "%s rm-data [OPTIONS] --pool <pool> --inode <inode> [--wait-list]\n"
+        "  Remove inode data without changing metadata.\n"
+        "  --wait-list means first retrieve objects listings and then remove it.\n"
+        "  --wait-list requires more memory, but allows to show correct stats.\n"
+        "\n"
+        "%s merge-data [OPTIONS] <from> <to> [--target <target>]\n"
+        "  Merge layer data without changing metadata. Merge <from>..<to> to <target>.\n"
+        "  <to> must be a child of <from> and <target> may be one of the layers between\n"
+        "  <from> and <to>, including <from> and <to>.\n"
+        "\n"
+        "%s flatten [OPTIONS] <layer>\n"
+        "  Flatten a layer, i.e. merge data and detach it from parents\n"
+        "\n"
+        "%s rm [OPTIONS] <from> [<to>] [--writers-stopped 1]\n"
+        "  Remove <from> or all layers between <from> and <to> (<to> must be a child of <from>),\n"
+        "  rebasing all their children accordingly. One of deleted parents may be renamed to one\n"
+        "  of children \"to be rebased\", but only if that child itself is readonly or if\n"
+        "  --writers-stopped 1 is specified\n"
+        "\n"
+        "OPTIONS (global):\n"
+        "  --etcd_address <etcd_address>\n"
+        "  --iodepth N         Send N operations in parallel to each OSD when possible (default 32)\n"
+        "  --parallel_osds M   Work with M osds in parallel when possible (default 4)\n"
+        "  --progress 1|0      Report progress (default 1)\n"
+        "  --cas 1|0           Use online CAS writes when possible (default auto)\n"
+        ,
+        exe_name, exe_name, exe_name, exe_name
+    );
+    exit(0);
+}
+
+void cli_tool_t::change_parent(inode_t cur, inode_t new_parent)
+{
+    auto cur_cfg_it = cli->st_cli.inode_config.find(cur);
+    if (cur_cfg_it == cli->st_cli.inode_config.end())
+    {
+        fprintf(stderr, "Inode 0x%lx disappeared\n", cur);
+        exit(1);
+    }
+    inode_config_t new_cfg = cur_cfg_it->second;
+    std::string cur_name = new_cfg.name;
+    std::string cur_cfg_key = base64_encode(cli->st_cli.etcd_prefix+
+        "/config/inode/"+std::to_string(INODE_POOL(cur))+
+        "/"+std::to_string(INODE_NO_POOL(cur)));
+    new_cfg.parent_id = new_parent;
+    json11::Json::object cur_cfg_json = cli->st_cli.serialize_inode_cfg(&new_cfg);
+    waiting++;
+    cli->st_cli.etcd_txn(json11::Json::object {
+        { "compare", json11::Json::array {
+            json11::Json::object {
+                { "target", "MOD" },
+                { "key", cur_cfg_key },
+                { "result", "LESS" },
+                { "mod_revision", new_cfg.mod_revision+1 },
+            },
+        } },
+        { "success", json11::Json::array {
+            json11::Json::object {
+                { "request_put", json11::Json::object {
+                    { "key", cur_cfg_key },
+                    { "value", base64_encode(json11::Json(cur_cfg_json).dump()) },
+                } }
+            },
+        } },
+    }, ETCD_SLOW_TIMEOUT, [this, new_parent, cur, cur_name](std::string err, json11::Json res)
+    {
+        if (err != "")
+        {
+            fprintf(stderr, "Error changing parent of %s: %s\n", cur_name.c_str(), err.c_str());
+            exit(1);
+        }
+        if (!res["succeeded"].bool_value())
+        {
+            fprintf(stderr, "Inode %s was modified during snapshot deletion\n", cur_name.c_str());
+            exit(1);
+        }
+        if (new_parent)
+        {
+            auto new_parent_it = cli->st_cli.inode_config.find(new_parent);
+            std::string new_parent_name = new_parent_it != cli->st_cli.inode_config.end()
+                ? new_parent_it->second.name : "<unknown>";
+            printf(
+                "Parent of layer %s (inode %lu in pool %u) changed to %s (inode %lu in pool %u)\n",
+                cur_name.c_str(), INODE_NO_POOL(cur), INODE_POOL(cur),
+                new_parent_name.c_str(), INODE_NO_POOL(new_parent), INODE_POOL(new_parent)
+            );
+        }
+        else
+        {
+            printf(
+                "Parent of layer %s (inode %lu in pool %u) detached\n",
+                cur_name.c_str(), INODE_NO_POOL(cur), INODE_POOL(cur)
+            );
+        }
+        waiting--;
+        ringloop->wakeup();
+    });
+}
+
+inode_config_t* cli_tool_t::get_inode_cfg(const std::string & name)
+{
+    for (auto & ic: cli->st_cli.inode_config)
+    {
+        if (ic.second.name == name)
+        {
+            return &ic.second;
+        }
+    }
+    fprintf(stderr, "Layer %s not found\n", name.c_str());
+    exit(1);
+}
+
+void cli_tool_t::run(json11::Json cfg)
+{
+    json11::Json::array cmd = cfg["command"].array_items();
+    if (!cmd.size())
+    {
+        fprintf(stderr, "command is missing\n");
+        exit(1);
+    }
+    else if (cmd[0] == "rm-data")
+    {
+        // Delete inode data
+        action_cb = start_rm(cfg);
+    }
+    else if (cmd[0] == "merge-data")
+    {
+        // Merge layer data without affecting metadata
+        action_cb = start_merge(cfg);
+    }
+    else if (cmd[0] == "flatten")
+    {
+        // Merge layer data without affecting metadata
+        action_cb = start_flatten(cfg);
+    }
+    else if (cmd[0] == "rm")
+    {
+        // Remove multiple snapshots and rebase their children
+        action_cb = start_snap_rm(cfg);
+    }
+    else
+    {
+        fprintf(stderr, "unknown command: %s\n", cmd[0].string_value().c_str());
+        exit(1);
+    }
+    iodepth = cfg["iodepth"].uint64_value();
+    if (!iodepth)
+        iodepth = 32;
+    parallel_osds = cfg["parallel_osds"].uint64_value();
+    if (!parallel_osds)
+        parallel_osds = 4;
+    log_level = cfg["log_level"].int64_value();
+    progress = cfg["progress"].uint64_value() ? true : false;
+    list_first = cfg["wait-list"].uint64_value() ? true : false;
+    // Create client
+    ringloop = new ring_loop_t(512);
+    epmgr = new epoll_manager_t(ringloop);
+    cli = new cluster_client_t(ringloop, epmgr->tfd, cfg);
+    cli->on_ready([this]()
+    {
+        // Initialize job
+        consumer.loop = [this]()
+        {
+            if (action_cb != NULL)
+            {
+                bool done = action_cb();
+                if (done)
+                {
+                    action_cb = NULL;
+                }
+            }
+            ringloop->submit();
+        };
+        ringloop->register_consumer(&consumer);
+        consumer.loop();
+    });
+    // Loop until it completes
+    while (action_cb != NULL)
+    {
+        ringloop->loop();
+        ringloop->wait();
+    }
+}
+
+int main(int narg, const char *args[])
+{
+    setvbuf(stdout, NULL, _IONBF, 0);
+    setvbuf(stderr, NULL, _IONBF, 0);
+    exe_name = args[0];
+    cli_tool_t *p = new cli_tool_t();
+    p->run(cli_tool_t::parse_args(narg, args));
+    return 0;
+}

src/cli.h

@@ -0,0 +1,56 @@
+// Copyright (c) Vitaliy Filippov, 2019+
+// License: VNPL-1.1 (see README.md for details)
+
+// Common CLI tool header
+
+#pragma once
+
+#include "json11/json11.hpp"
+#include "object_id.h"
+#include "ringloop.h"
+#include <functional>
+
+struct rm_inode_t;
+struct snap_merger_t;
+struct snap_flattener_t;
+struct snap_remover_t;
+
+class epoll_manager_t;
+class cluster_client_t;
+struct inode_config_t;
+
+class cli_tool_t
+{
+public:
+    uint64_t iodepth = 0, parallel_osds = 0;
+    bool progress = true;
+    bool list_first = false;
+    int log_level = 0;
+    int mode = 0;
+
+    ring_loop_t *ringloop = NULL;
+    epoll_manager_t *epmgr = NULL;
+    cluster_client_t *cli = NULL;
+
+    int waiting = 0;
+    ring_consumer_t consumer;
+    std::function<bool(void)> action_cb;
+
+    void run(json11::Json cfg);
+
+    void change_parent(inode_t cur, inode_t new_parent);
+    inode_config_t* get_inode_cfg(const std::string & name);
+
+    static json11::Json::object parse_args(int narg, const char *args[]);
+    static void help();
+
+    friend struct rm_inode_t;
+    friend struct snap_merger_t;
+    friend struct snap_flattener_t;
+    friend struct snap_remover_t;
+
+    std::function<bool(void)> start_rm(json11::Json);
+    std::function<bool(void)> start_merge(json11::Json);
+    std::function<bool(void)> start_flatten(json11::Json);
+    std::function<bool(void)> start_snap_rm(json11::Json);
+};

src/cli_flatten.cpp

@@ -0,0 +1,124 @@
+// Copyright (c) Vitaliy Filippov, 2019+
+// License: VNPL-1.1 (see README.md for details)
+
+#include "cli.h"
+#include "cluster_client.h"
+
+// Flatten a layer: merge all parents into a layer and break the connection completely
+struct snap_flattener_t
+{
+    cli_tool_t *parent;
+
+    // target to flatten
+    std::string target_name;
+    // writers are stopped, we can safely change writable layers
+    bool writers_stopped = false;
+    // use CAS writes (0 = never, 1 = auto, 2 = always)
+    int use_cas = 1;
+    // interval between fsyncs
+    int fsync_interval = 128;
+
+    std::string top_parent_name;
+    inode_t target_id = 0;
+    int state = 0;
+    std::function<bool(void)> merger_cb;
+
+    void get_merge_parents()
+    {
+        // Get all parents of target
+        inode_config_t *target_cfg = parent->get_inode_cfg(target_name);
+        target_id = target_cfg->num;
+        std::vector<inode_t> chain_list;
+        inode_config_t *cur = target_cfg;
+        chain_list.push_back(cur->num);
+        while (cur->parent_id != 0 && cur->parent_id != target_cfg->num)
+        {
+            auto it = parent->cli->st_cli.inode_config.find(cur->parent_id);
+            if (it == parent->cli->st_cli.inode_config.end())
+            {
+                fprintf(stderr, "Parent inode of layer %s (id %ld) not found\n", cur->name.c_str(), cur->parent_id);
+                exit(1);
+            }
+            cur = &it->second;
+            chain_list.push_back(cur->num);
+        }
+        if (cur->parent_id != 0)
+        {
+            fprintf(stderr, "Layer %s has a loop in parents\n", target_name.c_str());
+            exit(1);
+        }
+        top_parent_name = cur->name;
+    }
+
+    bool is_done()
+    {
+        return state == 5;
+    }
+
+    void loop()
+    {
+        if (state == 1)
+            goto resume_1;
+        else if (state == 2)
+            goto resume_2;
+        else if (state == 3)
+            goto resume_3;
+        // Get parent layers
+        get_merge_parents();
+        // Start merger
+        merger_cb = parent->start_merge(json11::Json::object {
+            { "command", json11::Json::array{ "merge-data", top_parent_name, target_name } },
+            { "target", target_name },
+            { "delete-source", false },
+            { "cas", use_cas },
+            { "fsync-interval", fsync_interval },
+        });
+        // Wait for it
+resume_1:
+        while (!merger_cb())
+        {
+            state = 1;
+            return;
+        }
+        merger_cb = NULL;
+        // Change parent
+        parent->change_parent(target_id, 0);
+        // Wait for it to complete
+        state = 2;
+resume_2:
+        if (parent->waiting > 0)
+            return;
+        state = 3;
+resume_3:
+        // Done
+        return;
+    }
+};
+
+std::function<bool(void)> cli_tool_t::start_flatten(json11::Json cfg)
+{
+    json11::Json::array cmd = cfg["command"].array_items();
+    auto flattener = new snap_flattener_t();
+    flattener->parent = this;
+    flattener->target_name = cmd.size() > 1 ? cmd[1].string_value() : "";
+    if (flattener->target_name == "")
+    {
+        fprintf(stderr, "Layer to flatten argument is missing\n");
+        exit(1);
+    }
+    flattener->fsync_interval = cfg["fsync-interval"].uint64_value();
+    if (!flattener->fsync_interval)
+        flattener->fsync_interval = 128;
+    if (!cfg["cas"].is_null())
+        flattener->use_cas = cfg["cas"].uint64_value() ? 2 : 0;
+    return [flattener]()
+    {
+        flattener->loop();
+        if (flattener->is_done())
+        {
+            delete flattener;
+            return true;
+        }
+        return false;
+    };
+}

src/cli_merge.cpp

@@ -0,0 +1,583 @@
+// Copyright (c) Vitaliy Filippov, 2019+
+// License: VNPL-1.1 (see README.md for details)
+
+#include "cli.h"
+#include "cluster_client.h"
+#include "cpp-btree/safe_btree_set.h"
+
+struct snap_rw_op_t
+{
+    uint64_t offset = 0;
+    void *buf = NULL;
+    cluster_op_t op;
+    int todo = 0;
+    uint32_t start = 0, end = 0;
+};
+
+// Layer merge is the base for multiple operations:
+// 1) Delete snapshot "up" = merge child layer into the parent layer, remove the child
+//    and rename the parent to the child
+// 2) Delete snapshot "down" = merge parent layer into the child layer and remove the parent
+// 3) Flatten image = merge parent layers into the child layer and break the connection
+struct snap_merger_t
+{
+    cli_tool_t *parent;
+
+    // -- CONFIGURATION --
+    // merge from..to into target (target may be one of from..to)
+    std::string from_name, to_name, target_name;
+    // inode=>rank (bigger rank means child layers)
+    std::map<inode_t,int> sources;
+    // delete merged source inode data during merge
+    bool delete_source = false;
+    // use CAS writes (0 = never, 1 = auto, 2 = always)
+    int use_cas = 1;
+    // don't necessarily delete source data, but perform checks as if we were to do it
+    bool check_delete_source = false;
+    // interval between fsyncs
+    int fsync_interval = 128;
+
+    // -- STATE --
+    inode_t target;
+    int target_rank;
+    bool inside_continue = false;
+    int state = 0;
+    int lists_todo = 0;
+    uint64_t target_block_size = 0;
+    btree::safe_btree_set<uint64_t> merge_offsets;
+    btree::safe_btree_set<uint64_t>::iterator oit;
+    std::map<inode_t, std::vector<uint64_t>> layer_lists;
+    std::map<inode_t, uint64_t> layer_block_size;
+    std::map<inode_t, uint64_t> layer_list_pos;
+    int in_flight = 0;
+    uint64_t last_fsync_offset = 0;
+    uint64_t last_written_offset = 0;
+    int deleted_unsynced = 0;
+    uint64_t processed = 0, to_process = 0;
+
+    void start_merge()
+    {
+        check_delete_source = delete_source || check_delete_source;
+        inode_config_t *from_cfg = parent->get_inode_cfg(from_name);
+        inode_config_t *to_cfg = parent->get_inode_cfg(to_name);
+        inode_config_t *target_cfg = target_name == "" ? from_cfg : parent->get_inode_cfg(target_name);
+        if (to_cfg->num == from_cfg->num)
+        {
+            fprintf(stderr, "Only one layer specified, nothing to merge\n");
+            exit(1);
+        }
+        // Check that to_cfg is actually a child of from_cfg and target_cfg is somewhere between them
+        std::vector<inode_t> chain_list;
+        inode_config_t *cur = to_cfg;
+        chain_list.push_back(cur->num);
+        layer_block_size[cur->num] = get_block_size(cur->num);
+        while (cur->parent_id != from_cfg->num &&
+            cur->parent_id != to_cfg->num &&
+            cur->parent_id != 0)
+        {
+            auto it = parent->cli->st_cli.inode_config.find(cur->parent_id);
+            if (it == parent->cli->st_cli.inode_config.end())
+            {
+                fprintf(stderr, "Parent inode of layer %s (id %ld) not found\n", cur->name.c_str(), cur->parent_id);
+                exit(1);
+            }
+            cur = &it->second;
+            chain_list.push_back(cur->num);
+            layer_block_size[cur->num] = get_block_size(cur->num);
+        }
+        if (cur->parent_id != from_cfg->num)
+        {
+            fprintf(stderr, "Layer %s is not a child of %s\n", to_name.c_str(), from_name.c_str());
+            exit(1);
+        }
+        chain_list.push_back(from_cfg->num);
+        layer_block_size[from_cfg->num] = get_block_size(from_cfg->num);
+        int i = chain_list.size()-1;
+        for (inode_t item: chain_list)
+        {
+            sources[item] = i--;
+        }
+        if (sources.find(target_cfg->num) == sources.end())
+        {
+            fprintf(stderr, "Layer %s is not between %s and %s\n", target_name.c_str(), to_name.c_str(), from_name.c_str());
+            exit(1);
+        }
+        target = target_cfg->num;
+        target_rank = sources.at(target);
+        int to_rank = sources.at(to_cfg->num);
+        bool to_has_children = false;
+        // Check that there are no other inodes dependent on altered layers
+        //
+        // 1) everything between <target> and <to> except <to> is not allowed
+        //    to have children other than <to> if <to> is a child of <target>:
+        //
+        //    <target> - <layer 3> - <to>
+        //            \- <layer 4> <--------X--------- NOT ALLOWED
+        //
+        // 2) everything between <from> and <target>, except <target>, is not allowed
+        //    to have children other than <target> if sources are to be deleted after merging:
+        //
+        //    <from> - <layer 1> - <target> - <to>
+        //          \- <layer 2> <---------X-------- NOT ALLOWED
+        for (auto & ic: parent->cli->st_cli.inode_config)
+        {
+            auto it = sources.find(ic.second.num);
+            if (it == sources.end() && ic.second.parent_id != 0)
+            {
+                it = sources.find(ic.second.parent_id);
+                if (it != sources.end())
+                {
+                    int parent_rank = it->second;
+                    if (parent_rank < to_rank && (parent_rank >= target_rank || check_delete_source))
+                    {
+                        fprintf(
+                            stderr, "Layers at or above %s, but below %s are not allowed"
+                                " to have other children, but %s is a child of %s\n",
+                            (check_delete_source ? from_name.c_str() : target_name.c_str()),
+                            to_name.c_str(), ic.second.name.c_str(),
+                            parent->cli->st_cli.inode_config.at(ic.second.parent_id).name.c_str()
+                        );
+                        exit(1);
+                    }
+                    if (parent_rank >= to_rank)
+                    {
+                        to_has_children = true;
+                    }
+                }
+            }
+        }
+        if ((target_rank < to_rank || to_has_children) && use_cas == 1)
+        {
+            // <to> has children itself, no need for CAS
+            use_cas = 0;
+        }
+        sources.erase(target);
+        printf(
+            "Merging %ld layer(s) into target %s%s (inode %lu in pool %u)\n",
+            sources.size(), target_cfg->name.c_str(),
+            use_cas ? " online (with CAS)" : "", INODE_NO_POOL(target), INODE_POOL(target)
+        );
+        target_block_size = get_block_size(target);
+    }
+
+    uint64_t get_block_size(inode_t inode)
+    {
+        auto & pool_cfg = parent->cli->st_cli.pool_config.at(INODE_POOL(inode));
+        uint64_t pg_data_size = (pool_cfg.scheme == POOL_SCHEME_REPLICATED ? 1 : pool_cfg.pg_size-pool_cfg.parity_chunks);
+        return parent->cli->get_bs_block_size() * pg_data_size;
+    }
+
+    void continue_merge_reent()
+    {
+        if (!inside_continue)
+        {
+            inside_continue = true;
+            continue_merge();
+            inside_continue = false;
+        }
+    }
+
+    bool is_done()
+    {
+        return state == 6;
+    }
+
+    void continue_merge()
+    {
+        if (state == 1)
+            goto resume_1;
+        else if (state == 2)
+            goto resume_2;
+        else if (state == 3)
+            goto resume_3;
+        else if (state == 4)
+            goto resume_4;
+        else if (state == 5)
+            goto resume_5;
+        else if (state == 6)
+            goto resume_6;
+        // Get parents and so on
+        start_merge();
+        // First list lower layers
+        list_layers(true);
+        state = 1;
+    resume_1:
+        while (lists_todo > 0)
+        {
+            // Wait for lists
+            return;
+        }
+        if (merge_offsets.size() > 0)
+        {
+            state = 2;
+            oit = merge_offsets.begin();
+            processed = 0;
+            to_process = merge_offsets.size();
+    resume_2:
+            // Then remove blocks already filled in target by issuing zero-length reads and checking bitmaps
+            while (in_flight < parent->iodepth*parent->parallel_osds && oit != merge_offsets.end())
+            {
+                in_flight++;
+                check_if_full(*oit);
+                oit++;
+                processed++;
+                if (parent->progress && !(processed % 128))
+                {
+                    printf("\rFiltering target blocks: %lu/%lu", processed, to_process);
+                }
+            }
+            if (in_flight > 0 || oit != merge_offsets.end())
+            {
+                // Wait until reads finish
+                return;
+            }
+            if (parent->progress)
+            {
+                printf("\r%lu full blocks of target filtered out\n", to_process-merge_offsets.size());
+            }
+        }
+        state = 3;
+    resume_3:
+        // Then list upper layers
+        list_layers(false);
+        state = 4;
+    resume_4:
+        while (lists_todo > 0)
+        {
+            // Wait for lists
+            return;
+        }
+        state = 5;
+        processed = 0;
+        to_process = merge_offsets.size();
+        oit = merge_offsets.begin();
+    resume_5:
+        // Now read, overwrite and optionally delete offsets one by one
+        while (in_flight < parent->iodepth*parent->parallel_osds && oit != merge_offsets.end())
+        {
+            in_flight++;
+            read_and_write(*oit);
+            oit++;
+            processed++;
+            if (parent->progress && !(processed % 128))
+            {
+                printf("\rOverwriting blocks: %lu/%lu", processed, to_process);
+            }
+        }
+        if (in_flight > 0 || oit != merge_offsets.end())
+        {
+            // Wait until overwrites finish
+            return;
+        }
+        if (parent->progress)
+        {
+            printf("\rOverwriting blocks: %lu/%lu\n", to_process, to_process);
+        }
+        // Done
+        printf("Done, layers from %s to %s merged into %s\n", from_name.c_str(), to_name.c_str(), target_name.c_str());
+        state = 6;
+    resume_6:
+        return;
+    }
+
+    void list_layers(bool lower)
+    {
+        for (auto & sp: sources)
+        {
+            inode_t src = sp.first;
+            if (lower ? (sp.second < target_rank) : (sp.second > target_rank))
+            {
+                lists_todo++;
+                inode_list_t* lst = parent->cli->list_inode_start(src, [this, src](
+                    inode_list_t *lst, std::set<object_id>&& objects, pg_num_t pg_num, osd_num_t primary_osd, int status)
+                {
+                    uint64_t layer_block = layer_block_size.at(src);
+                    for (object_id obj: objects)
+                    {
+                        merge_offsets.insert(obj.stripe - obj.stripe % target_block_size);
+                        for (int i = target_block_size; i < layer_block; i += target_block_size)
+                        {
+                            merge_offsets.insert(obj.stripe - obj.stripe % target_block_size + i);
+                        }
+                    }
+                    if (delete_source)
+                    {
+                        // Also store individual lists
+                        auto & layer_list = layer_lists[src];
+                        int pos = layer_list.size();
+                        layer_list.resize(pos + objects.size());
+                        for (object_id obj: objects)
+                        {
+                            layer_list[pos++] = obj.stripe;
+                        }
+                    }
+                    if (status & INODE_LIST_DONE)
+                    {
+                        auto & name = parent->cli->st_cli.inode_config.at(src).name;
+                        printf("Got listing of layer %s (inode %lu in pool %u)\n", name.c_str(), INODE_NO_POOL(src), INODE_POOL(src));
+                        if (delete_source)
+                        {
+                            // Sort the inode listing
+                            std::sort(layer_lists[src].begin(), layer_lists[src].end());
+                        }
+                        lists_todo--;
+                        continue_merge_reent();
+                    }
+                    else
+                    {
+                        parent->cli->list_inode_next(lst, 1);
+                    }
+                });
+                parent->cli->list_inode_next(lst, parent->parallel_osds);
+            }
+        }
+    }
+
+    // Check if <offset> is fully written in <target> and remove it from merge_offsets if so
+    void check_if_full(uint64_t offset)
+    {
+        cluster_op_t *op = new cluster_op_t;
+        op->opcode = OSD_OP_READ_BITMAP;
+        op->inode = target;
+        op->offset = offset;
+        op->len = 0;
+        op->callback = [this](cluster_op_t *op)
+        {
+            if (op->retval < 0)
+            {
+                fprintf(stderr, "error reading target bitmap at offset %lx: %s\n", op->offset, strerror(-op->retval));
+            }
+            else
+            {
+                uint64_t bitmap_bytes = target_block_size/parent->cli->get_bs_bitmap_granularity()/8;
+                int i;
+                for (i = 0; i < bitmap_bytes; i++)
+                {
+                    if (((uint8_t*)op->bitmap_buf)[i] != 0xff)
+                    {
+                        break;
+                    }
+                }
+                if (i == bitmap_bytes)
+                {
+                    // full
+                    merge_offsets.erase(op->offset);
+                }
+            }
+            delete op;
+            in_flight--;
+            continue_merge_reent();
+        };
+        parent->cli->execute(op);
+    }
+
+    // Read <offset> from <to>, write it to <target> and optionally delete it
+    // from all layers except <target> after fsync'ing
+    void read_and_write(uint64_t offset)
+    {
+        snap_rw_op_t *rwo = new snap_rw_op_t;
+        // Initialize counter to 1 to later allow write_subop() to return immediately
+        // (even though it shouldn't really do that)
+        rwo->todo = 1;
+        rwo->buf = malloc(target_block_size);
+        rwo->offset = offset;
+        rwo_read(rwo);
+    }
+
+    void rwo_read(snap_rw_op_t *rwo)
+    {
+        cluster_op_t *op = &rwo->op;
+        op->opcode = OSD_OP_READ;
+        op->inode = target;
+        op->offset = rwo->offset;
+        op->len = target_block_size;
+        op->iov.push_back(rwo->buf, target_block_size);
+        op->callback = [this, rwo](cluster_op_t *op)
+        {
+            if (op->retval != op->len)
+            {
+                fprintf(stderr, "error reading target at offset %lx: %s\n", op->offset, strerror(-op->retval));
+                exit(1);
+            }
+            next_write(rwo);
+        };
+        parent->cli->execute(op);
+    }
+
+    void next_write(snap_rw_op_t *rwo)
+    {
+        // Write each non-empty range using an individual operation
+        // FIXME: Allow to use single write with "holes" (OSDs don't allow it yet)
+        uint32_t gran = parent->cli->get_bs_bitmap_granularity();
+        uint64_t bitmap_size = target_block_size / gran;
+        while (rwo->end < bitmap_size)
+        {
+            auto bit = ((*(uint8_t*)(rwo->op.bitmap_buf + (rwo->end >> 3))) & (1 << (rwo->end & 0x7)));
+            if (!bit)
+            {
+                if (rwo->end > rwo->start)
+                {
+                    // write start->end
+                    rwo->todo++;
+                    write_subop(rwo, rwo->start*gran, rwo->end*gran, use_cas ? 1+rwo->op.version : 0);
+                    rwo->start = rwo->end;
+                    if (use_cas)
+                    {
+                        // Submit one by one if using CAS writes
+                        return;
+                    }
+                }
+                rwo->start = rwo->end = rwo->end+1;
+            }
+            else
+            {
+                rwo->end++;
+            }
+        }
+        if (rwo->end > rwo->start)
+        {
+            // write start->end
+            rwo->todo++;
+            write_subop(rwo, rwo->start*gran, rwo->end*gran, use_cas ? 1+rwo->op.version : 0);
+            rwo->start = rwo->end;
+            if (use_cas)
+            {
+                return;
+            }
+        }
+        rwo->todo--;
+        // Just in case, if everything is done
+        autofree_op(rwo);
+    }
+
+    void write_subop(snap_rw_op_t *rwo, uint32_t start, uint32_t end, uint64_t version)
+    {
+        cluster_op_t *subop = new cluster_op_t;
+        subop->opcode = OSD_OP_WRITE;
+        subop->inode = target;
+        subop->offset = rwo->offset+start;
+        subop->len = end-start;
+        subop->version = version;
+        subop->flags = OSD_OP_IGNORE_READONLY;
+        subop->iov.push_back(rwo->buf+start, end-start);
+        subop->callback = [this, rwo](cluster_op_t *subop)
+        {
+            rwo->todo--;
+            if (subop->retval != subop->len)
+            {
+                if (use_cas && subop->retval == -EINTR)
+                {
+                    // CAS failure - reread and repeat optimistically
+                    rwo->start = subop->offset - rwo->offset;
+                    rwo_read(rwo);
+                    delete subop;
+                    return;
+                }
+                fprintf(stderr, "error writing target at offset %lx: %s\n", subop->offset, strerror(-subop->retval));
+                exit(1);
+            }
+            // Increment CAS version
+            rwo->op.version++;
+            if (use_cas)
+                next_write(rwo);
+            else
+                autofree_op(rwo);
+            delete subop;
+        };
+        parent->cli->execute(subop);
+    }
+
+    void delete_offset(inode_t inode_num, uint64_t offset)
+    {
+        cluster_op_t *subop = new cluster_op_t;
+        subop->opcode = OSD_OP_DELETE;
+        subop->inode = inode_num;
+        subop->offset = offset;
+        subop->len = 0;
+        subop->flags = OSD_OP_IGNORE_READONLY;
+        subop->callback = [this](cluster_op_t *subop)
+        {
+            if (subop->retval != 0)
+            {
+                fprintf(stderr, "error deleting from layer 0x%lx at offset %lx: %s", subop->inode, subop->offset, strerror(-subop->retval));
+            }
+            delete subop;
+        };
+        parent->cli->execute(subop);
+    }
+
+    void autofree_op(snap_rw_op_t *rwo)
+    {
+        if (!rwo->todo)
+        {
+            if (last_written_offset < rwo->op.offset+target_block_size)
+            {
+                last_written_offset = rwo->op.offset+target_block_size;
+            }
+            if (delete_source)
+            {
+                deleted_unsynced++;
+                if (deleted_unsynced >= fsync_interval)
+                {
+                    uint64_t from = last_fsync_offset, to = last_written_offset;
+                    cluster_op_t *subop = new cluster_op_t;
+                    subop->opcode = OSD_OP_SYNC;
+                    subop->callback = [this, from, to](cluster_op_t *subop)
+                    {
+                        delete subop;
+                        // We can now delete source data between <from> and <to>
+                        // But to do this we have to keep all object lists in memory :-(
+                        for (auto & lp: layer_list_pos)
+                        {
+                            auto & layer_list = layer_lists.at(lp.first);
+                            uint64_t layer_block = layer_block_size.at(lp.first);
+                            int cur_pos = lp.second;
+                            while (cur_pos < layer_list.size() && layer_list[cur_pos]+layer_block < to)
+                            {
+                                delete_offset(lp.first, layer_list[cur_pos]);
+                                cur_pos++;
+                            }
+                            lp.second = cur_pos;
+                        }
+                    };
+                    parent->cli->execute(subop);
+                }
+            }
+            free(rwo->buf);
+            delete rwo;
+            in_flight--;
+            continue_merge_reent();
+        }
+    }
+};
+
+std::function<bool(void)> cli_tool_t::start_merge(json11::Json cfg)
+{
+    json11::Json::array cmd = cfg["command"].array_items();
+    auto merger = new snap_merger_t();
+    merger->parent = this;
+    merger->from_name = cmd.size() > 1 ? cmd[1].string_value() : "";
+    merger->to_name = cmd.size() > 2 ? cmd[2].string_value() : "";
+    merger->target_name = cfg["target"].string_value();
+    if (merger->from_name == "" || merger->to_name == "")
+    {
+        fprintf(stderr, "Beginning or end of the merge sequence is missing\n");
+        exit(1);
+    }
+    merger->delete_source = cfg["delete-source"].string_value() != "";
+    merger->fsync_interval = cfg["fsync-interval"].uint64_value();
+    if (!merger->fsync_interval)
+        merger->fsync_interval = 128;
+    if (!cfg["cas"].is_null())
+        merger->use_cas = cfg["cas"].uint64_value() ? 2 : 0;
+    return [merger]()
+    {
+        merger->continue_merge_reent();
+        if (merger->is_done())
+        {
+            delete merger;
+            return true;
+        }
+        return false;
+    };
+}

src/cli_rm.cpp

@@ -0,0 +1,195 @@
+// Copyright (c) Vitaliy Filippov, 2019+
+// License: VNPL-1.1 (see README.md for details)
+
+#include "cli.h"
+#include "cluster_client.h"
+
+#define RM_LISTING 1
+#define RM_REMOVING 2
+#define RM_END 3
+
+struct rm_pg_t
+{
+    pg_num_t pg_num;
+    osd_num_t rm_osd_num;
+    std::set<object_id> objects;
+    std::set<object_id>::iterator obj_pos;
+    uint64_t obj_count = 0, obj_done = 0, obj_prev_done = 0;
+    int state = 0;
+    int in_flight = 0;
+};
+
+struct rm_inode_t
+{
+    uint64_t inode = 0;
+    pool_id_t pool_id = 0;
+
+    cli_tool_t *parent = NULL;
+    inode_list_t *lister = NULL;
+    std::vector<rm_pg_t*> lists;
+    uint64_t total_count = 0, total_done = 0, total_prev_pct = 0;
+    uint64_t pgs_to_list = 0;
+    bool lists_done = false;
+    int state = 0;
+
+    void start_delete()
+    {
+        lister = parent->cli->list_inode_start(inode, [this](inode_list_t *lst,
+            std::set<object_id>&& objects, pg_num_t pg_num, osd_num_t primary_osd, int status)
+        {
+            rm_pg_t *rm = new rm_pg_t((rm_pg_t){
+                .pg_num = pg_num,
+                .rm_osd_num = primary_osd,
+                .objects = objects,
+                .obj_count = objects.size(),
+                .obj_done = 0,
+                .obj_prev_done = 0,
+            });
+            rm->obj_pos = rm->objects.begin();
+            lists.push_back(rm);
+            if (parent->list_first)
+            {
+                parent->cli->list_inode_next(lister, 1);
+            }
+            if (status & INODE_LIST_DONE)
+            {
+                lists_done = true;
+            }
+            pgs_to_list--;
+            continue_delete();
+        });
+        if (!lister)
+        {
+            fprintf(stderr, "Failed to list inode %lu from pool %u objects\n", INODE_NO_POOL(inode), INODE_POOL(inode));
+            exit(1);
+        }
+        pgs_to_list = parent->cli->list_pg_count(lister);
+        parent->cli->list_inode_next(lister, parent->parallel_osds);
+    }
+
+    void send_ops(rm_pg_t *cur_list)
+    {
+        if (parent->cli->msgr.osd_peer_fds.find(cur_list->rm_osd_num) ==
+            parent->cli->msgr.osd_peer_fds.end())
+        {
+            // Initiate connection
+            parent->cli->msgr.connect_peer(cur_list->rm_osd_num, parent->cli->st_cli.peer_states[cur_list->rm_osd_num]);
+            return;
+        }
+        while (cur_list->in_flight < parent->iodepth && cur_list->obj_pos != cur_list->objects.end())
+        {
+            osd_op_t *op = new osd_op_t();
+            op->op_type = OSD_OP_OUT;
+            op->peer_fd = parent->cli->msgr.osd_peer_fds[cur_list->rm_osd_num];
+            op->req = (osd_any_op_t){
+                .rw = {
+                    .header = {
+                        .magic = SECONDARY_OSD_OP_MAGIC,
+                        .id = parent->cli->next_op_id(),
+                        .opcode = OSD_OP_DELETE,
+                    },
+                    .inode = cur_list->obj_pos->inode,
+                    .offset = cur_list->obj_pos->stripe,
+                    .len = 0,
+                },
+            };
+            op->callback = [this, cur_list](osd_op_t *op)
+            {
+                cur_list->in_flight--;
+                if (op->reply.hdr.retval < 0)
+                {
+                    fprintf(stderr, "Failed to remove object %lx:%lx from PG %u (OSD %lu) (retval=%ld)\n",
+                        op->req.rw.inode, op->req.rw.offset,
+                        cur_list->pg_num, cur_list->rm_osd_num, op->reply.hdr.retval);
+                }
+                delete op;
+                cur_list->obj_done++;
+                total_done++;
+                continue_delete();
+            };
+            cur_list->obj_pos++;
+            cur_list->in_flight++;
+            parent->cli->msgr.outbox_push(op);
+        }
+    }
+
+    void continue_delete()
+    {
+        if (parent->list_first && !lists_done)
+        {
+            return;
+        }
+        for (int i = 0; i < lists.size(); i++)
+        {
+            if (!lists[i]->in_flight && lists[i]->obj_pos == lists[i]->objects.end())
+            {
+                delete lists[i];
+                lists.erase(lists.begin()+i, lists.begin()+i+1);
+                i--;
+                if (!lists_done)
+                {
+                    parent->cli->list_inode_next(lister, 1);
+                }
+            }
+            else
+            {
+                send_ops(lists[i]);
+            }
+        }
+        if (parent->progress && total_count > 0 && total_done*1000/total_count != total_prev_pct)
+        {
+            printf("\rRemoved %lu/%lu objects, %lu more PGs to list...", total_done, total_count, pgs_to_list);
+            total_prev_pct = total_done*1000/total_count;
+        }
+        if (lists_done && !lists.size())
+        {
+            printf("Done, inode %lu in pool %u data removed\n", INODE_NO_POOL(inode), pool_id);
+            state = 2;
+        }
+    }
+
+    bool loop()
+    {
+        if (state == 0)
+        {
+            start_delete();
+            state = 1;
+        }
+        else if (state == 1)
+        {
+            continue_delete();
+        }
+        else if (state == 2)
+        {
+            return true;
+        }
+        return false;
+    }
+};
+
+std::function<bool(void)> cli_tool_t::start_rm(json11::Json cfg)
+{
+    auto remover = new rm_inode_t();
+    remover->parent = this;
+    remover->inode = cfg["inode"].uint64_value();
+    remover->pool_id = cfg["pool"].uint64_value();
+    if (remover->pool_id)
+    {
+        remover->inode = (remover->inode & ((1l << (64-POOL_ID_BITS)) - 1)) | (((uint64_t)remover->pool_id) << (64-POOL_ID_BITS));
+    }
+    remover->pool_id = INODE_POOL(remover->inode);
+    if (!remover->pool_id)
+    {
+        fprintf(stderr, "pool is missing\n");
+        exit(1);
+    }
+    return [remover]()
+    {
+        if (remover->loop())
+        {
+            delete remover;
+            return true;
+        }
+        return false;
+    };
+}

src/cli_snap_rm.cpp

@@ -0,0 +1,565 @@
+// Copyright (c) Vitaliy Filippov, 2019+
+// License: VNPL-1.1 (see README.md for details)
+
+#include "cli.h"
+#include "cluster_client.h"
+#include "base64.h"
+
+// Remove layer(s): similar to merge, but alters metadata and processes multiple merge targets
+//
+// Exactly one child of the requested layers may be merged using the "inverted" workflow,
+// where we merge it "down" into one of the "to-be-removed" layers and then rename the
+// "to-be-removed" layer to the child. It may be done either if all writers are stopped
+// before trying to delete layers (which is signaled by --writers-stopped) or if that child
+// is a read-only layer (snapshot) itself.
+//
+// This "inverted" workflow trades copying data of one of the deleted layers for copying
+// data of one child of the chain which is also a child of the "traded" layer. So we
+// choose the (parent,child) pair which has the largest difference between "parent" and
+// "child" inode sizes.
+//
+// All other children of the chain are processed by iterating though them, merging removed
+// parents into them and rebasing them to the last layer which isn't a member of the removed
+// chain.
+//
+// Example:
+//
+// <parent> - <from> - <layer 2> - <to> - <child 1>
+//                 \           \       \- <child 2>
+//                  \           \- <child 3>
+//                   \-<child 4>
+//
+// 1) Find optimal pair for the "reverse" scenario
+//    Imagine that it's (<layer 2>, <child 1>) in this example
+// 2) Process all children except <child 1>:
+//    - Merge <from>..<to> to <child 2>
+//    - Set <child 2> parent to <parent>
+//    - Repeat for others
+// 3) Process <child 1>:
+//    - Merge <from>..<child 1> to <layer 2>
+//    - Set <layer 2> parent to <parent>
+//    - Rename <layer 2> to <child 1>
+// 4) Delete other layers of the chain (<from>, <to>)
+struct snap_remover_t
+{
+    cli_tool_t *parent;
+
+    // remove from..to
+    std::string from_name, to_name;
+    // writers are stopped, we can safely change writable layers
+    bool writers_stopped = false;
+    // use CAS writes (0 = never, 1 = auto, 2 = always)
+    int use_cas = 1;
+    // interval between fsyncs
+    int fsync_interval = 128;
+
+    std::map<inode_t,int> sources;
+    std::map<inode_t,uint64_t> inode_used;
+    std::vector<inode_t> merge_children;
+    std::vector<inode_t> chain_list;
+    std::map<inode_t,int> inverse_candidates;
+    inode_t inverse_parent = 0, inverse_child = 0;
+    inode_t new_parent = 0;
+    int state = 0;
+    int current_child = 0;
+    std::function<bool(void)> cb;
+
+    bool is_done()
+    {
+        return state == 9;
+    }
+
+    void loop()
+    {
+        if (state == 1)
+            goto resume_1;
+        else if (state == 2)
+            goto resume_2;
+        else if (state == 3)
+            goto resume_3;
+        else if (state == 4)
+            goto resume_4;
+        else if (state == 5)
+            goto resume_5;
+        else if (state == 6)
+            goto resume_6;
+        else if (state == 7)
+            goto resume_7;
+        else if (state == 8)
+            goto resume_8;
+        else if (state == 9)
+            goto resume_9;
+        // Get children to merge
+        get_merge_children();
+        // Try to select an inode for the "inverse" optimized scenario
+        // Read statistics from etcd to do it
+        read_stats();
+        state = 1;
+resume_1:
+        if (parent->waiting > 0)
+            return;
+        choose_inverse_candidate();
+        // Merge children one by one, except our "inverse" child
+        for (current_child = 0; current_child < merge_children.size(); current_child++)
+        {
+            if (merge_children[current_child] == inverse_child)
+                continue;
+            start_merge_child(merge_children[current_child], merge_children[current_child]);
+resume_2:
+            while (!cb())
+            {
+                state = 2;
+                return;
+            }
+            cb = NULL;
+            parent->change_parent(merge_children[current_child], new_parent);
+            state = 3;
+resume_3:
+            if (parent->waiting > 0)
+                return;
+        }
+        // Merge our "inverse" child into our "inverse" parent
+        if (inverse_child != 0)
+        {
+            start_merge_child(inverse_child, inverse_parent);
+resume_4:
+            while (!cb())
+            {
+                state = 4;
+                return;
+            }
+            cb = NULL;
+            // Delete "inverse" child data
+            start_delete_source(inverse_child);
+resume_5:
+            while (!cb())
+            {
+                state = 5;
+                return;
+            }
+            cb = NULL;
+            // Delete "inverse" child metadata, rename parent over it,
+            // and also change parent links of the previous "inverse" child
+            rename_inverse_parent();
+            state = 6;
+resume_6:
+            if (parent->waiting > 0)
+                return;
+        }
+        // Delete parents, except the "inverse" one
+        for (current_child = 0; current_child < chain_list.size(); current_child++)
+        {
+            if (chain_list[current_child] == inverse_parent)
+                continue;
+            start_delete_source(chain_list[current_child]);
+resume_7:
+            while (!cb())
+            {
+                state = 7;
+                return;
+            }
+            cb = NULL;
+            delete_inode_config(chain_list[current_child]);
+            state = 8;
+resume_8:
+            if (parent->waiting > 0)
+                return;
+        }
+        state = 9;
+resume_9:
+        // Done
+        return;
+    }
+
+    void get_merge_children()
+    {
+        // Get all children of from..to
+        inode_config_t *from_cfg = parent->get_inode_cfg(from_name);
+        inode_config_t *to_cfg = parent->get_inode_cfg(to_name);
+        // Check that to_cfg is actually a child of from_cfg
+        // FIXME de-copypaste the following piece of code with snap_merger_t
+        inode_config_t *cur = to_cfg;
+        chain_list.push_back(cur->num);
+        while (cur->num != from_cfg->num && cur->parent_id != 0)
+        {
+            auto it = parent->cli->st_cli.inode_config.find(cur->parent_id);
+            if (it == parent->cli->st_cli.inode_config.end())
+            {
+                fprintf(stderr, "Parent inode of layer %s (id %ld) not found\n", cur->name.c_str(), cur->parent_id);
+                exit(1);
+            }
+            cur = &it->second;
+            chain_list.push_back(cur->num);
+        }
+        if (cur->num != from_cfg->num)
+        {
+            fprintf(stderr, "Layer %s is not a child of %s\n", to_name.c_str(), from_name.c_str());
+            exit(1);
+        }
+        new_parent = from_cfg->parent_id;
+        // Calculate ranks
+        int i = chain_list.size()-1;
+        for (inode_t item: chain_list)
+        {
+            sources[item] = i--;
+        }
+        for (auto & ic: parent->cli->st_cli.inode_config)
+        {
+            if (!ic.second.parent_id)
+            {
+                continue;
+            }
+            auto it = sources.find(ic.second.parent_id);
+            if (it != sources.end() && sources.find(ic.second.num) == sources.end())
+            {
+                merge_children.push_back(ic.second.num);
+                if (ic.second.readonly || writers_stopped)
+                {
+                    inverse_candidates[ic.second.num] = it->second;
+                }
+            }
+        }
+    }
+
+    void read_stats()
+    {
+        if (inverse_candidates.size() == 0)
+        {
+            return;
+        }
+        json11::Json::array reads;
+        for (auto cp: inverse_candidates)
+        {
+            inode_t inode = cp.first;
+            reads.push_back(json11::Json::object {
+                { "request_range", json11::Json::object {
+                    { "key", base64_encode(
+                        parent->cli->st_cli.etcd_prefix+
+                        "/inode/stats/"+std::to_string(INODE_POOL(inode))+
+                        "/"+std::to_string(INODE_NO_POOL(inode))
+                    ) },
+                } }
+            });
+        }
+        for (auto cp: sources)
+        {
+            inode_t inode = cp.first;
+            reads.push_back(json11::Json::object {
+                { "request_range", json11::Json::object {
+                    { "key", base64_encode(
+                        parent->cli->st_cli.etcd_prefix+
+                        "/inode/stats/"+std::to_string(INODE_POOL(inode))+
+                        "/"+std::to_string(INODE_NO_POOL(inode))
+                    ) },
+                } }
+            });
+        }
+        parent->waiting++;
+        parent->cli->st_cli.etcd_txn(json11::Json::object {
+            { "success", reads },
+        }, ETCD_SLOW_TIMEOUT, [this](std::string err, json11::Json data)
+        {
+            parent->waiting--;
+            if (err != "")
+            {
+                fprintf(stderr, "Error reading layer statistics from etcd: %s\n", err.c_str());
+                exit(1);
+            }
+            for (auto inode_result: data["responses"].array_items())
+            {
+                auto kv = parent->cli->st_cli.parse_etcd_kv(inode_result["kvs"][0]);
+                pool_id_t pool_id = 0;
+                inode_t inode = 0;
+                char null_byte = 0;
+                sscanf(kv.key.c_str() + parent->cli->st_cli.etcd_prefix.length()+13, "%u/%lu%c", &pool_id, &inode, &null_byte);
+                if (!inode || null_byte != 0)
+                {
+                    fprintf(stderr, "Bad key returned from etcd: %s\n", kv.key.c_str());
+                    exit(1);
+                }
+                auto pool_cfg_it = parent->cli->st_cli.pool_config.find(pool_id);
+                if (pool_cfg_it == parent->cli->st_cli.pool_config.end())
+                {
+                    fprintf(stderr, "Pool %u does not exist\n", pool_id);
+                    exit(1);
+                }
+                inode = INODE_WITH_POOL(pool_id, inode);
+                auto & pool_cfg = pool_cfg_it->second;
+                uint64_t used_bytes = kv.value["raw_used"].uint64_value() / pool_cfg.pg_size;
+                if (pool_cfg.scheme != POOL_SCHEME_REPLICATED)
+                {
+                    used_bytes *= (pool_cfg.pg_size - pool_cfg.parity_chunks);
+                }
+                inode_used[inode] = used_bytes;
+            }
+            parent->ringloop->wakeup();
+        });
+    }
+
+    void choose_inverse_candidate()
+    {
+        uint64_t max_diff = 0;
+        for (auto cp: inverse_candidates)
+        {
+            inode_t child = cp.first;
+            uint64_t child_used = inode_used[child];
+            int rank = cp.second;
+            for (int i = chain_list.size()-rank; i < chain_list.size(); i++)
+            {
+                inode_t parent = chain_list[i];
+                uint64_t parent_used = inode_used[parent];
+                if (parent_used > child_used && (!max_diff || max_diff < (parent_used-child_used)))
+                {
+                    max_diff = (parent_used-child_used);
+                    inverse_parent = parent;
+                    inverse_child = child;
+                }
+            }
+        }
+    }
+
+    void rename_inverse_parent()
+    {
+        auto child_it = parent->cli->st_cli.inode_config.find(inverse_child);
+        if (child_it == parent->cli->st_cli.inode_config.end())
+        {
+            fprintf(stderr, "Inode %ld disappeared\n", inverse_child);
+            exit(1);
+        }
+        auto target_it = parent->cli->st_cli.inode_config.find(inverse_parent);
+        if (target_it == parent->cli->st_cli.inode_config.end())
+        {
+            fprintf(stderr, "Inode %ld disappeared\n", inverse_parent);
+            exit(1);
+        }
+        inode_config_t *child_cfg = &child_it->second;
+        inode_config_t *target_cfg = &target_it->second;
+        std::string child_name = child_cfg->name;
+        std::string target_name = target_cfg->name;
+        std::string child_cfg_key = base64_encode(
+            parent->cli->st_cli.etcd_prefix+
+            "/config/inode/"+std::to_string(INODE_POOL(inverse_child))+
+            "/"+std::to_string(INODE_NO_POOL(inverse_child))
+        );
+        std::string target_cfg_key = base64_encode(
+            parent->cli->st_cli.etcd_prefix+
+            "/config/inode/"+std::to_string(INODE_POOL(inverse_parent))+
+            "/"+std::to_string(INODE_NO_POOL(inverse_parent))
+        );
+        // Fill new configuration
+        inode_config_t new_cfg = *child_cfg;
+        new_cfg.num = target_cfg->num;
+        new_cfg.parent_id = new_parent;
+        json11::Json::array cmp = json11::Json::array {
+            json11::Json::object {
+                { "target", "MOD" },
+                { "key", child_cfg_key },
+                { "result", "LESS" },
+                { "mod_revision", child_cfg->mod_revision+1 },
+            },
+            json11::Json::object {
+                { "target", "MOD" },
+                { "key", target_cfg_key },
+                { "result", "LESS" },
+                { "mod_revision", target_cfg->mod_revision+1 },
+            },
+        };
+        json11::Json::array txn = json11::Json::array {
+            json11::Json::object {
+                { "request_delete_range", json11::Json::object {
+                    { "key", child_cfg_key },
+                } },
+            },
+            json11::Json::object {
+                { "request_put", json11::Json::object {
+                    { "key", target_cfg_key },
+                    { "value", base64_encode(json11::Json(parent->cli->st_cli.serialize_inode_cfg(&new_cfg)).dump()) },
+                } },
+            },
+            json11::Json::object {
+                { "request_put", json11::Json::object {
+                    { "key", base64_encode(parent->cli->st_cli.etcd_prefix+"/index/image/"+child_cfg->name) },
+                    { "value", base64_encode(json11::Json({
+                        { "id", INODE_NO_POOL(inverse_parent) },
+                        { "pool_id", (uint64_t)INODE_POOL(inverse_parent) },
+                    }).dump()) },
+                } },
+            },
+        };
+        // Reparent children of inverse_child
+        for (auto & cp: parent->cli->st_cli.inode_config)
+        {
+            if (cp.second.parent_id == child_cfg->num)
+            {
+                auto cp_cfg = cp.second;
+                cp_cfg.parent_id = inverse_parent;
+                auto cp_key = base64_encode(
+                    parent->cli->st_cli.etcd_prefix+
+                    "/config/inode/"+std::to_string(INODE_POOL(cp.second.num))+
+                    "/"+std::to_string(INODE_NO_POOL(cp.second.num))
+                );
+                cmp.push_back(json11::Json::object {
+                    { "target", "MOD" },
+                    { "key", cp_key },
+                    { "result", "LESS" },
+                    { "mod_revision", cp.second.mod_revision+1 },
+                });
+                txn.push_back(json11::Json::object {
+                    { "request_put", json11::Json::object {
+                        { "key", cp_key },
+                        { "value", base64_encode(json11::Json(parent->cli->st_cli.serialize_inode_cfg(&cp_cfg)).dump()) },
+                    } },
+                });
+            }
+        }
+        parent->waiting++;
+        parent->cli->st_cli.etcd_txn(json11::Json::object {
+            { "compare", cmp },
+            { "success", txn },
+        }, ETCD_SLOW_TIMEOUT, [this, target_name, child_name](std::string err, json11::Json res)
+        {
+            parent->waiting--;
+            if (err != "")
+            {
+                fprintf(stderr, "Error renaming %s to %s: %s\n", target_name.c_str(), child_name.c_str(), err.c_str());
+                exit(1);
+            }
+            if (!res["succeeded"].bool_value())
+            {
+                fprintf(
+                    stderr, "Parent (%s), child (%s), or one of its children"
+                    " configuration was modified during rename\n", target_name.c_str(), child_name.c_str()
+                );
+                exit(1);
+            }
+            printf("Layer %s renamed to %s\n", target_name.c_str(), child_name.c_str());
+            parent->ringloop->wakeup();
+        });
+    }
+
+    void delete_inode_config(inode_t cur)
+    {
+        auto cur_cfg_it = parent->cli->st_cli.inode_config.find(cur);
+        if (cur_cfg_it == parent->cli->st_cli.inode_config.end())
+        {
+            fprintf(stderr, "Inode 0x%lx disappeared\n", cur);
+            exit(1);
+        }
+        inode_config_t *cur_cfg = &cur_cfg_it->second;
+        std::string cur_name = cur_cfg->name;
+        std::string cur_cfg_key = base64_encode(
+            parent->cli->st_cli.etcd_prefix+
+            "/config/inode/"+std::to_string(INODE_POOL(cur))+
+            "/"+std::to_string(INODE_NO_POOL(cur))
+        );
+        parent->waiting++;
+        parent->cli->st_cli.etcd_txn(json11::Json::object {
+            { "compare", json11::Json::array {
+                json11::Json::object {
+                    { "target", "MOD" },
+                    { "key", cur_cfg_key },
+                    { "result", "LESS" },
+                    { "mod_revision", cur_cfg->mod_revision+1 },
+                },
+            } },
+            { "success", json11::Json::array {
+                json11::Json::object {
+                    { "request_delete_range", json11::Json::object {
+                        { "key", cur_cfg_key },
+                    } },
+                    { "request_delete_range", json11::Json::object {
+                        { "key", base64_encode(parent->cli->st_cli.etcd_prefix+"/index/image/"+cur_name) },
+                    } },
+                },
+            } },
+        }, ETCD_SLOW_TIMEOUT, [this, cur_name](std::string err, json11::Json res)
+        {
+            parent->waiting--;
+            if (err != "")
+            {
+                fprintf(stderr, "Error deleting %s: %s\n", cur_name.c_str(), err.c_str());
+                exit(1);
+            }
+            if (!res["succeeded"].bool_value())
+            {
+                fprintf(stderr, "Layer %s configuration was modified during deletion\n", cur_name.c_str());
+                exit(1);
+            }
+            printf("Layer %s deleted\n", cur_name.c_str());
+            parent->ringloop->wakeup();
+        });
+    }
+
+    void start_merge_child(inode_t child_inode, inode_t target_inode)
+    {
+        auto child_it = parent->cli->st_cli.inode_config.find(child_inode);
+        if (child_it == parent->cli->st_cli.inode_config.end())
+        {
+            fprintf(stderr, "Inode %ld disappeared\n", child_inode);
+            exit(1);
+        }
+        auto target_it = parent->cli->st_cli.inode_config.find(target_inode);
+        if (target_it == parent->cli->st_cli.inode_config.end())
+        {
+            fprintf(stderr, "Inode %ld disappeared\n", target_inode);
+            exit(1);
+        }
+        cb = parent->start_merge(json11::Json::object {
+            { "command", json11::Json::array{ "merge-data", from_name, child_it->second.name } },
+            { "target", target_it->second.name },
+            { "delete-source", false },
+            { "cas", use_cas },
+            { "fsync-interval", fsync_interval },
+        });
+    }
+
+    void start_delete_source(inode_t inode)
+    {
+        auto source = parent->cli->st_cli.inode_config.find(inode);
+        if (source == parent->cli->st_cli.inode_config.end())
+        {
+            fprintf(stderr, "Inode %ld disappeared\n", inode);
+            exit(1);
+        }
+        cb = parent->start_rm(json11::Json::object {
+            { "inode", inode },
+            { "pool", (uint64_t)INODE_POOL(inode) },
+            { "fsync-interval", fsync_interval },
+        });
+    }
+};
+
+std::function<bool(void)> cli_tool_t::start_snap_rm(json11::Json cfg)
+{
+    json11::Json::array cmd = cfg["command"].array_items();
+    auto snap_remover = new snap_remover_t();
+    snap_remover->parent = this;
+    snap_remover->from_name = cmd.size() > 1 ? cmd[1].string_value() : "";
+    snap_remover->to_name = cmd.size() > 2 ? cmd[2].string_value() : "";
+    if (snap_remover->from_name == "")
+    {
+        fprintf(stderr, "Layer to remove argument is missing\n");
+        exit(1);
+    }
+    if (snap_remover->to_name == "")
+    {
+        snap_remover->to_name = snap_remover->from_name;
+    }
+    snap_remover->fsync_interval = cfg["fsync-interval"].uint64_value();
+    if (!snap_remover->fsync_interval)
+        snap_remover->fsync_interval = 128;
+    if (!cfg["cas"].is_null())
+        snap_remover->use_cas = cfg["cas"].uint64_value() ? 2 : 0;
+    if (!cfg["writers_stopped"].is_null())
+        snap_remover->writers_stopped = true;
+    return [snap_remover]()
+    {
+        snap_remover->loop();
+        if (snap_remover->is_done())
+        {
+            delete snap_remover;
+            return true;
+        }
+        return false;
+    };
+}

src/cluster_client.cpp

@@ -12,7 +12,7 @@
 #define CACHE_DIRTY 1
 #define CACHE_FLUSHING 2
 #define CACHE_REPEATING 3
-#define OP_FLUSH_BUFFER 2
+#define OP_FLUSH_BUFFER 0x02
 
 cluster_client_t::cluster_client_t(ring_loop_t *ringloop, timerfd_manager_t *tfd, json11::Json & config)
 {
@@ -31,6 +31,7 @@ cluster_client_t::cluster_client_t(ring_loop_t *ringloop, timerfd_manager_t *tfd
         {
             // peer_osd just connected
             continue_ops();
+            continue_lists();
         }
         else if (dirty_buffers.size())
         {
@@ -51,7 +52,7 @@ cluster_client_t::cluster_client_t(ring_loop_t *ringloop, timerfd_manager_t *tfd
     msgr.exec_op = [this](osd_op_t *op)
     {
         // Garbage in
-        printf("Incoming garbage from peer %d\n", op->peer_fd);
+        fprintf(stderr, "Incoming garbage from peer %d\n", op->peer_fd);
         msgr.stop_client(op->peer_fd);
         delete op;
     };
@@ -139,7 +140,7 @@ void cluster_client_t::calc_wait(cluster_op_t *op)
         if (!op->prev_wait && pgs_loaded)
             continue_sync(op);
     }
-    else
+    else /* if (op->opcode == OSD_OP_READ || op->opcode == OSD_OP_READ_BITMAP) */
     {
         for (auto prev = op->prev; prev; prev = prev->prev)
         {
@@ -147,7 +148,7 @@ void cluster_client_t::calc_wait(cluster_op_t *op)
             {
                 op->prev_wait++;
             }
-            else if (prev->opcode == OSD_OP_WRITE || prev->opcode == OSD_OP_READ)
+            else if (prev->opcode == OSD_OP_WRITE || prev->opcode == OSD_OP_READ || prev->opcode == OSD_OP_READ_BITMAP)
             {
                 // Flushes are always in the beginning
                 break;
@@ -167,7 +168,7 @@ void cluster_client_t::inc_wait(uint64_t opcode, uint64_t flags, cluster_op_t *n
             auto n2 = next->next;
             if (next->opcode == OSD_OP_SYNC ||
                 next->opcode == OSD_OP_WRITE && (flags & OP_FLUSH_BUFFER) && !(next->flags & OP_FLUSH_BUFFER) ||
-                next->opcode == OSD_OP_READ && (flags & OP_FLUSH_BUFFER))
+                (next->opcode == OSD_OP_READ || next->opcode == OSD_OP_READ_BITMAP) && (flags & OP_FLUSH_BUFFER))
             {
                 next->prev_wait += inc;
                 if (!next->prev_wait)
@@ -357,7 +358,7 @@ void cluster_client_t::on_change_hook(std::map<std::string, etcd_kv_t> & changes
             // And now they have to be resliced!
             for (auto op = op_queue_head; op; op = op->next)
             {
-                if ((op->opcode == OSD_OP_WRITE || op->opcode == OSD_OP_READ) &&
+                if ((op->opcode == OSD_OP_WRITE || op->opcode == OSD_OP_READ || op->opcode == OSD_OP_READ_BITMAP) &&
                     INODE_POOL(op->cur_inode) == pool_item.first)
                 {
                     op->needs_reslice = true;
@@ -417,7 +418,8 @@ void cluster_client_t::on_ready(std::function<void(void)> fn)
  */
 void cluster_client_t::execute(cluster_op_t *op)
 {
-    if (op->opcode != OSD_OP_SYNC && op->opcode != OSD_OP_READ && op->opcode != OSD_OP_WRITE)
+    if (op->opcode != OSD_OP_SYNC && op->opcode != OSD_OP_READ &&
+        op->opcode != OSD_OP_READ_BITMAP && op->opcode != OSD_OP_WRITE)
     {
         op->retval = -EINVAL;
         std::function<void(cluster_op_t*)>(op->callback)(op);
@@ -557,7 +559,7 @@ void cluster_client_t::flush_buffer(const object_id & oid, cluster_buffer_t *wr)
 {
     wr->state = CACHE_REPEATING;
     cluster_op_t *op = new cluster_op_t;
-    op->flags = OP_FLUSH_BUFFER;
+    op->flags = OSD_OP_IGNORE_READONLY|OP_FLUSH_BUFFER;
     op->opcode = OSD_OP_WRITE;
     op->cur_inode = op->inode = oid.inode;
     op->offset = oid.stripe;
@@ -594,7 +596,8 @@ int cluster_client_t::continue_rw(cluster_op_t *op)
     else if (op->state == 3)
         goto resume_3;
 resume_0:
-    if (!op->len || op->offset % bs_bitmap_granularity || op->len % bs_bitmap_granularity)
+    if ((op->opcode == OSD_OP_READ || op->opcode == OSD_OP_WRITE) && !op->len ||
+        op->offset % bs_bitmap_granularity || op->len % bs_bitmap_granularity)
     {
         op->retval = -EINVAL;
         erase_op(op);
@@ -615,16 +618,19 @@ resume_0:
             return 0;
         }
     }
-    if (op->opcode == OSD_OP_WRITE)
+    if (op->opcode == OSD_OP_WRITE || op->opcode == OSD_OP_DELETE)
     {
-        auto ino_it = st_cli.inode_config.find(op->inode);
-        if (ino_it != st_cli.inode_config.end() && ino_it->second.readonly)
+        if (!(op->flags & OSD_OP_IGNORE_READONLY))
         {
-            op->retval = -EINVAL;
-            erase_op(op);
-            return 1;
+            auto ino_it = st_cli.inode_config.find(op->inode);
+            if (ino_it != st_cli.inode_config.end() && ino_it->second.readonly)
+            {
+                op->retval = -EINVAL;
+                erase_op(op);
+                return 1;
+            }
         }
-        if (!immediate_commit && !(op->flags & OP_FLUSH_BUFFER))
+        if (op->opcode == OSD_OP_WRITE && !immediate_commit && !(op->flags & OP_FLUSH_BUFFER))
         {
             copy_write(op, dirty_buffers);
         }
@@ -633,6 +639,13 @@ resume_1:
     // Slice the operation into parts
     slice_rw(op);
     op->needs_reslice = false;
+    if ((op->opcode == OSD_OP_WRITE || op->opcode == OSD_OP_DELETE) && op->version && op->parts.size() > 1)
+    {
+        // Atomic writes to multiple stripes are unsupported
+        op->retval = -EINVAL;
+        erase_op(op);
+        return 1;
+    }
 resume_2:
     // Send unsent parts, if they're not subject to change
     op->state = 3;
@@ -688,13 +701,16 @@ resume_3:
             // Check parent inode
             auto ino_it = st_cli.inode_config.find(op->cur_inode);
             while (ino_it != st_cli.inode_config.end() && ino_it->second.parent_id &&
-                INODE_POOL(ino_it->second.parent_id) == INODE_POOL(op->cur_inode))
+                INODE_POOL(ino_it->second.parent_id) == INODE_POOL(op->cur_inode) &&
+                // Check for loops
+                ino_it->second.parent_id != op->inode)
             {
                 // Skip parents from the same pool
                 ino_it = st_cli.inode_config.find(ino_it->second.parent_id);
             }
             if (ino_it != st_cli.inode_config.end() &&
-                ino_it->second.parent_id)
+                ino_it->second.parent_id &&
+                ino_it->second.parent_id != op->inode)
             {
                 // Continue reading from the parent inode
                 op->cur_inode = ino_it->second.parent_id;
@@ -783,13 +799,13 @@ void cluster_client_t::slice_rw(cluster_op_t *op)
     uint32_t pg_data_size = (pool_cfg.scheme == POOL_SCHEME_REPLICATED ? 1 : pool_cfg.pg_size-pool_cfg.parity_chunks);
     uint64_t pg_block_size = bs_block_size * pg_data_size;
     uint64_t first_stripe = (op->offset / pg_block_size) * pg_block_size;
-    uint64_t last_stripe = ((op->offset + op->len + pg_block_size - 1) / pg_block_size - 1) * pg_block_size;
+    uint64_t last_stripe = op->len > 0 ? ((op->offset + op->len - 1) / pg_block_size) * pg_block_size : first_stripe;
     op->retval = 0;
     op->parts.resize((last_stripe - first_stripe) / pg_block_size + 1);
-    if (op->opcode == OSD_OP_READ)
+    if (op->opcode == OSD_OP_READ || op->opcode == OSD_OP_READ_BITMAP)
     {
         // Allocate memory for the bitmap
-        unsigned object_bitmap_size = ((op->len / bs_bitmap_granularity + 7) / 8);
+        unsigned object_bitmap_size = (((op->opcode == OSD_OP_READ_BITMAP ? pg_block_size : op->len) / bs_bitmap_granularity + 7) / 8);
         object_bitmap_size = (object_bitmap_size < 8 ? 8 : object_bitmap_size);
         unsigned bitmap_mem = object_bitmap_size + (bs_bitmap_size * pg_data_size) * op->parts.size();
         if (op->bitmap_buf_size < bitmap_mem)
@@ -853,13 +869,13 @@ void cluster_client_t::slice_rw(cluster_op_t *op)
             if (end == begin)
                 op->done_count++;
         }
-        else
+        else if (op->opcode != OSD_OP_READ_BITMAP && op->opcode != OSD_OP_DELETE)
         {
             add_iov(end-begin, false, op, iov_idx, iov_pos, op->parts[i].iov, NULL, 0);
         }
         op->parts[i].parent = op;
         op->parts[i].offset = begin;
-        op->parts[i].len = (uint32_t)(end - begin);
+        op->parts[i].len = op->opcode == OSD_OP_READ_BITMAP || op->opcode == OSD_OP_DELETE ? 0 : (uint32_t)(end - begin);
         op->parts[i].pg_num = pg_num;
         op->parts[i].osd_num = 0;
         op->parts[i].flags = 0;
@@ -873,7 +889,7 @@ bool cluster_client_t::affects_osd(uint64_t inode, uint64_t offset, uint64_t len
     uint32_t pg_data_size = (pool_cfg.scheme == POOL_SCHEME_REPLICATED ? 1 : pool_cfg.pg_size-pool_cfg.parity_chunks);
     uint64_t pg_block_size = bs_block_size * pg_data_size;
     uint64_t first_stripe = (offset / pg_block_size) * pg_block_size;
-    uint64_t last_stripe = ((offset + len + pg_block_size - 1) / pg_block_size - 1) * pg_block_size;
+    uint64_t last_stripe = len > 0 ? ((offset + len - 1) / pg_block_size) * pg_block_size : first_stripe;
     for (uint64_t stripe = first_stripe; stripe <= last_stripe; stripe += pg_block_size)
     {
         pg_num_t pg_num = (stripe/pool_cfg.pg_stripe_size) % pool_cfg.real_pg_count + 1; // like map_to_pg()
@@ -906,9 +922,12 @@ bool cluster_client_t::try_send(cluster_op_t *op, int i)
                 pool_cfg.scheme == POOL_SCHEME_REPLICATED ? 1 : pool_cfg.pg_size-pool_cfg.parity_chunks
             );
             uint64_t meta_rev = 0;
-            auto ino_it = st_cli.inode_config.find(op->inode);
-            if (ino_it != st_cli.inode_config.end())
-                meta_rev = ino_it->second.mod_revision;
+            if (op->opcode != OSD_OP_READ_BITMAP && op->opcode != OSD_OP_DELETE)
+            {
+                auto ino_it = st_cli.inode_config.find(op->inode);
+                if (ino_it != st_cli.inode_config.end())
+                    meta_rev = ino_it->second.mod_revision;
+            }
             part->op = (osd_op_t){
                 .op_type = OSD_OP_OUT,
                 .peer_fd = peer_fd,
@@ -916,15 +935,16 @@ bool cluster_client_t::try_send(cluster_op_t *op, int i)
                     .header = {
                         .magic = SECONDARY_OSD_OP_MAGIC,
                         .id = op_id++,
-                        .opcode = op->opcode,
+                        .opcode = op->opcode == OSD_OP_READ_BITMAP ? OSD_OP_READ : op->opcode,
                     },
                     .inode = op->cur_inode,
                     .offset = part->offset,
                     .len = part->len,
                     .meta_revision = meta_rev,
+                    .version = op->opcode == OSD_OP_WRITE || op->opcode == OSD_OP_DELETE ? op->version : 0,
                 } },
-                .bitmap = op->opcode == OSD_OP_WRITE ? NULL : op->part_bitmaps + pg_bitmap_size*i,
-                .bitmap_len = (unsigned)(op->opcode == OSD_OP_WRITE ? 0 : pg_bitmap_size),
+                .bitmap = (op->opcode == OSD_OP_READ || op->opcode == OSD_OP_READ_BITMAP ? op->part_bitmaps + pg_bitmap_size*i : NULL),
+                .bitmap_len = (unsigned)(op->opcode == OSD_OP_READ || op->opcode == OSD_OP_READ_BITMAP ? pg_bitmap_size : 0),
                 .callback = [this, part](osd_op_t *op_part)
                 {
                     handle_op_part(part);
@@ -1072,10 +1092,6 @@ void cluster_client_t::handle_op_part(cluster_op_part_t *part)
     if (part->op.reply.hdr.retval != expected)
     {
         // Operation failed, retry
-        printf(
-            "%s operation failed on OSD %lu: retval=%ld (expected %d), dropping connection\n",
-            osd_op_names[part->op.req.hdr.opcode], part->osd_num, part->op.reply.hdr.retval, expected
-        );
         if (part->op.reply.hdr.retval == -EPIPE)
         {
             // Mark op->up_wait = true before stopping the client
@@ -1094,7 +1110,14 @@ void cluster_client_t::handle_op_part(cluster_op_part_t *part)
             // Don't overwrite other errors with -EPIPE
             op->retval = part->op.reply.hdr.retval;
         }
-        msgr.stop_client(part->op.peer_fd);
+        if (op->retval != -EINTR && op->retval != -EIO)
+        {
+            fprintf(
+                stderr, "%s operation failed on OSD %lu: retval=%ld (expected %d), dropping connection\n",
+                osd_op_names[part->op.req.hdr.opcode], part->osd_num, part->op.reply.hdr.retval, expected
+            );
+            msgr.stop_client(part->op.peer_fd);
+        }
         part->flags |= PART_ERROR;
     }
     else
@@ -1103,9 +1126,10 @@ void cluster_client_t::handle_op_part(cluster_op_part_t *part)
         dirty_osds.insert(part->osd_num);
         part->flags |= PART_DONE;
         op->done_count++;
-        if (op->opcode == OSD_OP_READ)
+        if (op->opcode == OSD_OP_READ || op->opcode == OSD_OP_READ_BITMAP)
         {
             copy_part_bitmap(op, part);
+            op->version = op->parts.size() == 1 ? part->op.reply.rw.version : 0;
         }
     }
     if (op->inflight_count == 0)
@@ -1126,7 +1150,7 @@ void cluster_client_t::copy_part_bitmap(cluster_op_t *op, cluster_op_part_t *par
     );
     uint32_t object_offset = (part->op.req.rw.offset - op->offset) / bs_bitmap_granularity;
     uint32_t part_offset = (part->op.req.rw.offset % pg_block_size) / bs_bitmap_granularity;
-    uint32_t part_len = part->op.req.rw.len / bs_bitmap_granularity;
+    uint32_t part_len = (op->opcode == OSD_OP_READ_BITMAP ? pg_block_size : part->op.req.rw.len) / bs_bitmap_granularity;
     if (!(object_offset & 0x7) && !(part_offset & 0x7) && (part_len >= 8))
     {
         // Copy bytes
@@ -1146,3 +1170,8 @@ void cluster_client_t::copy_part_bitmap(cluster_op_t *op, cluster_op_part_t *par
         part_len--;
     }
 }
+
+uint64_t cluster_client_t::next_op_id()
+{
+    return op_id++;
+}

src/cluster_client.h

@@ -10,6 +10,11 @@
 #define MAX_BLOCK_SIZE 128*1024*1024
 #define DEFAULT_CLIENT_MAX_DIRTY_BYTES 32*1024*1024
 #define DEFAULT_CLIENT_MAX_DIRTY_OPS 1024
+#define INODE_LIST_DONE 1
+#define INODE_LIST_HAS_UNSTABLE 2
+#define OSD_OP_READ_BITMAP OSD_OP_SEC_READ_BMP
+
+#define OSD_OP_IGNORE_READONLY 0x08
 
 struct cluster_op_t;
 
@@ -27,16 +32,22 @@ struct cluster_op_part_t
 
 struct cluster_op_t
 {
-    uint64_t opcode; // OSD_OP_READ, OSD_OP_WRITE, OSD_OP_SYNC
+    uint64_t opcode; // OSD_OP_READ, OSD_OP_WRITE, OSD_OP_SYNC, OSD_OP_DELETE, OSD_OP_READ_BITMAP
     uint64_t inode;
     uint64_t offset;
     uint64_t len;
+    // for reads and writes within a single object (stripe),
+    // reads can return current version and writes can use "CAS" semantics
+    uint64_t version = 0;
+    // now only OSD_OP_IGNORE_READONLY is supported
+    uint64_t flags = 0;
     int retval;
     osd_op_buf_list_t iov;
+    // READ and READ_BITMAP return the bitmap here
+    void *bitmap_buf = NULL;
     std::function<void(cluster_op_t*)> callback;
     ~cluster_op_t();
 protected:
-    uint64_t flags = 0;
     int state = 0;
     uint64_t cur_inode; // for snapshot reads
     void *buf = NULL;
@@ -45,7 +56,7 @@ protected:
     bool up_wait = false;
     int inflight_count = 0, done_count = 0;
     std::vector<cluster_op_part_t> parts;
-    void *bitmap_buf = NULL, *part_bitmaps = NULL;
+    void *part_bitmaps = NULL;
     unsigned bitmap_buf_size = 0;
     cluster_op_t *prev = NULL, *next = NULL;
     int prev_wait = 0;
@@ -59,6 +70,9 @@ struct cluster_buffer_t
     int state;
 };
 
+struct inode_list_t;
+struct inode_list_osd_t;
+
 // FIXME: Split into public and private interfaces
 class cluster_client_t
 {
@@ -90,6 +104,7 @@ class cluster_client_t
     bool pgs_loaded = false;
     ring_consumer_t consumer;
     std::vector<std::function<void(void)>> on_ready_hooks;
+    std::vector<inode_list_t*> lists;
     int continuing_ops = 0;
 
 public:
@@ -105,6 +120,14 @@ public:
 
     static void copy_write(cluster_op_t *op, std::map<object_id, cluster_buffer_t> & dirty_buffers);
     void continue_ops(bool up_retry = false);
+    inode_list_t *list_inode_start(inode_t inode,
+        std::function<void(inode_list_t* lst, std::set<object_id>&& objects, pg_num_t pg_num, osd_num_t primary_osd, int status)> callback);
+    int list_pg_count(inode_list_t *lst);
+    void list_inode_next(inode_list_t *lst, int next_pgs);
+    inline uint32_t get_bs_bitmap_granularity() { return bs_bitmap_granularity; }
+    inline uint64_t get_bs_block_size() { return bs_block_size; }
+    uint64_t next_op_id();
+
 protected:
     bool affects_osd(uint64_t inode, uint64_t offset, uint64_t len, osd_num_t osd);
     void flush_buffer(const object_id & oid, cluster_buffer_t *wr);
@@ -122,4 +145,7 @@ protected:
     void erase_op(cluster_op_t *op);
     void calc_wait(cluster_op_t *op);
     void inc_wait(uint64_t opcode, uint64_t flags, cluster_op_t *next, int inc);
+    void continue_lists();
+    void continue_listing(inode_list_t *lst);
+    void send_list(inode_list_osd_t *cur_list);
 };

src/cluster_client_list.cpp

@@ -0,0 +1,285 @@
+// Copyright (c) Vitaliy Filippov, 2019+
+// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
+
+#include <algorithm>
+#include "pg_states.h"
+#include "cluster_client.h"
+
+struct inode_list_t;
+
+struct inode_list_pg_t;
+
+struct inode_list_osd_t
+{
+    inode_list_pg_t *pg = NULL;
+    osd_num_t osd_num = 0;
+    bool sent = false;
+};
+
+struct inode_list_pg_t
+{
+    inode_list_t *lst = NULL;
+    int pos = 0;
+    pg_num_t pg_num;
+    osd_num_t cur_primary;
+    bool has_unstable = false;
+    int sent = 0;
+    int done = 0;
+    std::vector<inode_list_osd_t> list_osds;
+    std::set<object_id> objects;
+};
+
+struct inode_list_t
+{
+    cluster_client_t *cli = NULL;
+    pool_id_t pool_id = 0;
+    inode_t inode = 0;
+    int done_pgs = 0;
+    int want = 0;
+    std::vector<inode_list_pg_t*> pgs;
+    std::function<void(inode_list_t* lst, std::set<object_id>&& objects, pg_num_t pg_num, osd_num_t primary_osd, int status)> callback;
+};
+
+inode_list_t* cluster_client_t::list_inode_start(inode_t inode,
+    std::function<void(inode_list_t* lst, std::set<object_id>&& objects, pg_num_t pg_num, osd_num_t primary_osd, int status)> callback)
+{
+    int skipped_pgs = 0;
+    pool_id_t pool_id = INODE_POOL(inode);
+    if (!pool_id || st_cli.pool_config.find(pool_id) == st_cli.pool_config.end())
+    {
+        if (log_level > 0)
+        {
+            fprintf(stderr, "Pool %u does not exist\n", pool_id);
+        }
+        return NULL;
+    }
+    inode_list_t *lst = new inode_list_t();
+    lst->cli = this;
+    lst->pool_id = pool_id;
+    lst->inode = inode;
+    lst->callback = callback;
+    auto pool_cfg = st_cli.pool_config[pool_id];
+    for (auto & pg_item: pool_cfg.pg_config)
+    {
+        auto & pg = pg_item.second;
+        if (pg.pause || !pg.cur_primary || !(pg.cur_state & PG_ACTIVE))
+        {
+            skipped_pgs++;
+            if (log_level > 0)
+            {
+                fprintf(stderr, "PG %u is inactive, skipping\n", pg_item.first);
+            }
+            continue;
+        }
+        inode_list_pg_t *r = new inode_list_pg_t();
+        r->lst = lst;
+        r->pg_num = pg_item.first;
+        r->cur_primary = pg.cur_primary;
+        if (pg.cur_state != PG_ACTIVE)
+        {
+            // Not clean
+            std::set<osd_num_t> all_peers;
+            for (osd_num_t pg_osd: pg.target_set)
+            {
+                if (pg_osd != 0)
+                {
+                    all_peers.insert(pg_osd);
+                }
+            }
+            for (osd_num_t pg_osd: pg.all_peers)
+            {
+                if (pg_osd != 0)
+                {
+                    all_peers.insert(pg_osd);
+                }
+            }
+            for (auto & hist_item: pg.target_history)
+            {
+                for (auto pg_osd: hist_item)
+                {
+                    if (pg_osd != 0)
+                    {
+                        all_peers.insert(pg_osd);
+                    }
+                }
+            }
+            for (osd_num_t peer_osd: all_peers)
+            {
+                r->list_osds.push_back((inode_list_osd_t){
+                    .pg = r,
+                    .osd_num = peer_osd,
+                    .sent = false,
+                });
+            }
+        }
+        else
+        {
+            // Clean
+            r->list_osds.push_back((inode_list_osd_t){
+                .pg = r,
+                .osd_num = pg.cur_primary,
+                .sent = false,
+            });
+        }
+        lst->pgs.push_back(r);
+    }
+    std::sort(lst->pgs.begin(), lst->pgs.end(), [](inode_list_pg_t *a, inode_list_pg_t *b)
+    {
+        return a->cur_primary < b->cur_primary ? true : false;
+    });
+    for (int i = 0; i < lst->pgs.size(); i++)
+    {
+        lst->pgs[i]->pos = i;
+    }
+    lists.push_back(lst);
+    return lst;
+}
+
+int cluster_client_t::list_pg_count(inode_list_t *lst)
+{
+    return lst->pgs.size();
+}
+
+void cluster_client_t::list_inode_next(inode_list_t *lst, int next_pgs)
+{
+    if (next_pgs >= 0)
+    {
+        lst->want += next_pgs;
+    }
+    continue_listing(lst);
+}
+
+void cluster_client_t::continue_listing(inode_list_t *lst)
+{
+    if (lst->done_pgs >= lst->pgs.size())
+    {
+        // All done
+        for (int i = 0; i < lists.size(); i++)
+        {
+            if (lists[i] == lst)
+            {
+                lists.erase(lists.begin()+i, lists.begin()+i+1);
+                break;
+            }
+        }
+        delete lst;
+        return;
+    }
+    if (lst->want <= 0)
+    {
+        return;
+    }
+    for (int i = 0; i < lst->pgs.size(); i++)
+    {
+        if (lst->pgs[i] && lst->pgs[i]->sent < lst->pgs[i]->list_osds.size())
+        {
+            for (int j = 0; j < lst->pgs[i]->list_osds.size(); j++)
+            {
+                send_list(&lst->pgs[i]->list_osds[j]);
+                if (lst->want <= 0)
+                {
+                    break;
+                }
+            }
+        }
+    }
+}
+
+void cluster_client_t::send_list(inode_list_osd_t *cur_list)
+{
+    if (cur_list->sent)
+    {
+        return;
+    }
+    if (msgr.osd_peer_fds.find(cur_list->osd_num) == msgr.osd_peer_fds.end())
+    {
+        // Initiate connection
+        msgr.connect_peer(cur_list->osd_num, st_cli.peer_states[cur_list->osd_num]);
+        return;
+    }
+    auto & pool_cfg = st_cli.pool_config[cur_list->pg->lst->pool_id];
+    osd_op_t *op = new osd_op_t();
+    op->op_type = OSD_OP_OUT;
+    op->peer_fd = msgr.osd_peer_fds[cur_list->osd_num];
+    op->req = (osd_any_op_t){
+        .sec_list = {
+            .header = {
+                .magic = SECONDARY_OSD_OP_MAGIC,
+                .id = op_id++,
+                .opcode = OSD_OP_SEC_LIST,
+            },
+            .list_pg = cur_list->pg->pg_num,
+            .pg_count = (pg_num_t)pool_cfg.real_pg_count,
+            .pg_stripe_size = pool_cfg.pg_stripe_size,
+            .min_inode = cur_list->pg->lst->inode,
+            .max_inode = cur_list->pg->lst->inode,
+        },
+    };
+    op->callback = [this, cur_list](osd_op_t *op)
+    {
+        if (op->reply.hdr.retval < 0)
+        {
+            fprintf(stderr, "Failed to get PG %u/%u object list from OSD %lu (retval=%ld), skipping\n",
+                cur_list->pg->lst->pool_id, cur_list->pg->pg_num, cur_list->osd_num, op->reply.hdr.retval);
+        }
+        else
+        {
+            if (op->reply.sec_list.stable_count < op->reply.hdr.retval)
+            {
+                // Unstable objects, if present, mean that someone still writes into the inode. Warn the user about it.
+                cur_list->pg->has_unstable = true;
+                fprintf(
+                    stderr, "[PG %u/%u] Inode still has %lu unstable object versions out of total %lu - is it still open?\n",
+                    cur_list->pg->lst->pool_id, cur_list->pg->pg_num, op->reply.hdr.retval - op->reply.sec_list.stable_count,
+                    op->reply.hdr.retval
+                );
+            }
+            if (log_level > 0)
+            {
+                fprintf(
+                    stderr, "[PG %u/%u] Got inode object list from OSD %lu: %ld object versions\n",
+                    cur_list->pg->lst->pool_id, cur_list->pg->pg_num, cur_list->osd_num, op->reply.hdr.retval
+                );
+            }
+            for (uint64_t i = 0; i < op->reply.hdr.retval; i++)
+            {
+                object_id oid = ((obj_ver_id*)op->buf)[i].oid;
+                oid.stripe = oid.stripe & ~STRIPE_MASK;
+                cur_list->pg->objects.insert(oid);
+            }
+        }
+        delete op;
+        auto lst = cur_list->pg->lst;
+        auto pg = cur_list->pg;
+        pg->done++;
+        if (pg->done >= pg->list_osds.size())
+        {
+            int status = 0;
+            lst->done_pgs++;
+            if (lst->done_pgs >= lst->pgs.size())
+            {
+                status |= INODE_LIST_DONE;
+            }
+            if (pg->has_unstable)
+            {
+                status |= INODE_LIST_HAS_UNSTABLE;
+            }
+            lst->callback(lst, std::move(pg->objects), pg->pg_num, pg->cur_primary, status);
+            lst->pgs[pg->pos] = NULL;
+            delete pg;
+        }
+        continue_listing(lst);
+    };
+    msgr.outbox_push(op);
+    cur_list->sent = true;
+    cur_list->pg->sent++;
+    cur_list->pg->lst->want--;
+}
+
+void cluster_client_t::continue_lists()
+{
+    for (auto lst: lists)
+    {
+        continue_listing(lst);
+    }
+}

src/etcd_state_client.cpp

@@ -35,7 +35,7 @@ etcd_kv_t etcd_state_client_t::parse_etcd_kv(const json11::Json & kv_json)
     kv.value = json_text == "" ? json11::Json() : json11::Json::parse(json_text, json_err);
     if (json_err != "")
     {
-        printf("Bad JSON in etcd key %s: %s (value: %s)\n", kv.key.c_str(), json_err.c_str(), json_text.c_str());
+        fprintf(stderr, "Bad JSON in etcd key %s: %s (value: %s)\n", kv.key.c_str(), json_err.c_str(), json_text.c_str());
         kv.key = "";
     }
     else
@@ -81,10 +81,10 @@ void etcd_state_client_t::add_etcd_url(std::string addr)
             addr = addr.substr(7);
         else if (strtolower(addr.substr(0, 8)) == "https://")
         {
-            printf("HTTPS is unsupported for etcd. Either use plain HTTP or setup a local proxy for etcd interaction\n");
+            fprintf(stderr, "HTTPS is unsupported for etcd. Either use plain HTTP or setup a local proxy for etcd interaction\n");
             exit(1);
         }
-        if (addr.find('/') < 0)
+        if (addr.find('/') == std::string::npos)
             addr += "/v3";
         this->etcd_addresses.push_back(addr);
     }
@@ -149,7 +149,7 @@ void etcd_state_client_t::start_etcd_watcher()
             json11::Json data = json11::Json::parse(msg->body, json_err);
             if (json_err != "")
             {
-                printf("Bad JSON in etcd event: %s, ignoring event\n", json_err.c_str());
+                fprintf(stderr, "Bad JSON in etcd event: %s, ignoring event\n", json_err.c_str());
             }
             else
             {
@@ -175,7 +175,7 @@ void etcd_state_client_t::start_etcd_watcher()
                 {
                     if (this->log_level > 3)
                     {
-                        printf("Incoming event: %s -> %s\n", kv.first.c_str(), kv.second.value.dump().c_str());
+                        fprintf(stderr, "Incoming event: %s -> %s\n", kv.first.c_str(), kv.second.value.dump().c_str());
                     }
                     parse_state(kv.second);
                 }
@@ -250,7 +250,7 @@ void etcd_state_client_t::load_global_config()
     {
         if (err != "")
         {
-            printf("Error reading OSD configuration from etcd: %s\n", err.c_str());
+            fprintf(stderr, "Error reading OSD configuration from etcd: %s\n", err.c_str());
             tfd->set_timer(ETCD_SLOW_TIMEOUT, false, [this](int timer_id)
             {
                 load_global_config();
@@ -323,7 +323,7 @@ void etcd_state_client_t::load_pgs()
     {
         if (err != "")
         {
-            printf("Error loading PGs from etcd: %s\n", err.c_str());
+            fprintf(stderr, "Error loading PGs from etcd: %s\n", err.c_str());
             tfd->set_timer(ETCD_SLOW_TIMEOUT, false, [this](int timer_id)
             {
                 load_pgs();
@@ -386,7 +386,7 @@ void etcd_state_client_t::parse_state(const etcd_kv_t & kv)
             sscanf(pool_item.first.c_str(), "%u%c", &pool_id, &null_byte);
             if (!pool_id || pool_id >= POOL_ID_MAX || null_byte != 0)
             {
-                printf("Pool ID %s is invalid (must be a number less than 0x%x), skipping pool\n", pool_item.first.c_str(), POOL_ID_MAX);
+                fprintf(stderr, "Pool ID %s is invalid (must be a number less than 0x%x), skipping pool\n", pool_item.first.c_str(), POOL_ID_MAX);
                 continue;
             }
             pc.id = pool_id;
@@ -394,7 +394,7 @@ void etcd_state_client_t::parse_state(const etcd_kv_t & kv)
             pc.name = pool_item.second["name"].string_value();
             if (pc.name == "")
             {
-                printf("Pool %u has empty name, skipping pool\n", pool_id);
+                fprintf(stderr, "Pool %u has empty name, skipping pool\n", pool_id);
                 continue;
             }
             // Failure Domain
@@ -408,7 +408,7 @@ void etcd_state_client_t::parse_state(const etcd_kv_t & kv)
                 pc.scheme = POOL_SCHEME_JERASURE;
             else
             {
-                printf("Pool %u has invalid coding scheme (one of \"xor\", \"replicated\" or \"jerasure\" required), skipping pool\n", pool_id);
+                fprintf(stderr, "Pool %u has invalid coding scheme (one of \"xor\", \"replicated\" or \"jerasure\" required), skipping pool\n", pool_id);
                 continue;
             }
             // PG Size
@@ -418,7 +418,7 @@ void etcd_state_client_t::parse_state(const etcd_kv_t & kv)
                 (pc.scheme == POOL_SCHEME_XOR || pc.scheme == POOL_SCHEME_JERASURE) ||
                 pool_item.second["pg_size"].uint64_value() > 256)
             {
-                printf("Pool %u has invalid pg_size, skipping pool\n", pool_id);
+                fprintf(stderr, "Pool %u has invalid pg_size, skipping pool\n", pool_id);
                 continue;
             }
             // Parity Chunks
@@ -427,7 +427,7 @@ void etcd_state_client_t::parse_state(const etcd_kv_t & kv)
             {
                 if (pc.parity_chunks > 1)
                 {
-                    printf("Pool %u has invalid parity_chunks (must be 1), skipping pool\n", pool_id);
+                    fprintf(stderr, "Pool %u has invalid parity_chunks (must be 1), skipping pool\n", pool_id);
                     continue;
                 }
                 pc.parity_chunks = 1;
@@ -435,7 +435,7 @@ void etcd_state_client_t::parse_state(const etcd_kv_t & kv)
             if (pc.scheme == POOL_SCHEME_JERASURE &&
                 (pc.parity_chunks < 1 || pc.parity_chunks > pc.pg_size-2))
             {
-                printf("Pool %u has invalid parity_chunks (must be between 1 and pg_size-2), skipping pool\n", pool_id);
+                fprintf(stderr, "Pool %u has invalid parity_chunks (must be between 1 and pg_size-2), skipping pool\n", pool_id);
                 continue;
             }
             // PG MinSize
@@ -444,14 +444,14 @@ void etcd_state_client_t::parse_state(const etcd_kv_t & kv)
                 (pc.scheme == POOL_SCHEME_XOR || pc.scheme == POOL_SCHEME_JERASURE) &&
                 pc.pg_minsize < (pc.pg_size-pc.parity_chunks))
             {
-                printf("Pool %u has invalid pg_minsize, skipping pool\n", pool_id);
+                fprintf(stderr, "Pool %u has invalid pg_minsize, skipping pool\n", pool_id);
                 continue;
             }
             // PG Count
             pc.pg_count = pool_item.second["pg_count"].uint64_value();
             if (pc.pg_count < 1)
             {
-                printf("Pool %u has invalid pg_count, skipping pool\n", pool_id);
+                fprintf(stderr, "Pool %u has invalid pg_count, skipping pool\n", pool_id);
                 continue;
             }
             // Max OSD Combinations
@@ -460,7 +460,7 @@ void etcd_state_client_t::parse_state(const etcd_kv_t & kv)
                 pc.max_osd_combinations = 10000;
             if (pc.max_osd_combinations > 0 && pc.max_osd_combinations < 100)
             {
-                printf("Pool %u has invalid max_osd_combinations (must be at least 100), skipping pool\n", pool_id);
+                fprintf(stderr, "Pool %u has invalid max_osd_combinations (must be at least 100), skipping pool\n", pool_id);
                 continue;
             }
             // PG Stripe Size
@@ -478,7 +478,7 @@ void etcd_state_client_t::parse_state(const etcd_kv_t & kv)
             {
                 if (pg_item.second.target_set.size() != parsed_cfg.pg_size)
                 {
-                    printf("Pool %u PG %u configuration is invalid: osd_set size %lu != pool pg_size %lu\n",
+                    fprintf(stderr, "Pool %u PG %u configuration is invalid: osd_set size %lu != pool pg_size %lu\n",
                         pool_id, pg_item.first, pg_item.second.target_set.size(), parsed_cfg.pg_size);
                     pg_item.second.pause = true;
                 }
@@ -501,7 +501,7 @@ void etcd_state_client_t::parse_state(const etcd_kv_t & kv)
             sscanf(pool_item.first.c_str(), "%u%c", &pool_id, &null_byte);
             if (!pool_id || pool_id >= POOL_ID_MAX || null_byte != 0)
             {
-                printf("Pool ID %s is invalid in PG configuration (must be a number less than 0x%x), skipping pool\n", pool_item.first.c_str(), POOL_ID_MAX);
+                fprintf(stderr, "Pool ID %s is invalid in PG configuration (must be a number less than 0x%x), skipping pool\n", pool_item.first.c_str(), POOL_ID_MAX);
                 continue;
             }
             for (auto & pg_item: pool_item.second.object_items())
@@ -510,7 +510,7 @@ void etcd_state_client_t::parse_state(const etcd_kv_t & kv)
                 sscanf(pg_item.first.c_str(), "%u%c", &pg_num, &null_byte);
                 if (!pg_num || null_byte != 0)
                 {
-                    printf("Bad key in pool %u PG configuration: %s (must be a number), skipped\n", pool_id, pg_item.first.c_str());
+                    fprintf(stderr, "Bad key in pool %u PG configuration: %s (must be a number), skipped\n", pool_id, pg_item.first.c_str());
                     continue;
                 }
                 auto & parsed_cfg = this->pool_config[pool_id].pg_config[pg_num];
@@ -524,7 +524,7 @@ void etcd_state_client_t::parse_state(const etcd_kv_t & kv)
                 }
                 if (parsed_cfg.target_set.size() != pool_config[pool_id].pg_size)
                 {
-                    printf("Pool %u PG %u configuration is invalid: osd_set size %lu != pool pg_size %lu\n",
+                    fprintf(stderr, "Pool %u PG %u configuration is invalid: osd_set size %lu != pool pg_size %lu\n",
                         pool_id, pg_num, parsed_cfg.target_set.size(), pool_config[pool_id].pg_size);
                     parsed_cfg.pause = true;
                 }
@@ -537,8 +537,8 @@ void etcd_state_client_t::parse_state(const etcd_kv_t & kv)
             {
                 if (pg_it->second.exists && pg_it->first != ++n)
                 {
-                    printf(
-                        "Invalid pool %u PG configuration: PG numbers don't cover whole 1..%lu range\n",
+                    fprintf(
+                        stderr, "Invalid pool %u PG configuration: PG numbers don't cover whole 1..%lu range\n",
                         pool_item.second.id, pool_item.second.pg_config.size()
                     );
                     for (pg_it = pool_item.second.pg_config.begin(); pg_it != pool_item.second.pg_config.end(); pg_it++)
@@ -561,7 +561,7 @@ void etcd_state_client_t::parse_state(const etcd_kv_t & kv)
         sscanf(key.c_str() + etcd_prefix.length()+12, "%u/%u%c", &pool_id, &pg_num, &null_byte);
         if (!pool_id || pool_id >= POOL_ID_MAX || !pg_num || null_byte != 0)
         {
-            printf("Bad etcd key %s, ignoring\n", key.c_str());
+            fprintf(stderr, "Bad etcd key %s, ignoring\n", key.c_str());
         }
         else
         {
@@ -600,7 +600,7 @@ void etcd_state_client_t::parse_state(const etcd_kv_t & kv)
         sscanf(key.c_str() + etcd_prefix.length()+10, "%u/%u%c", &pool_id, &pg_num, &null_byte);
         if (!pool_id || pool_id >= POOL_ID_MAX || !pg_num || null_byte != 0)
         {
-            printf("Bad etcd key %s, ignoring\n", key.c_str());
+            fprintf(stderr, "Bad etcd key %s, ignoring\n", key.c_str());
         }
         else if (value.is_null())
         {
@@ -624,7 +624,7 @@ void etcd_state_client_t::parse_state(const etcd_kv_t & kv)
                 }
                 if (i >= pg_state_bit_count)
                 {
-                    printf("Unexpected pool %u PG %u state keyword in etcd: %s\n", pool_id, pg_num, e.dump().c_str());
+                    fprintf(stderr, "Unexpected pool %u PG %u state keyword in etcd: %s\n", pool_id, pg_num, e.dump().c_str());
                     return;
                 }
             }
@@ -633,7 +633,7 @@ void etcd_state_client_t::parse_state(const etcd_kv_t & kv)
                 (state & PG_PEERING) && state != PG_PEERING ||
                 (state & PG_INCOMPLETE) && state != PG_INCOMPLETE)
             {
-                printf("Unexpected pool %u PG %u state in etcd: primary=%lu, state=%s\n", pool_id, pg_num, cur_primary, value["state"].dump().c_str());
+                fprintf(stderr, "Unexpected pool %u PG %u state in etcd: primary=%lu, state=%s\n", pool_id, pg_num, cur_primary, value["state"].dump().c_str());
                 return;
             }
             this->pool_config[pool_id].pg_config[pg_num].cur_primary = cur_primary;
@@ -671,7 +671,7 @@ void etcd_state_client_t::parse_state(const etcd_kv_t & kv)
         sscanf(key.c_str() + etcd_prefix.length()+14, "%lu/%lu%c", &pool_id, &inode_num, &null_byte);
         if (!pool_id || pool_id >= POOL_ID_MAX || !inode_num || (inode_num >> (64-POOL_ID_BITS)) || null_byte != 0)
         {
-            printf("Bad etcd key %s, ignoring\n", key.c_str());
+            fprintf(stderr, "Bad etcd key %s, ignoring\n", key.c_str());
         }
         else
         {
@@ -706,8 +706,8 @@ void etcd_state_client_t::parse_state(const etcd_kv_t & kv)
                         parent_inode_num |= pool_id << (64-POOL_ID_BITS);
                     else if (parent_pool_id >= POOL_ID_MAX)
                     {
-                        printf(
-                            "Inode %lu/%lu parent_pool value is invalid, ignoring parent setting\n",
+                        fprintf(
+                            stderr, "Inode %lu/%lu parent_pool value is invalid, ignoring parent setting\n",
                             inode_num >> (64-POOL_ID_BITS), inode_num & ((1l << (64-POOL_ID_BITS)) - 1)
                         );
                         parent_inode_num = 0;
@@ -765,3 +765,22 @@ void etcd_state_client_t::close_watch(inode_watch_t* watch)
     }
     delete watch;
 }
+
+json11::Json::object & etcd_state_client_t::serialize_inode_cfg(inode_config_t *cfg)
+{
+    json11::Json::object new_cfg = json11::Json::object {
+        { "name", cfg->name },
+        { "size", cfg->size },
+    };
+    if (cfg->parent_id)
+    {
+        if (INODE_POOL(cfg->num) != INODE_POOL(cfg->parent_id))
+            new_cfg["parent_pool"] = (uint64_t)INODE_POOL(cfg->parent_id);
+        new_cfg["parent_id"] = (uint64_t)INODE_NO_POOL(cfg->parent_id);
+    }
+    if (cfg->readonly)
+    {
+        new_cfg["readonly"] = true;
+    }
+    return new_cfg;
+}

src/etcd_state_client.h

@@ -99,6 +99,7 @@ public:
     std::function<void(pool_id_t, pg_num_t)> on_change_pg_history_hook;
     std::function<void(osd_num_t)> on_change_osd_state_hook;
 
+    json11::Json::object & serialize_inode_cfg(inode_config_t *cfg);
     etcd_kv_t parse_etcd_kv(const json11::Json & kv_json);
     void etcd_call(std::string api, json11::Json payload, int timeout, std::function<void(std::string, json11::Json)> callback);
     void etcd_txn(json11::Json txn, int timeout, std::function<void(std::string, json11::Json)> callback);

src/fio_cluster.cpp

@@ -25,20 +25,17 @@
 
 #include <vector>
 
-#include "epoll_manager.h"
-#include "cluster_client.h"
+#include "vitastor_c.h"
 #include "fio_headers.h"
 
 struct sec_data
 {
-    ring_loop_t *ringloop = NULL;
-    epoll_manager_t *epmgr = NULL;
-    cluster_client_t *cli = NULL;
-    inode_watch_t *watch = NULL;
+    vitastor_c *cli = NULL;
+    void *watch = NULL;
     bool last_sync = false;
     /* The list of completed io_u structs. */
     std::vector<io_u*> completed;
-    uint64_t op_n = 0, inflight = 0;
+    uint64_t inflight = 0;
     bool trace = false;
 };
 
@@ -189,6 +186,12 @@ static struct fio_option options[] = {
     },
 };
 
+static void watch_callback(void *opaque, long watch)
+{
+    struct sec_data *bsd = (struct sec_data*)opaque;
+    bsd->watch = (void*)watch;
+}
+
 static int sec_setup(struct thread_data *td)
 {
     sec_options *o = (sec_options*)td->eo;
@@ -209,27 +212,6 @@ static int sec_setup(struct thread_data *td)
         td->o.open_files++;
     }
 
-    json11::Json::object cfg;
-    if (o->config_path)
-        cfg["config_path"] = std::string(o->config_path);
-    if (o->etcd_host)
-        cfg["etcd_address"] = std::string(o->etcd_host);
-    if (o->etcd_prefix)
-        cfg["etcd_prefix"] = std::string(o->etcd_prefix);
-    if (o->rdma_device)
-        cfg["rdma_device"] = std::string(o->rdma_device);
-    if (o->rdma_port_num)
-        cfg["rdma_port_num"] = o->rdma_port_num;
-    if (o->rdma_gid_index)
-        cfg["rdma_gid_index"] = o->rdma_gid_index;
-    if (o->rdma_mtu)
-        cfg["rdma_mtu"] = o->rdma_mtu;
-    if (o->cluster_log)
-        cfg["log_level"] = o->cluster_log;
-    if (o->use_rdma != -1)
-        cfg["use_rdma"] = o->use_rdma;
-    json11::Json cfg_json(cfg);
-
     if (!o->image)
     {
         if (!(o->inode & ((1l << (64-POOL_ID_BITS)) - 1)))
@@ -251,20 +233,20 @@ static int sec_setup(struct thread_data *td)
     {
         o->inode = 0;
     }
-    bsd->ringloop = new ring_loop_t(512);
-    bsd->epmgr = new epoll_manager_t(bsd->ringloop);
-    bsd->cli = new cluster_client_t(bsd->ringloop, bsd->epmgr->tfd, cfg_json);
+    bsd->cli = vitastor_c_create_uring(o->config_path, o->etcd_host, o->etcd_prefix,
+        o->use_rdma, o->rdma_device, o->rdma_port_num, o->rdma_gid_index, o->rdma_mtu, o->cluster_log);
     if (o->image)
     {
-        while (!bsd->cli->is_ready())
+        bsd->watch = NULL;
+        vitastor_c_watch_inode(bsd->cli, o->image, watch_callback, bsd);
+        while (true)
         {
-            bsd->ringloop->loop();
-            if (bsd->cli->is_ready())
+            vitastor_c_uring_handle_events(bsd->cli);
+            if (bsd->watch)
                 break;
-            bsd->ringloop->wait();
+            vitastor_c_uring_wait_events(bsd->cli);
         }
-        bsd->watch = bsd->cli->st_cli.watch_inode(std::string(o->image));
-        td->files[0]->real_file_size = bsd->watch->cfg.size;
+        td->files[0]->real_file_size = vitastor_c_inode_get_size(bsd->watch);
     }
 
     bsd->trace = o->trace ? true : false;
@@ -279,11 +261,9 @@ static void sec_cleanup(struct thread_data *td)
     {
         if (bsd->watch)
         {
-            bsd->cli->st_cli.close_watch(bsd->watch);
+            vitastor_c_close_watch(bsd->cli, bsd->watch);
         }
-        delete bsd->cli;
-        delete bsd->epmgr;
-        delete bsd->ringloop;
+        vitastor_c_destroy(bsd->cli);
         delete bsd;
     }
 }
@@ -294,12 +274,31 @@ static int sec_init(struct thread_data *td)
     return 0;
 }
 
+static void io_callback(void *opaque, long retval)
+{
+    struct io_u *io = (struct io_u*)opaque;
+    io->error = retval < 0 ? -retval : 0;
+    sec_data *bsd = (sec_data*)io->engine_data;
+    bsd->inflight--;
+    bsd->completed.push_back(io);
+    if (bsd->trace)
+    {
+        printf("--- %s 0x%lx retval=%ld\n", io->ddir == DDIR_READ ? "READ" :
+            (io->ddir == DDIR_WRITE ? "WRITE" : "SYNC"), (uint64_t)io, retval);
+    }
+}
+
+static void read_callback(void *opaque, long retval, uint64_t version)
+{
+    io_callback(opaque, retval);
+}
+
 /* Begin read or write request. */
 static enum fio_q_status sec_queue(struct thread_data *td, struct io_u *io)
 {
     sec_options *opt = (sec_options*)td->eo;
     sec_data *bsd = (sec_data*)td->io_ops_data;
-    int n = bsd->op_n;
+    struct iovec iov;
 
     fio_ro_check(td, io);
     if (io->ddir == DDIR_SYNC && bsd->last_sync)
@@ -308,32 +307,29 @@ static enum fio_q_status sec_queue(struct thread_data *td, struct io_u *io)
     }
 
     io->engine_data = bsd;
-    cluster_op_t *op = new cluster_op_t;
+    io->error = 0;
+    bsd->inflight++;
 
-    op->inode = opt->image ? bsd->watch->cfg.num : opt->inode;
+    uint64_t inode = opt->image ? vitastor_c_inode_get_num(bsd->watch) : opt->inode;
     switch (io->ddir)
     {
     case DDIR_READ:
-        op->opcode = OSD_OP_READ;
-        op->offset = io->offset;
-        op->len = io->xfer_buflen;
-        op->iov.push_back(io->xfer_buf, io->xfer_buflen);
+        iov = { .iov_base = io->xfer_buf, .iov_len = io->xfer_buflen };
+        vitastor_c_read(bsd->cli, inode, io->offset, io->xfer_buflen, &iov, 1, read_callback, io);
         bsd->last_sync = false;
         break;
     case DDIR_WRITE:
-        if (opt->image && bsd->watch->cfg.readonly)
+        if (opt->image && vitastor_c_inode_get_readonly(bsd->watch))
         {
             io->error = EROFS;
             return FIO_Q_COMPLETED;
         }
-        op->opcode = OSD_OP_WRITE;
-        op->offset = io->offset;
-        op->len = io->xfer_buflen;
-        op->iov.push_back(io->xfer_buf, io->xfer_buflen);
+        iov = { .iov_base = io->xfer_buf, .iov_len = io->xfer_buflen };
+        vitastor_c_write(bsd->cli, inode, io->offset, io->xfer_buflen, 0, &iov, 1, io_callback, io);
         bsd->last_sync = false;
         break;
     case DDIR_SYNC:
-        op->opcode = OSD_OP_SYNC;
+        vitastor_c_sync(bsd->cli, io_callback, io);
         bsd->last_sync = true;
         break;
     default:
@@ -341,39 +337,20 @@ static enum fio_q_status sec_queue(struct thread_data *td, struct io_u *io)
         return FIO_Q_COMPLETED;
     }
 
-    op->callback = [io, n](cluster_op_t *op)
-    {
-        io->error = op->retval < 0 ? -op->retval : 0;
-        sec_data *bsd = (sec_data*)io->engine_data;
-        bsd->inflight--;
-        bsd->completed.push_back(io);
-        if (bsd->trace)
-        {
-            printf("--- %s n=%d retval=%d\n", io->ddir == DDIR_READ ? "READ" :
-                (io->ddir == DDIR_WRITE ? "WRITE" : "SYNC"), n, op->retval);
-        }
-        delete op;
-    };
-
     if (opt->trace)
     {
         if (io->ddir == DDIR_SYNC)
         {
-            printf("+++ SYNC # %d\n", n);
+            printf("+++ SYNC 0x%lx\n", (uint64_t)io);
         }
         else
         {
-            printf("+++ %s # %d 0x%llx+%llx\n",
+            printf("+++ %s 0x%lx 0x%llx+%llx\n",
                 io->ddir == DDIR_READ ? "READ" : "WRITE",
-                n, io->offset, io->xfer_buflen);
+                (uint64_t)io, io->offset, io->xfer_buflen);
         }
     }
 
-    io->error = 0;
-    bsd->inflight++;
-    bsd->op_n++;
-    bsd->cli->execute(op);
-
     if (io->error != 0)
         return FIO_Q_COMPLETED;
     return FIO_Q_QUEUED;
@@ -384,10 +361,10 @@ static int sec_getevents(struct thread_data *td, unsigned int min, unsigned int
     sec_data *bsd = (sec_data*)td->io_ops_data;
     while (true)
     {
-        bsd->ringloop->loop();
+        vitastor_c_uring_handle_events(bsd->cli);
         if (bsd->completed.size() >= min)
             break;
-        bsd->ringloop->wait();
+        vitastor_c_uring_wait_events(bsd->cli);
     }
     return bsd->completed.size();
 }

src/messenger.cpp

@@ -21,13 +21,13 @@ void osd_messenger_t::init()
         );
         if (!rdma_context)
         {
-            printf("[OSD %lu] Couldn't initialize RDMA, proceeding with TCP only\n", osd_num);
+            fprintf(stderr, "[OSD %lu] Couldn't initialize RDMA, proceeding with TCP only\n", osd_num);
         }
         else
         {
             rdma_max_sge = rdma_max_sge < rdma_context->attrx.orig_attr.max_sge
                 ? rdma_max_sge : rdma_context->attrx.orig_attr.max_sge;
-            printf("[OSD %lu] RDMA initialized successfully\n", osd_num);
+            fprintf(stderr, "[OSD %lu] RDMA initialized successfully\n", osd_num);
             fcntl(rdma_context->channel->fd, F_SETFL, fcntl(rdma_context->channel->fd, F_GETFL, 0) | O_NONBLOCK);
             tfd->set_fd_handler(rdma_context->channel->fd, false, [this](int notify_fd, int epoll_events)
             {
@@ -55,7 +55,7 @@ void osd_messenger_t::init()
                 if (!cl->ping_time_remaining)
                 {
                     // Ping timed out, stop the client
-                    printf("Ping timed out for OSD %lu (client %d), disconnecting peer\n", cl->osd_num, cl->peer_fd);
+                    fprintf(stderr, "Ping timed out for OSD %lu (client %d), disconnecting peer\n", cl->osd_num, cl->peer_fd);
                     to_stop.push_back(cl->peer_fd);
                 }
             }
@@ -82,7 +82,7 @@ void osd_messenger_t::init()
                         delete op;
                         if (fail_fd >= 0)
                         {
-                            printf("Ping failed for OSD %lu (client %d), disconnecting peer\n", cl->osd_num, cl->peer_fd);
+                            fprintf(stderr, "Ping failed for OSD %lu (client %d), disconnecting peer\n", cl->osd_num, cl->peer_fd);
                             stop_client(fail_fd, true);
                         }
                     };
@@ -117,7 +117,7 @@ osd_messenger_t::~osd_messenger_t()
     }
     while (clients.size() > 0)
     {
-        stop_client(clients.begin()->first, true);
+        stop_client(clients.begin()->first, true, true);
     }
 #ifdef WITH_RDMA
     if (rdma_context)
@@ -261,7 +261,7 @@ void osd_messenger_t::try_connect_peer_addr(osd_num_t peer_osd, const char *peer
         {
             osd_num_t peer_osd = clients.at(peer_fd)->osd_num;
             stop_client(peer_fd, true);
-            on_connect_peer(peer_osd, -EIO);
+            on_connect_peer(peer_osd, -EPIPE);
             return;
         });
     }
@@ -305,7 +305,7 @@ void osd_messenger_t::handle_peer_epoll(int peer_fd, int epoll_events)
     if (epoll_events & EPOLLRDHUP)
     {
         // Stop client
-        printf("[OSD %lu] client %d disconnected\n", this->osd_num, peer_fd);
+        fprintf(stderr, "[OSD %lu] client %d disconnected\n", this->osd_num, peer_fd);
         stop_client(peer_fd, true);
     }
     else if (epoll_events & EPOLLIN)
@@ -330,7 +330,7 @@ void osd_messenger_t::on_connect_peer(osd_num_t peer_osd, int peer_fd)
     wp.connecting = false;
     if (peer_fd < 0)
     {
-        printf("Failed to connect to peer OSD %lu address %s port %d: %s\n", peer_osd, wp.cur_addr.c_str(), wp.cur_port, strerror(-peer_fd));
+        fprintf(stderr, "Failed to connect to peer OSD %lu address %s port %d: %s\n", peer_osd, wp.cur_addr.c_str(), wp.cur_port, strerror(-peer_fd));
         if (wp.address_changed)
         {
             wp.address_changed = false;
@@ -357,7 +357,7 @@ void osd_messenger_t::on_connect_peer(osd_num_t peer_osd, int peer_fd)
     }
     if (log_level > 0)
     {
-        printf("[OSD %lu] Connected with peer OSD %lu (client %d)\n", osd_num, peer_osd, peer_fd);
+        fprintf(stderr, "[OSD %lu] Connected with peer OSD %lu (client %d)\n", osd_num, peer_osd, peer_fd);
     }
     wanted_peers.erase(peer_osd);
     repeer_pgs(peer_osd);
@@ -403,7 +403,7 @@ void osd_messenger_t::check_peer_config(osd_client_t *cl)
         if (op->reply.hdr.retval < 0)
         {
             err = true;
-            printf("Failed to get config from OSD %lu (retval=%ld), disconnecting peer\n", cl->osd_num, op->reply.hdr.retval);
+            fprintf(stderr, "Failed to get config from OSD %lu (retval=%ld), disconnecting peer\n", cl->osd_num, op->reply.hdr.retval);
         }
         else
         {
@@ -411,18 +411,18 @@ void osd_messenger_t::check_peer_config(osd_client_t *cl)
             if (json_err != "")
             {
                 err = true;
-                printf("Failed to get config from OSD %lu: bad JSON: %s, disconnecting peer\n", cl->osd_num, json_err.c_str());
+                fprintf(stderr, "Failed to get config from OSD %lu: bad JSON: %s, disconnecting peer\n", cl->osd_num, json_err.c_str());
             }
             else if (config["osd_num"].uint64_value() != cl->osd_num)
             {
                 err = true;
-                printf("Connected to OSD %lu instead of OSD %lu, peer state is outdated, disconnecting peer\n", config["osd_num"].uint64_value(), cl->osd_num);
+                fprintf(stderr, "Connected to OSD %lu instead of OSD %lu, peer state is outdated, disconnecting peer\n", config["osd_num"].uint64_value(), cl->osd_num);
             }
             else if (config["protocol_version"].uint64_value() != OSD_PROTOCOL_VERSION)
             {
                 err = true;
-                printf(
-                    "OSD %lu protocol version is %lu, but only version %u is supported.\n"
+                fprintf(
+                    stderr, "OSD %lu protocol version is %lu, but only version %u is supported.\n"
                     " If you need to upgrade from 0.5.x please request it via the issue tracker.\n",
                     cl->osd_num, config["protocol_version"].uint64_value(), OSD_PROTOCOL_VERSION
                 );
@@ -443,8 +443,8 @@ void osd_messenger_t::check_peer_config(osd_client_t *cl)
             if (!msgr_rdma_address_t::from_string(config["rdma_address"].string_value().c_str(), &addr) ||
                 cl->rdma_conn->connect(&addr) != 0)
             {
-                printf(
-                    "Failed to connect to OSD %lu (address %s) using RDMA\n",
+                fprintf(
+                    stderr, "Failed to connect to OSD %lu (address %s) using RDMA\n",
                     cl->osd_num, config["rdma_address"].string_value().c_str()
                 );
                 delete cl->rdma_conn;
@@ -465,7 +465,7 @@ void osd_messenger_t::check_peer_config(osd_client_t *cl)
                 }
                 if (log_level > 0)
                 {
-                    printf("Connected to OSD %lu using RDMA\n", cl->osd_num);
+                    fprintf(stderr, "Connected to OSD %lu using RDMA\n", cl->osd_num);
                 }
                 cl->peer_state = PEER_RDMA;
                 tfd->set_fd_handler(cl->peer_fd, false, NULL);
@@ -491,7 +491,7 @@ void osd_messenger_t::accept_connections(int listen_fd)
     {
         assert(peer_fd != 0);
         char peer_str[256];
-        printf("[OSD %lu] new client %d: connection from %s port %d\n", this->osd_num, peer_fd,
+        fprintf(stderr, "[OSD %lu] new client %d: connection from %s port %d\n", this->osd_num, peer_fd,
             inet_ntop(AF_INET, &addr.sin_addr, peer_str, 256), ntohs(addr.sin_port));
         fcntl(peer_fd, F_SETFL, fcntl(peer_fd, F_GETFL, 0) | O_NONBLOCK);
         int one = 1;
@@ -516,10 +516,12 @@ void osd_messenger_t::accept_connections(int listen_fd)
     }
 }
 
+#ifdef WITH_RDMA
 bool osd_messenger_t::is_rdma_enabled()
 {
     return rdma_context != NULL;
 }
+#endif
 
 json11::Json osd_messenger_t::read_config(const json11::Json & config)
 {

src/messenger.h

@@ -156,7 +156,7 @@ public:
     void init();
     void parse_config(const json11::Json & config);
     void connect_peer(uint64_t osd_num, json11::Json peer_state);
-    void stop_client(int peer_fd, bool force = false);
+    void stop_client(int peer_fd, bool force = false, bool force_delete = false);
     void outbox_push(osd_op_t *cur_op);
     std::function<void(osd_op_t*)> exec_op;
     std::function<void(osd_num_t)> repeer_pgs;

src/mock/messenger.cpp

@@ -15,7 +15,7 @@ osd_messenger_t::~osd_messenger_t()
 {
     while (clients.size() > 0)
     {
-        stop_client(clients.begin()->first, true);
+        stop_client(clients.begin()->first, true, true);
     }
 }
 

src/msgr_op.h

@@ -76,7 +76,7 @@ struct osd_op_buf_list_t
                 buf = (iovec*)malloc(sizeof(iovec) * alloc);
                 if (!buf)
                 {
-                    printf("Failed to allocate %lu bytes\n", sizeof(iovec) * alloc);
+                    fprintf(stderr, "Failed to allocate %lu bytes\n", sizeof(iovec) * alloc);
                     exit(1);
                 }
                 memcpy(buf, inline_buf, sizeof(iovec) * old);
@@ -87,7 +87,7 @@ struct osd_op_buf_list_t
                 buf = (iovec*)realloc(buf, sizeof(iovec) * alloc);
                 if (!buf)
                 {
-                    printf("Failed to allocate %lu bytes\n", sizeof(iovec) * alloc);
+                    fprintf(stderr, "Failed to allocate %lu bytes\n", sizeof(iovec) * alloc);
                     exit(1);
                 }
             }
@@ -109,7 +109,7 @@ struct osd_op_buf_list_t
                 buf = (iovec*)malloc(sizeof(iovec) * alloc);
                 if (!buf)
                 {
-                    printf("Failed to allocate %lu bytes\n", sizeof(iovec) * alloc);
+                    fprintf(stderr, "Failed to allocate %lu bytes\n", sizeof(iovec) * alloc);
                     exit(1);
                 }
                 memcpy(buf, inline_buf, sizeof(iovec)*old);
@@ -120,7 +120,7 @@ struct osd_op_buf_list_t
                 buf = (iovec*)realloc(buf, sizeof(iovec) * alloc);
                 if (!buf)
                 {
-                    printf("Failed to allocate %lu bytes\n", sizeof(iovec) * alloc);
+                    fprintf(stderr, "Failed to allocate %lu bytes\n", sizeof(iovec) * alloc);
                     exit(1);
                 }
             }

src/msgr_rdma.cpp

@@ -46,6 +46,9 @@ msgr_rdma_connection_t::~msgr_rdma_connection_t()
     ctx->used_max_cqe -= max_send+max_recv;
     if (qp)
         ibv_destroy_qp(qp);
+    if (recv_buffers.size())
+        for (auto b: recv_buffers)
+            free(b);
 }
 
 msgr_rdma_context_t *msgr_rdma_context_t::create(const char *ib_devname, uint8_t ib_port, uint8_t gid_index, uint32_t mtu)
@@ -55,6 +58,7 @@ msgr_rdma_context_t *msgr_rdma_context_t::create(const char *ib_devname, uint8_t
     msgr_rdma_context_t *ctx = new msgr_rdma_context_t();
     ctx->mtu = mtu;
 
+    srand48(time(NULL));
     dev_list = ibv_get_device_list(NULL);
     if (!dev_list)
     {
@@ -315,8 +319,8 @@ bool osd_messenger_t::connect_rdma(int peer_fd, std::string rdma_address, uint64
             if (r != 0)
             {
                 delete rdma_conn;
-                printf(
-                    "Failed to connect RDMA queue pair to %s (client %d)\n",
+                fprintf(
+                    stderr, "Failed to connect RDMA queue pair to %s (client %d)\n",
                     addr.to_string().c_str(), peer_fd
                 );
             }
@@ -346,7 +350,7 @@ static void try_send_rdma_wr(osd_client_t *cl, ibv_sge *sge, int op_sge)
     int err = ibv_post_send(cl->rdma_conn->qp, &wr, &bad_wr);
     if (err || bad_wr)
     {
-        printf("RDMA send failed: %s\n", strerror(err));
+        fprintf(stderr, "RDMA send failed: %s\n", strerror(err));
         exit(1);
     }
     cl->rdma_conn->cur_send++;
@@ -408,7 +412,7 @@ static void try_recv_rdma_wr(osd_client_t *cl, ibv_sge *sge, int op_sge)
     int err = ibv_post_recv(cl->rdma_conn->qp, &wr, &bad_wr);
     if (err || bad_wr)
     {
-        printf("RDMA receive failed: %s\n", strerror(err));
+        fprintf(stderr, "RDMA receive failed: %s\n", strerror(err));
         exit(1);
     }
     cl->rdma_conn->cur_recv++;
@@ -445,7 +449,7 @@ void osd_messenger_t::handle_rdma_events()
     }
     if (ibv_req_notify_cq(rdma_context->cq, 0) != 0)
     {
-        printf("Failed to request RDMA completion notification, exiting\n");
+        fprintf(stderr, "Failed to request RDMA completion notification, exiting\n");
         exit(1);
     }
     ibv_wc wc[RDMA_EVENTS_AT_ONCE];
@@ -465,19 +469,23 @@ void osd_messenger_t::handle_rdma_events()
             osd_client_t *cl = cl_it->second;
             if (wc[i].status != IBV_WC_SUCCESS)
             {
-                printf("RDMA work request failed for client %d", client_id);
+                fprintf(stderr, "RDMA work request failed for client %d", client_id);
                 if (cl->osd_num)
                 {
-                    printf(" (OSD %lu)", cl->osd_num);
+                    fprintf(stderr, " (OSD %lu)", cl->osd_num);
                 }
-                printf(" with status: %s, stopping client\n", ibv_wc_status_str(wc[i].status));
+                fprintf(stderr, " with status: %s, stopping client\n", ibv_wc_status_str(wc[i].status));
                 stop_client(client_id);
                 continue;
             }
             if (!is_send)
             {
                 cl->rdma_conn->cur_recv--;
-                handle_read_buffer(cl, cl->rdma_conn->recv_buffers[0], wc[i].byte_len);
+                if (!handle_read_buffer(cl, cl->rdma_conn->recv_buffers[0], wc[i].byte_len))
+                {
+                    // handle_read_buffer may stop the client
+                    continue;
+                }
                 free(cl->rdma_conn->recv_buffers[0]);
                 cl->rdma_conn->recv_buffers.erase(cl->rdma_conn->recv_buffers.begin(), cl->rdma_conn->recv_buffers.begin()+1);
                 try_recv_rdma(cl);

src/msgr_receive.cpp

@@ -72,7 +72,7 @@ bool osd_messenger_t::handle_read(int result, osd_client_t *cl)
         // this is a client socket, so don't panic on error. just disconnect it
         if (result != 0)
         {
-            printf("Client %d socket read error: %d (%s). Disconnecting client\n", cl->peer_fd, -result, strerror(-result));
+            fprintf(stderr, "Client %d socket read error: %d (%s). Disconnecting client\n", cl->peer_fd, -result, strerror(-result));
         }
         stop_client(cl->peer_fd);
         return false;
@@ -177,7 +177,7 @@ bool osd_messenger_t::handle_finished_read(osd_client_t *cl)
             handle_op_hdr(cl);
         else
         {
-            printf("Received garbage: magic=%lx id=%lu opcode=%lx from %d\n", cl->read_op->req.hdr.magic, cl->read_op->req.hdr.id, cl->read_op->req.hdr.opcode, cl->peer_fd);
+            fprintf(stderr, "Received garbage: magic=%lx id=%lu opcode=%lx from %d\n", cl->read_op->req.hdr.magic, cl->read_op->req.hdr.id, cl->read_op->req.hdr.opcode, cl->peer_fd);
             stop_client(cl->peer_fd);
             return false;
         }
@@ -292,7 +292,7 @@ bool osd_messenger_t::handle_reply_hdr(osd_client_t *cl)
     if (req_it == cl->sent_ops.end())
     {
         // Command out of sync. Drop connection
-        printf("Client %d command out of sync: id %lu\n", cl->peer_fd, cl->read_op->req.hdr.id);
+        fprintf(stderr, "Client %d command out of sync: id %lu\n", cl->peer_fd, cl->read_op->req.hdr.id);
         stop_client(cl->peer_fd);
         return false;
     }
@@ -307,7 +307,7 @@ bool osd_messenger_t::handle_reply_hdr(osd_client_t *cl)
         if (op->reply.hdr.retval >= 0 && (op->reply.hdr.retval != expected_size || bmp_len > op->bitmap_len))
         {
             // Check reply length to not overflow the buffer
-            printf("Client %d read reply of different length: expected %u+%u, got %ld+%u\n",
+            fprintf(stderr, "Client %d read reply of different length: expected %u+%u, got %ld+%u\n",
                 cl->peer_fd, expected_size, op->bitmap_len, op->reply.hdr.retval, bmp_len);
             cl->sent_ops[op->req.hdr.id] = op;
             stop_client(cl->peer_fd);

src/msgr_send.cpp

@@ -227,7 +227,7 @@ void osd_messenger_t::handle_send(int result, osd_client_t *cl)
     if (result < 0 && result != -EAGAIN)
     {
         // this is a client socket, so don't panic. just disconnect it
-        printf("Client %d socket write error: %d (%s). Disconnecting client\n", cl->peer_fd, -result, strerror(-result));
+        fprintf(stderr, "Client %d socket write error: %d (%s). Disconnecting client\n", cl->peer_fd, -result, strerror(-result));
         stop_client(cl->peer_fd);
         return;
     }
@@ -274,7 +274,7 @@ void osd_messenger_t::handle_send(int result, osd_client_t *cl)
             // FIXME: Ignore pings during RDMA state transition
             if (log_level > 0)
             {
-                printf("Successfully connected with client %d using RDMA\n", cl->peer_fd);
+                fprintf(stderr, "Successfully connected with client %d using RDMA\n", cl->peer_fd);
             }
             cl->peer_state = PEER_RDMA;
             tfd->set_fd_handler(cl->peer_fd, false, NULL);

src/msgr_stop.cpp

@@ -41,7 +41,7 @@ void osd_messenger_t::cancel_op(osd_op_t *op)
     }
 }
 
-void osd_messenger_t::stop_client(int peer_fd, bool force)
+void osd_messenger_t::stop_client(int peer_fd, bool force, bool force_delete)
 {
     assert(peer_fd != 0);
     auto it = clients.find(peer_fd);
@@ -58,11 +58,11 @@ void osd_messenger_t::stop_client(int peer_fd, bool force)
     {
         if (cl->osd_num)
         {
-            printf("[OSD %lu] Stopping client %d (OSD peer %lu)\n", osd_num, peer_fd, cl->osd_num);
+            fprintf(stderr, "[OSD %lu] Stopping client %d (OSD peer %lu)\n", osd_num, peer_fd, cl->osd_num);
         }
         else
         {
-            printf("[OSD %lu] Stopping client %d (regular client)\n", osd_num, peer_fd);
+            fprintf(stderr, "[OSD %lu] Stopping client %d (regular client)\n", osd_num, peer_fd);
         }
     }
     // First set state to STOPPED so another stop_client() call doesn't try to free it again
@@ -136,7 +136,7 @@ void osd_messenger_t::stop_client(int peer_fd, bool force)
         clients.erase(it);
     }
     cl->refs--;
-    if (cl->refs <= 0)
+    if (cl->refs <= 0 || force_delete)
     {
         delete cl;
     }

src/nbd_proxy.cpp

@@ -10,6 +10,7 @@
 #include <netinet/tcp.h>
 #include <arpa/inet.h>
 #include <sys/un.h>
+#include <sys/epoll.h>
 #include <unistd.h>
 #include <fcntl.h>
 #include <signal.h>
@@ -200,9 +201,10 @@ public:
         fcntl(sockfd[0], F_SETFL, fcntl(sockfd[0], F_GETFL, 0) | O_NONBLOCK);
         nbd_fd = sockfd[0];
         load_module();
+        bool bg = cfg["foreground"].is_null();
         if (!cfg["dev_num"].is_null())
         {
-            if (run_nbd(sockfd, cfg["dev_num"].int64_value(), device_size, NBD_FLAG_SEND_FLUSH, 30) < 0)
+            if (run_nbd(sockfd, cfg["dev_num"].int64_value(), device_size, NBD_FLAG_SEND_FLUSH, 30, bg) < 0)
             {
                 perror("run_nbd");
                 exit(1);
@@ -214,7 +216,7 @@ public:
             int i = 0;
             while (true)
             {
-                int r = run_nbd(sockfd, i, device_size, NBD_FLAG_SEND_FLUSH, 30);
+                int r = run_nbd(sockfd, i, device_size, NBD_FLAG_SEND_FLUSH, 30, bg);
                 if (r == 0)
                 {
                     printf("/dev/nbd%d\n", i);
@@ -237,7 +239,7 @@ public:
                 }
             }
         }
-        if (cfg["foreground"].is_null())
+        if (bg)
         {
             daemonize();
         }
@@ -254,22 +256,47 @@ public:
         };
         ringloop->register_consumer(&consumer);
         // Add FD to epoll
-        epmgr->tfd->set_fd_handler(sockfd[0], false, [this](int peer_fd, int epoll_events)
+        bool stop = false;
+        epmgr->tfd->set_fd_handler(sockfd[0], false, [this, &stop](int peer_fd, int epoll_events)
         {
-            read_ready++;
-            submit_read();
+            if (epoll_events & EPOLLRDHUP)
+            {
+                close(peer_fd);
+                stop = true;
+            }
+            else
+            {
+                read_ready++;
+                submit_read();
+            }
         });
-        while (1)
+        while (!stop)
         {
             ringloop->loop();
             ringloop->wait();
         }
-        // FIXME: Cleanup when exiting
+        stop = false;
+        cluster_op_t *close_sync = new cluster_op_t;
+        close_sync->opcode = OSD_OP_SYNC;
+        close_sync->callback = [this, &stop](cluster_op_t *op)
+        {
+            stop = true;
+            delete op;
+        };
+        cli->execute(close_sync);
+        while (!stop)
+        {
+            ringloop->loop();
+            ringloop->wait();
+        }
+        delete cli;
+        delete epmgr;
+        delete ringloop;
     }
 
     void load_module()
     {
-        if (access("/sys/module/nbd", F_OK))
+        if (access("/sys/module/nbd", F_OK) == 0)
         {
             return;
         }
@@ -411,7 +438,7 @@ public:
     }
 
 protected:
-    int run_nbd(int sockfd[2], int dev_num, uint64_t size, uint64_t flags, unsigned timeout)
+    int run_nbd(int sockfd[2], int dev_num, uint64_t size, uint64_t flags, unsigned timeout, bool bg)
     {
         // Check handle size
         assert(sizeof(cur_req.handle) == 8);
@@ -459,11 +486,14 @@ protected:
         {
             // Run in child
             close(sockfd[0]);
+            if (bg)
+            {
+                daemonize();
+            }
             r = ioctl(nbd, NBD_DO_IT);
             if (r < 0)
             {
                 fprintf(stderr, "NBD device terminated with error: %s\n", strerror(errno));
-                kill(getppid(), SIGTERM);
             }
             close(sockfd[1]);
             ioctl(nbd, NBD_CLEAR_QUE);

src/osd.cpp

@@ -7,6 +7,8 @@
 #include <netinet/tcp.h>
 #include <arpa/inet.h>
 
+#include "blockstore_impl.h"
+#include "osd_primary.h"
 #include "osd.h"
 #include "http_client.h"
 
@@ -365,6 +367,7 @@ void osd_t::print_stats()
 
 void osd_t::print_slow()
 {
+    bool has_slow = false;
     char alloc[1024];
     timespec now;
     clock_gettime(CLOCK_REALTIME, &now);
@@ -426,9 +429,32 @@ void osd_t::print_slow()
                 {
                     bufprintf(" inode=%lx offset=%lx len=%x", op->req.rw.inode, op->req.rw.offset, op->req.rw.len);
                 }
+                if (op->req.hdr.opcode == OSD_OP_SEC_READ || op->req.hdr.opcode == OSD_OP_SEC_WRITE ||
+                    op->req.hdr.opcode == OSD_OP_SEC_WRITE_STABLE || op->req.hdr.opcode == OSD_OP_SEC_DELETE ||
+                    op->req.hdr.opcode == OSD_OP_SEC_SYNC || op->req.hdr.opcode == OSD_OP_SEC_LIST ||
+                    op->req.hdr.opcode == OSD_OP_SEC_STABILIZE || op->req.hdr.opcode == OSD_OP_SEC_ROLLBACK ||
+                    op->req.hdr.opcode == OSD_OP_SEC_READ_BMP)
+                {
+                    bufprintf(" state=%d", PRIV(op->bs_op)->op_state);
+                    int wait_for = PRIV(op->bs_op)->wait_for;
+                    if (wait_for)
+                    {
+                        bufprintf(" wait=%d (detail=%lu)", wait_for, PRIV(op->bs_op)->wait_detail);
+                    }
+                }
+                else if (op->req.hdr.opcode == OSD_OP_READ || op->req.hdr.opcode == OSD_OP_WRITE ||
+                    op->req.hdr.opcode == OSD_OP_SYNC || op->req.hdr.opcode == OSD_OP_DELETE)
+                {
+                    bufprintf(" state=%d", !op->op_data ? -1 : op->op_data->st);
+                }
 #undef bufprintf
                 printf("%s\n", alloc);
+                has_slow = true;
             }
         }
     }
+    if (has_slow)
+    {
+        bs->dump_diagnostics();
+    }
 }

src/osd_id.h

@@ -9,6 +9,8 @@
 #define POOL_ID_MAX 0x10000
 #define POOL_ID_BITS 16
 #define INODE_POOL(inode) (pool_id_t)((inode) >> (64 - POOL_ID_BITS))
+#define INODE_NO_POOL(inode) (inode_t)(inode & ((1l << (64-POOL_ID_BITS)) - 1))
+#define INODE_WITH_POOL(pool_id, inode) (((inode_t)(pool_id) << (64-POOL_ID_BITS)) | INODE_NO_POOL(inode))
 
 // Pool ID is 16 bits long
 typedef uint32_t pool_id_t;

src/osd_ops.h

@@ -191,6 +191,9 @@ struct __attribute__((__packed__)) osd_op_rw_t
     uint32_t flags;
     // inode metadata revision
     uint64_t meta_revision;
+    // object version for atomic "CAS" (compare-and-set) writes
+    // writes and deletes fail with -EINTR if object version differs from (version-1)
+    uint64_t version;
 };
 
 struct __attribute__((__packed__)) osd_reply_rw_t
@@ -199,6 +202,8 @@ struct __attribute__((__packed__)) osd_reply_rw_t
     // for reads: bitmap length
     uint32_t bitmap_len;
     uint32_t pad0;
+    // for reads: object version
+    uint64_t version;
 };
 
 // sync to the primary OSD

src/osd_primary.cpp

@@ -67,7 +67,9 @@ bool osd_t::prepare_primary_rw(osd_op_t *cur_op)
         }
         // Find parents from the same pool. Optimized reads only work within pools
         while (inode_it != st_cli.inode_config.end() && inode_it->second.parent_id &&
-            INODE_POOL(inode_it->second.parent_id) == pg_it->second.pool_id)
+            INODE_POOL(inode_it->second.parent_id) == pg_it->second.pool_id &&
+            // Check for loops
+            inode_it->second.parent_id != cur_op->req.rw.inode)
         {
             chain_size++;
             inode_it = st_cli.inode_config.find(inode_it->second.parent_id);
@@ -123,7 +125,10 @@ bool osd_t::prepare_primary_rw(osd_op_t *cur_op)
         int chain_num = 0;
         op_data->read_chain[chain_num++] = cur_op->req.rw.inode;
         auto inode_it = st_cli.inode_config.find(cur_op->req.rw.inode);
-        while (inode_it != st_cli.inode_config.end() && inode_it->second.parent_id)
+        while (inode_it != st_cli.inode_config.end() && inode_it->second.parent_id &&
+            INODE_POOL(inode_it->second.parent_id) == pg_it->second.pool_id &&
+            // Check for loops
+            inode_it->second.parent_id != cur_op->req.rw.inode)
         {
             op_data->read_chain[chain_num++] = inode_it->second.parent_id;
             inode_it = st_cli.inode_config.find(inode_it->second.parent_id);
@@ -193,7 +198,7 @@ void osd_t::continue_primary_read(osd_op_t *cur_op)
         {
             // Fast happy-path
             cur_op->buf = alloc_read_buffer(op_data->stripes, op_data->pg_data_size, 0);
-            submit_primary_subops(SUBMIT_READ, op_data->target_ver, pg.cur_set.data(), cur_op);
+            submit_primary_subops(SUBMIT_RMW_READ, op_data->target_ver, pg.cur_set.data(), cur_op);
             op_data->st = 1;
         }
         else
@@ -210,7 +215,7 @@ void osd_t::continue_primary_read(osd_op_t *cur_op)
             op_data->scheme = pg.scheme;
             op_data->degraded = 1;
             cur_op->buf = alloc_read_buffer(op_data->stripes, pg.pg_size, 0);
-            submit_primary_subops(SUBMIT_READ, op_data->target_ver, cur_set, cur_op);
+            submit_primary_subops(SUBMIT_RMW_READ, op_data->target_ver, cur_set, cur_op);
             op_data->st = 1;
         }
     }
@@ -222,6 +227,7 @@ resume_2:
         finish_op(cur_op, op_data->epipe > 0 ? -EPIPE : -EIO);
         return;
     }
+    cur_op->reply.rw.version = op_data->fact_ver;
     cur_op->reply.rw.bitmap_len = op_data->pg_data_size * clean_entry_bitmap_size;
     if (op_data->degraded)
     {
@@ -343,6 +349,13 @@ resume_3:
         pg_cancel_write_queue(pg, cur_op, op_data->oid, op_data->epipe > 0 ? -EPIPE : -EIO);
         return;
     }
+    // Check CAS version
+    if (cur_op->req.rw.version && op_data->fact_ver != (cur_op->req.rw.version-1))
+    {
+        cur_op->reply.hdr.retval = -EINTR;
+        cur_op->reply.rw.version = op_data->fact_ver;
+        goto continue_others;
+    }
     // Save version override for parallel reads
     pg.ver_override[op_data->oid] = op_data->fact_ver;
     // Submit deletes
@@ -370,6 +383,8 @@ resume_5:
         free_object_state(pg, &op_data->object_state);
     }
     pg.total_count--;
+    cur_op->reply.hdr.retval = 0;
+continue_others:
     osd_op_t *next_op = NULL;
     auto next_it = pg.write_queue.find(op_data->oid);
     if (next_it != pg.write_queue.end() && next_it->second == cur_op)
@@ -378,7 +393,7 @@ resume_5:
         if (next_it != pg.write_queue.end() && next_it->first == op_data->oid)
             next_op = next_it->second;
     }
-    finish_op(cur_op, cur_op->req.rw.len);
+    finish_op(cur_op, cur_op->reply.hdr.retval);
     if (next_op)
     {
         // Continue next write to the same object

src/osd_primary_chain.cpp

@@ -65,7 +65,10 @@ int osd_t::read_bitmaps(osd_op_t *cur_op, pg_t & pg, int base_state)
             auto vo_it = pg.ver_override.find(cur_oid);
             auto read_version = (vo_it != pg.ver_override.end() ? vo_it->second : UINT64_MAX);
             // Read bitmap synchronously from the local database
-            bs->read_bitmap(cur_oid, read_version, op_data->snapshot_bitmaps + chain_num*clean_entry_bitmap_size, NULL);
+            bs->read_bitmap(
+                cur_oid, read_version, op_data->snapshot_bitmaps + chain_num*clean_entry_bitmap_size,
+                !chain_num ? &cur_op->reply.rw.version : NULL
+            );
         }
     }
     else
@@ -228,7 +231,10 @@ int osd_t::submit_bitmap_subops(osd_op_t *cur_op, pg_t & pg)
                 // Read bitmap synchronously from the local database
                 for (int j = prev; j <= i; j++)
                 {
-                    bs->read_bitmap((*bitmap_requests)[j].oid, (*bitmap_requests)[j].version, (*bitmap_requests)[j].bmp_buf, NULL);
+                    bs->read_bitmap(
+                        (*bitmap_requests)[j].oid, (*bitmap_requests)[j].version, (*bitmap_requests)[j].bmp_buf,
+                        (*bitmap_requests)[j].oid.inode == cur_op->req.rw.inode ? &cur_op->reply.rw.version : NULL
+                    );
                 }
             }
             else
@@ -264,6 +270,10 @@ int osd_t::submit_bitmap_subops(osd_op_t *cur_op, pg_t & pg)
                         for (int j = prev; j <= i; j++)
                         {
                             memcpy((*bitmap_requests)[j].bmp_buf, cur_buf, clean_entry_bitmap_size);
+                            if ((*bitmap_requests)[j].oid.inode == cur_op->req.rw.inode)
+                            {
+                                memcpy(&cur_op->reply.rw.version, cur_buf-8, 8);
+                            }
                             cur_buf += 8 + clean_entry_bitmap_size;
                         }
                     }

src/osd_primary_write.cpp

@@ -96,6 +96,13 @@ resume_3:
         pg_cancel_write_queue(pg, cur_op, op_data->oid, op_data->epipe > 0 ? -EPIPE : -EIO);
         return;
     }
+    // Check CAS version
+    if (cur_op->req.rw.version && op_data->fact_ver != (cur_op->req.rw.version-1))
+    {
+        cur_op->reply.hdr.retval = -EINTR;
+        cur_op->reply.rw.version = op_data->fact_ver;
+        goto continue_others;
+    }
     if (op_data->scheme == POOL_SCHEME_REPLICATED)
     {
         // Set bitmap bits
@@ -254,6 +261,7 @@ resume_9:
         }
     }
     cur_op->reply.hdr.retval = cur_op->req.rw.len;
+    cur_op->reply.rw.version = op_data->fact_ver;
 continue_others:
     osd_op_t *next_op = NULL;
     auto next_it = pg.write_queue.find(op_data->oid);
@@ -265,7 +273,7 @@ continue_others:
             next_op = next_it->second;
     }
     // finish_op would invalidate next_it if it cleared pg.write_queue, but it doesn't do that :)
-    finish_op(cur_op, cur_op->req.rw.len);
+    finish_op(cur_op, cur_op->reply.hdr.retval);
     if (next_op)
     {
         // Continue next write to the same object

src/qemu_driver.c

@@ -26,7 +26,7 @@
 #define qobject_unref QDECREF
 #endif
 
-#include "qemu_proxy.h"
+#include "vitastor_c.h"
 
 void qemu_module_dummy(void)
 {
@@ -48,6 +48,7 @@ typedef struct VitastorClient
     uint64_t pool;
     uint64_t size;
     long readonly;
+    int use_rdma;
     char *rdma_device;
     int rdma_port_num;
     int rdma_gid_index;
@@ -65,10 +66,11 @@ typedef struct VitastorRPC
 } VitastorRPC;
 
 static void vitastor_co_init_task(BlockDriverState *bs, VitastorRPC *task);
-static void vitastor_co_generic_bh_cb(long retval, void *opaque);
+static void vitastor_co_generic_bh_cb(void *opaque, long retval);
+static void vitastor_co_read_cb(void *opaque, long retval, uint64_t version);
 static void vitastor_close(BlockDriverState *bs);
 
-static char *qemu_rbd_next_tok(char *src, char delim, char **p)
+static char *qemu_vitastor_next_tok(char *src, char delim, char **p)
 {
     char *end;
     *p = NULL;
@@ -87,7 +89,7 @@ static char *qemu_rbd_next_tok(char *src, char delim, char **p)
     return src;
 }
 
-static void qemu_rbd_unescape(char *src)
+static void qemu_vitastor_unescape(char *src)
 {
     char *p;
     for (p = src; *src; ++src, ++p)
@@ -120,18 +122,19 @@ static void vitastor_parse_filename(const char *filename, QDict *options, Error
     while (p)
     {
         char *name, *value;
-        name = qemu_rbd_next_tok(p, '=', &p);
+        name = qemu_vitastor_next_tok(p, '=', &p);
         if (!p)
         {
             error_setg(errp, "conf option %s has no value", name);
             break;
         }
-        qemu_rbd_unescape(name);
-        value = qemu_rbd_next_tok(p, ':', &p);
-        qemu_rbd_unescape(value);
+        qemu_vitastor_unescape(name);
+        value = qemu_vitastor_next_tok(p, ':', &p);
+        qemu_vitastor_unescape(value);
         if (!strcmp(name, "inode") ||
             !strcmp(name, "pool") ||
             !strcmp(name, "size") ||
+            !strcmp(name, "use_rdma") ||
             !strcmp(name, "rdma_port_num") ||
             !strcmp(name, "rdma_gid_index") ||
             !strcmp(name, "rdma_mtu"))
@@ -181,7 +184,7 @@ static void coroutine_fn vitastor_co_get_metadata(VitastorRPC *task)
     task->co = qemu_coroutine_self();
 
     qemu_mutex_lock(&client->mutex);
-    vitastor_proxy_watch_metadata(client->proxy, client->image, vitastor_co_generic_bh_cb, task);
+    vitastor_c_watch_inode(client->proxy, client->image, vitastor_co_generic_bh_cb, task);
     qemu_mutex_unlock(&client->mutex);
 
     while (!task->complete)
@@ -196,15 +199,17 @@ static int vitastor_file_open(BlockDriverState *bs, QDict *options, int flags, E
     int64_t ret = 0;
     qemu_mutex_init(&client->mutex);
     client->config_path = g_strdup(qdict_get_try_str(options, "config_path"));
+    // FIXME: Rename to etcd_address
     client->etcd_host = g_strdup(qdict_get_try_str(options, "etcd_host"));
     client->etcd_prefix = g_strdup(qdict_get_try_str(options, "etcd_prefix"));
+    client->use_rdma = qdict_get_try_int(options, "use_rdma", -1);
     client->rdma_device = g_strdup(qdict_get_try_str(options, "rdma_device"));
     client->rdma_port_num = qdict_get_try_int(options, "rdma_port_num", 0);
     client->rdma_gid_index = qdict_get_try_int(options, "rdma_gid_index", 0);
     client->rdma_mtu = qdict_get_try_int(options, "rdma_mtu", 0);
-    client->proxy = vitastor_proxy_create(
-        bdrv_get_aio_context(bs), client->config_path, client->etcd_host, client->etcd_prefix,
-        client->rdma_device, client->rdma_port_num, client->rdma_gid_index, client->rdma_mtu
+    client->proxy = vitastor_c_create_qemu(
+        (QEMUSetFDHandler*)aio_set_fd_handler, bdrv_get_aio_context(bs), client->config_path, client->etcd_host, client->etcd_prefix,
+        client->use_rdma, client->rdma_device, client->rdma_port_num, client->rdma_gid_index, client->rdma_mtu, 0
     );
     client->image = g_strdup(qdict_get_try_str(options, "image"));
     client->readonly = (flags & BDRV_O_RDWR) ? 1 : 0;
@@ -224,9 +229,9 @@ static int vitastor_file_open(BlockDriverState *bs, QDict *options, int flags, E
         }
         BDRV_POLL_WHILE(bs, !task.complete);
         client->watch = (void*)task.ret;
-        client->readonly = client->readonly || vitastor_proxy_get_readonly(client->watch);
-        client->size = vitastor_proxy_get_size(client->watch);
-        if (!vitastor_proxy_get_inode_num(client->watch))
+        client->readonly = client->readonly || vitastor_c_inode_get_readonly(client->watch);
+        client->size = vitastor_c_inode_get_size(client->watch);
+        if (!vitastor_c_inode_get_num(client->watch))
         {
             error_setg(errp, "image does not exist");
             vitastor_close(bs);
@@ -255,6 +260,7 @@ static int vitastor_file_open(BlockDriverState *bs, QDict *options, int flags, E
     }
     bs->total_sectors = client->size / BDRV_SECTOR_SIZE;
     //client->aio_context = bdrv_get_aio_context(bs);
+    qdict_del(options, "use_rdma");
     qdict_del(options, "rdma_mtu");
     qdict_del(options, "rdma_gid_index");
     qdict_del(options, "rdma_port_num");
@@ -272,7 +278,7 @@ static int vitastor_file_open(BlockDriverState *bs, QDict *options, int flags, E
 static void vitastor_close(BlockDriverState *bs)
 {
     VitastorClient *client = bs->opaque;
-    vitastor_proxy_destroy(client->proxy);
+    vitastor_c_destroy(client->proxy);
     qemu_mutex_destroy(&client->mutex);
     if (client->config_path)
         g_free(client->config_path);
@@ -387,7 +393,7 @@ static void vitastor_co_init_task(BlockDriverState *bs, VitastorRPC *task)
     };
 }
 
-static void vitastor_co_generic_bh_cb(long retval, void *opaque)
+static void vitastor_co_generic_bh_cb(void *opaque, long retval)
 {
     VitastorRPC *task = opaque;
     task->ret = retval;
@@ -403,6 +409,11 @@ static void vitastor_co_generic_bh_cb(long retval, void *opaque)
     }
 }
 
+static void vitastor_co_read_cb(void *opaque, long retval, uint64_t version)
+{
+    vitastor_co_generic_bh_cb(opaque, retval);
+}
+
 static int coroutine_fn vitastor_co_preadv(BlockDriverState *bs, uint64_t offset, uint64_t bytes, QEMUIOVector *iov, int flags)
 {
     VitastorClient *client = bs->opaque;
@@ -410,9 +421,9 @@ static int coroutine_fn vitastor_co_preadv(BlockDriverState *bs, uint64_t offset
     vitastor_co_init_task(bs, &task);
     task.iov = iov;
 
-    uint64_t inode = client->watch ? vitastor_proxy_get_inode_num(client->watch) : client->inode;
+    uint64_t inode = client->watch ? vitastor_c_inode_get_num(client->watch) : client->inode;
     qemu_mutex_lock(&client->mutex);
-    vitastor_proxy_rw(0, client->proxy, inode, offset, bytes, iov->iov, iov->niov, vitastor_co_generic_bh_cb, &task);
+    vitastor_c_read(client->proxy, inode, offset, bytes, iov->iov, iov->niov, vitastor_co_read_cb, &task);
     qemu_mutex_unlock(&client->mutex);
 
     while (!task.complete)
@@ -430,9 +441,9 @@ static int coroutine_fn vitastor_co_pwritev(BlockDriverState *bs, uint64_t offse
     vitastor_co_init_task(bs, &task);
     task.iov = iov;
 
-    uint64_t inode = client->watch ? vitastor_proxy_get_inode_num(client->watch) : client->inode;
+    uint64_t inode = client->watch ? vitastor_c_inode_get_num(client->watch) : client->inode;
     qemu_mutex_lock(&client->mutex);
-    vitastor_proxy_rw(1, client->proxy, inode, offset, bytes, iov->iov, iov->niov, vitastor_co_generic_bh_cb, &task);
+    vitastor_c_write(client->proxy, inode, offset, bytes, 0, iov->iov, iov->niov, vitastor_co_generic_bh_cb, &task);
     qemu_mutex_unlock(&client->mutex);
 
     while (!task.complete)
@@ -462,7 +473,7 @@ static int coroutine_fn vitastor_co_flush(BlockDriverState *bs)
     vitastor_co_init_task(bs, &task);
 
     qemu_mutex_lock(&client->mutex);
-    vitastor_proxy_sync(client->proxy, vitastor_co_generic_bh_cb, &task);
+    vitastor_c_sync(client->proxy, vitastor_co_generic_bh_cb, &task);
     qemu_mutex_unlock(&client->mutex);
 
     while (!task.complete)

src/qemu_proxy.cpp

@@ -1,177 +0,0 @@
-// Copyright (c) Vitaliy Filippov, 2019+
-// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
-
-// C-C++ proxy for the QEMU driver
-// (QEMU headers don't compile with g++)
-
-#include <sys/epoll.h>
-
-#include "cluster_client.h"
-
-typedef void* AioContext;
-#include "qemu_proxy.h"
-
-extern "C"
-{
-    // QEMU
-    typedef void IOHandler(void *opaque);
-    void aio_set_fd_handler(AioContext *ctx, int fd, int is_external, IOHandler *fd_read, IOHandler *fd_write, void *poll_fn, void *opaque);
-}
-
-struct QemuProxyData
-{
-    int fd;
-    std::function<void(int, int)> callback;
-};
-
-class QemuProxy
-{
-    std::map<int, QemuProxyData> handlers;
-
-public:
-
-    timerfd_manager_t *tfd;
-    cluster_client_t *cli;
-    AioContext *ctx;
-
-    QemuProxy(AioContext *ctx, const char *config_path, const char *etcd_host, const char *etcd_prefix,
-        const char *rdma_device, int rdma_port_num, int rdma_gid_index, int rdma_mtu)
-    {
-        this->ctx = ctx;
-        json11::Json::object cfg;
-        if (config_path)
-            cfg["config_path"] = std::string(config_path);
-        if (etcd_host)
-            cfg["etcd_address"] = std::string(etcd_host);
-        if (etcd_prefix)
-            cfg["etcd_prefix"] = std::string(etcd_prefix);
-        if (rdma_device)
-            cfg["rdma_device"] = std::string(rdma_device);
-        if (rdma_port_num)
-            cfg["rdma_port_num"] = rdma_port_num;
-        if (rdma_gid_index)
-            cfg["rdma_gid_index"] = rdma_gid_index;
-        if (rdma_mtu)
-            cfg["rdma_mtu"] = rdma_mtu;
-        json11::Json cfg_json(cfg);
-        tfd = new timerfd_manager_t([this](int fd, bool wr, std::function<void(int, int)> callback) { set_fd_handler(fd, wr, callback); });
-        cli = new cluster_client_t(NULL, tfd, cfg_json);
-    }
-
-    ~QemuProxy()
-    {
-        delete cli;
-        delete tfd;
-    }
-
-    void set_fd_handler(int fd, bool wr, std::function<void(int, int)> callback)
-    {
-        if (callback != NULL)
-        {
-            handlers[fd] = { .fd = fd, .callback = callback };
-            aio_set_fd_handler(ctx, fd, false, &QemuProxy::read_handler, wr ? &QemuProxy::write_handler : NULL, NULL, &handlers[fd]);
-        }
-        else
-        {
-            handlers.erase(fd);
-            aio_set_fd_handler(ctx, fd, false, NULL, NULL, NULL, NULL);
-        }
-    }
-
-    static void read_handler(void *opaque)
-    {
-        QemuProxyData *data = (QemuProxyData *)opaque;
-        data->callback(data->fd, EPOLLIN);
-    }
-
-    static void write_handler(void *opaque)
-    {
-        QemuProxyData *data = (QemuProxyData *)opaque;
-        data->callback(data->fd, EPOLLOUT);
-    }
-};
-
-extern "C" {
-
-void* vitastor_proxy_create(AioContext *ctx, const char *config_path, const char *etcd_host, const char *etcd_prefix,
-    const char *rdma_device, int rdma_port_num, int rdma_gid_index, int rdma_mtu)
-{
-    QemuProxy *p = new QemuProxy(ctx, config_path, etcd_host, etcd_prefix, rdma_device, rdma_port_num, rdma_gid_index, rdma_mtu);
-    return p;
-}
-
-void vitastor_proxy_destroy(void *client)
-{
-    QemuProxy *p = (QemuProxy*)client;
-    delete p;
-}
-
-void vitastor_proxy_rw(int write, void *client, uint64_t inode, uint64_t offset, uint64_t len,
-    iovec *iov, int iovcnt, VitastorIOHandler cb, void *opaque)
-{
-    QemuProxy *p = (QemuProxy*)client;
-    cluster_op_t *op = new cluster_op_t;
-    op->opcode = write ? OSD_OP_WRITE : OSD_OP_READ;
-    op->inode = inode;
-    op->offset = offset;
-    op->len = len;
-    for (int i = 0; i < iovcnt; i++)
-    {
-        op->iov.push_back(iov[i].iov_base, iov[i].iov_len);
-    }
-    op->callback = [cb, opaque](cluster_op_t *op)
-    {
-        cb(op->retval, opaque);
-        delete op;
-    };
-    p->cli->execute(op);
-}
-
-void vitastor_proxy_sync(void *client, VitastorIOHandler cb, void *opaque)
-{
-    QemuProxy *p = (QemuProxy*)client;
-    cluster_op_t *op = new cluster_op_t;
-    op->opcode = OSD_OP_SYNC;
-    op->callback = [cb, opaque](cluster_op_t *op)
-    {
-        cb(op->retval, opaque);
-        delete op;
-    };
-    p->cli->execute(op);
-}
-
-void vitastor_proxy_watch_metadata(void *client, char *image, VitastorIOHandler cb, void *opaque)
-{
-    QemuProxy *p = (QemuProxy*)client;
-    p->cli->on_ready([=]()
-    {
-        auto watch = p->cli->st_cli.watch_inode(std::string(image));
-        cb((long)watch, opaque);
-    });
-}
-
-void vitastor_proxy_close_watch(void *client, void *watch)
-{
-    QemuProxy *p = (QemuProxy*)client;
-    p->cli->st_cli.close_watch((inode_watch_t*)watch);
-}
-
-uint64_t vitastor_proxy_get_size(void *watch_ptr)
-{
-    inode_watch_t *watch = (inode_watch_t*)watch_ptr;
-    return watch->cfg.size;
-}
-
-uint64_t vitastor_proxy_get_inode_num(void *watch_ptr)
-{
-    inode_watch_t *watch = (inode_watch_t*)watch_ptr;
-    return watch->cfg.num;
-}
-
-int vitastor_proxy_get_readonly(void *watch_ptr)
-{
-    inode_watch_t *watch = (inode_watch_t*)watch_ptr;
-    return watch->cfg.readonly;
-}
-
-}

src/qemu_proxy.h

@@ -1,35 +0,0 @@
-// Copyright (c) Vitaliy Filippov, 2019+
-// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
-
-#ifndef VITASTOR_QEMU_PROXY_H
-#define VITASTOR_QEMU_PROXY_H
-
-#ifndef POOL_ID_BITS
-#define POOL_ID_BITS 16
-#endif
-#include <stdint.h>
-#include <sys/uio.h>
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-// Our exports
-typedef void VitastorIOHandler(long retval, void *opaque);
-void* vitastor_proxy_create(AioContext *ctx, const char *config_path, const char *etcd_host, const char *etcd_prefix,
-    const char *rdma_device, int rdma_port_num, int rdma_gid_index, int rdma_mtu);
-void vitastor_proxy_destroy(void *client);
-void vitastor_proxy_rw(int write, void *client, uint64_t inode, uint64_t offset, uint64_t len,
-    struct iovec *iov, int iovcnt, VitastorIOHandler cb, void *opaque);
-void vitastor_proxy_sync(void *client, VitastorIOHandler cb, void *opaque);
-void vitastor_proxy_watch_metadata(void *client, char *image, VitastorIOHandler cb, void *opaque);
-void vitastor_proxy_close_watch(void *client, void *watch);
-uint64_t vitastor_proxy_get_size(void *watch);
-uint64_t vitastor_proxy_get_inode_num(void *watch);
-int vitastor_proxy_get_readonly(void *watch);
-
-#ifdef __cplusplus
-}
-#endif
-
-#endif

src/rm_inode.cpp

@@ -1,410 +0,0 @@
-// Copyright (c) Vitaliy Filippov, 2019+
-// License: VNPL-1.1 (see README.md for details)
-
-/**
- * Inode removal tool
- * May be included into a bigger "command-line management interface" in the future
- */
-
-#include <vector>
-#include <algorithm>
-
-#include "epoll_manager.h"
-#include "cluster_client.h"
-#include "pg_states.h"
-
-#define RM_LISTING 1
-#define RM_REMOVING 2
-#define RM_END 3
-
-const char *exe_name = NULL;
-
-struct rm_pg_t;
-
-struct rm_pg_osd_t
-{
-    rm_pg_t *pg = NULL;
-    osd_num_t osd_num;
-    bool sent = false;
-};
-
-struct rm_pg_t
-{
-    pg_num_t pg_num;
-    osd_num_t rm_osd_num;
-    std::vector<rm_pg_osd_t> list_osds;
-    int state = 0;
-    int to_list;
-    std::set<object_id> objects;
-    std::set<object_id>::iterator obj_pos;
-    uint64_t obj_count = 0, obj_done = 0, obj_prev_done = 0;
-    int in_flight = 0;
-};
-
-class rm_inode_t
-{
-protected:
-    uint64_t inode = 0;
-    pool_id_t pool_id = 0;
-    uint64_t iodepth = 0, parallel_osds = 0;
-
-    ring_loop_t *ringloop = NULL;
-    epoll_manager_t *epmgr = NULL;
-    cluster_client_t *cli = NULL;
-    ring_consumer_t consumer;
-
-    std::vector<rm_pg_t*> lists;
-    uint64_t total_count = 0, total_done = 0, total_prev_pct = 0;
-    uint64_t pgs_to_list = 0;
-    bool started = false;
-    bool progress = true;
-    bool list_first = false;
-    int log_level = 0;
-
-public:
-    static json11::Json::object parse_args(int narg, const char *args[])
-    {
-        json11::Json::object cfg;
-        cfg["progress"] = "1";
-        for (int i = 1; i < narg; i++)
-        {
-            if (!strcmp(args[i], "-h") || !strcmp(args[i], "--help"))
-            {
-                help();
-            }
-            else if (args[i][0] == '-' && args[i][1] == '-')
-            {
-                const char *opt = args[i]+2;
-                cfg[opt] = !strcmp(opt, "json") || !strcmp(opt, "wait-list") || i == narg-1 ? "1" : args[++i];
-            }
-        }
-        return cfg;
-    }
-
-    static void help()
-    {
-        printf(
-            "Vitastor inode removal tool\n"
-            "(c) Vitaliy Filippov, 2020 (VNPL-1.1)\n\n"
-            "USAGE:\n"
-            "  %s [--etcd_address <etcd_address>] --pool <pool> --inode <inode> [--wait-list]\n",
-            exe_name
-        );
-        exit(0);
-    }
-
-    void run(json11::Json cfg)
-    {
-        inode = cfg["inode"].uint64_value();
-        pool_id = cfg["pool"].uint64_value();
-        if (pool_id)
-            inode = (inode & ((1l << (64-POOL_ID_BITS)) - 1)) | (((uint64_t)pool_id) << (64-POOL_ID_BITS));
-        pool_id = INODE_POOL(inode);
-        if (!pool_id)
-        {
-            fprintf(stderr, "pool is missing");
-            exit(1);
-        }
-        iodepth = cfg["iodepth"].uint64_value();
-        if (!iodepth)
-            iodepth = 32;
-        parallel_osds = cfg["parallel_osds"].uint64_value();
-        if (!parallel_osds)
-            parallel_osds = 4;
-        log_level = cfg["log_level"].int64_value();
-        progress = cfg["progress"].uint64_value() ? true : false;
-        list_first = cfg["wait-list"].uint64_value() ? true : false;
-        // Create client
-        ringloop = new ring_loop_t(512);
-        epmgr = new epoll_manager_t(ringloop);
-        cli = new cluster_client_t(ringloop, epmgr->tfd, cfg);
-        cli->on_ready([this]() { start_delete(); });
-        // Initialize job
-        consumer.loop = [this]()
-        {
-            if (started)
-                continue_delete();
-            ringloop->submit();
-        };
-        ringloop->register_consumer(&consumer);
-        // Loop until it completes
-        while (1)
-        {
-            ringloop->loop();
-            ringloop->wait();
-        }
-    }
-
-    void start_delete()
-    {
-        if (cli->st_cli.pool_config.find(pool_id) == cli->st_cli.pool_config.end())
-        {
-            fprintf(stderr, "Pool %u does not exist\n", pool_id);
-            exit(1);
-        }
-        auto pool_cfg = cli->st_cli.pool_config[pool_id];
-        for (auto & pg_item: pool_cfg.pg_config)
-        {
-            auto & pg = pg_item.second;
-            if (pg.pause || !pg.cur_primary || !(pg.cur_state & PG_ACTIVE))
-            {
-                fprintf(stderr, "PG %u is inactive, skipping\n", pg_item.first);
-                continue;
-            }
-            rm_pg_t *r = new rm_pg_t();
-            r->pg_num = pg_item.first;
-            r->rm_osd_num = pg.cur_primary;
-            r->state = RM_LISTING;
-            if (pg.cur_state != PG_ACTIVE)
-            {
-                std::set<osd_num_t> all_peers;
-                for (osd_num_t pg_osd: pg.target_set)
-                {
-                    if (pg_osd != 0)
-                    {
-                        all_peers.insert(pg_osd);
-                    }
-                }
-                for (osd_num_t pg_osd: pg.all_peers)
-                {
-                    if (pg_osd != 0)
-                    {
-                        all_peers.insert(pg_osd);
-                    }
-                }
-                for (auto & hist_item: pg.target_history)
-                {
-                    for (auto pg_osd: hist_item)
-                    {
-                        if (pg_osd != 0)
-                        {
-                            all_peers.insert(pg_osd);
-                        }
-                    }
-                }
-                for (osd_num_t peer_osd: all_peers)
-                {
-                    r->list_osds.push_back((rm_pg_osd_t){ .pg = r, .osd_num = peer_osd, .sent = false });
-                }
-            }
-            else
-            {
-                r->list_osds.push_back((rm_pg_osd_t){ .pg = r, .osd_num = pg.cur_primary, .sent = false });
-            }
-            r->to_list = r->list_osds.size();
-            lists.push_back(r);
-        }
-        std::sort(lists.begin(), lists.end(), [](rm_pg_t *a, rm_pg_t *b)
-        {
-            return a->rm_osd_num < b->rm_osd_num ? true : false;
-        });
-        pgs_to_list = lists.size();
-        started = true;
-        continue_delete();
-    }
-
-    void send_list(rm_pg_osd_t *cur_list)
-    {
-        if (cur_list->sent)
-        {
-            return;
-        }
-        if (cli->msgr.osd_peer_fds.find(cur_list->osd_num) ==
-            cli->msgr.osd_peer_fds.end())
-        {
-            // Initiate connection
-            cli->msgr.connect_peer(cur_list->osd_num, cli->st_cli.peer_states[cur_list->osd_num]);
-            return;
-        }
-        osd_op_t *op = new osd_op_t();
-        op->op_type = OSD_OP_OUT;
-        op->peer_fd = cli->msgr.osd_peer_fds[cur_list->osd_num];
-        op->req = (osd_any_op_t){
-            .sec_list = {
-                .header = {
-                    .magic = SECONDARY_OSD_OP_MAGIC,
-                    .id = cli->msgr.next_subop_id++,
-                    .opcode = OSD_OP_SEC_LIST,
-                },
-                .list_pg = cur_list->pg->pg_num,
-                .pg_count = (pg_num_t)cli->st_cli.pool_config[pool_id].real_pg_count,
-                .pg_stripe_size = cli->st_cli.pool_config[pool_id].pg_stripe_size,
-                .min_inode = inode,
-                .max_inode = inode,
-            },
-        };
-        op->callback = [this, cur_list](osd_op_t *op)
-        {
-            cur_list->pg->to_list--;
-            if (op->reply.hdr.retval < 0)
-            {
-                fprintf(stderr, "Failed to get PG %u/%u object list from OSD %lu (retval=%ld), skipping\n",
-                    pool_id, cur_list->pg->pg_num, cur_list->osd_num, op->reply.hdr.retval);
-            }
-            else
-            {
-                if (op->reply.sec_list.stable_count < op->reply.hdr.retval)
-                {
-                    // Unstable objects, if present, mean that someone still writes into the inode. Warn the user about it.
-                    printf(
-                        "[PG %u/%u] Inode still has %lu unstable object versions - is it still open? Not a good idea to delete it.\n",
-                        pool_id, cur_list->pg->pg_num, op->reply.hdr.retval - op->reply.sec_list.stable_count
-                    );
-                }
-                if (log_level > 0)
-                {
-                    printf(
-                        "[PG %u/%u] Got inode object list from OSD %lu: %ld object versions\n",
-                        pool_id, cur_list->pg->pg_num, cur_list->osd_num, op->reply.hdr.retval
-                    );
-                }
-                for (uint64_t i = 0; i < op->reply.hdr.retval; i++)
-                {
-                    object_id oid = ((obj_ver_id*)op->buf)[i].oid;
-                    oid.stripe = oid.stripe & ~STRIPE_MASK;
-                    cur_list->pg->objects.insert(oid);
-                }
-            }
-            delete op;
-            if (cur_list->pg->to_list <= 0)
-            {
-                cur_list->pg->obj_done = cur_list->pg->obj_prev_done = 0;
-                cur_list->pg->obj_pos = cur_list->pg->objects.begin();
-                cur_list->pg->obj_count = cur_list->pg->objects.size();
-                total_count += cur_list->pg->obj_count;
-                total_prev_pct = 0;
-                cur_list->pg->state = RM_REMOVING;
-                pgs_to_list--;
-            }
-            continue_delete();
-        };
-        cli->msgr.outbox_push(op);
-        cur_list->sent = true;
-    }
-
-    void send_ops(rm_pg_t *cur_list)
-    {
-        if (cli->msgr.osd_peer_fds.find(cur_list->rm_osd_num) ==
-            cli->msgr.osd_peer_fds.end())
-        {
-            // Initiate connection
-            cli->msgr.connect_peer(cur_list->rm_osd_num, cli->st_cli.peer_states[cur_list->rm_osd_num]);
-            return;
-        }
-        while (cur_list->in_flight < iodepth && cur_list->obj_pos != cur_list->objects.end())
-        {
-            osd_op_t *op = new osd_op_t();
-            op->op_type = OSD_OP_OUT;
-            op->peer_fd = cli->msgr.osd_peer_fds[cur_list->rm_osd_num];
-            op->req = (osd_any_op_t){
-                .rw = {
-                    .header = {
-                        .magic = SECONDARY_OSD_OP_MAGIC,
-                        .id = cli->msgr.next_subop_id++,
-                        .opcode = OSD_OP_DELETE,
-                    },
-                    .inode = cur_list->obj_pos->inode,
-                    .offset = (cur_list->obj_pos->stripe & ~STRIPE_MASK),
-                    .len = 0,
-                },
-            };
-            op->callback = [this, cur_list](osd_op_t *op)
-            {
-                cur_list->in_flight--;
-                if (op->reply.hdr.retval < 0)
-                {
-                    fprintf(stderr, "Failed to remove object from PG %u (OSD %lu) (retval=%ld)\n",
-                        cur_list->pg_num, cur_list->rm_osd_num, op->reply.hdr.retval);
-                }
-                delete op;
-                cur_list->obj_done++;
-                total_done++;
-                continue_delete();
-            };
-            cli->msgr.outbox_push(op);
-            cur_list->obj_pos++;
-            cur_list->in_flight++;
-        }
-        if (!cur_list->in_flight && cur_list->obj_pos == cur_list->objects.end())
-        {
-            cur_list->obj_count = 0;
-            cur_list->obj_done = cur_list->obj_prev_done = 0;
-            cur_list->state = RM_END;
-        }
-    }
-
-    void continue_delete()
-    {
-        int par_osd = 0;
-        osd_num_t max_seen_osd = 0;
-        bool no_del = false;
-        if (list_first)
-        {
-            int i, n = 0;
-            for (i = 0; i < lists.size(); i++)
-            {
-                if (lists[i]->state == RM_LISTING)
-                {
-                    n++;
-                }
-            }
-            if (n > 0)
-            {
-                no_del = true;
-            }
-        }
-        for (int i = 0; i < lists.size(); i++)
-        {
-            if (lists[i]->state == RM_END)
-            {
-                delete lists[i];
-                lists.erase(lists.begin()+i, lists.begin()+i+1);
-                i--;
-            }
-            else if (lists[i]->rm_osd_num > max_seen_osd)
-            {
-                if (lists[i]->state == RM_LISTING)
-                {
-                    for (int j = 0; j < lists[i]->list_osds.size(); j++)
-                    {
-                        send_list(&lists[i]->list_osds[j]);
-                    }
-                }
-                else if (lists[i]->state == RM_REMOVING)
-                {
-                    if (no_del)
-                    {
-                        continue;
-                    }
-                    send_ops(lists[i]);
-                }
-                par_osd++;
-                max_seen_osd = lists[i]->rm_osd_num;
-                if (par_osd >= parallel_osds)
-                {
-                    break;
-                }
-            }
-        }
-        if (progress && total_count > 0 && total_done*1000/total_count != total_prev_pct)
-        {
-            printf("\rRemoved %lu/%lu objects, %lu more PGs to list...", total_done, total_count, pgs_to_list);
-            total_prev_pct = total_done*1000/total_count;
-        }
-        if (!lists.size())
-        {
-            printf("Done, inode %lu in pool %u removed\n", (inode & ((1l << (64-POOL_ID_BITS)) - 1)), pool_id);
-            exit(0);
-        }
-    }
-};
-
-int main(int narg, const char *args[])
-{
-    setvbuf(stdout, NULL, _IONBF, 0);
-    setvbuf(stderr, NULL, _IONBF, 0);
-    exe_name = args[0];
-    rm_inode_t *p = new rm_inode_t();
-    p->run(rm_inode_t::parse_args(narg, args));
-    return 0;
-}

src/test_cas.cpp

@@ -0,0 +1,135 @@
+// Copyright (c) Vitaliy Filippov, 2019+
+// License: VNPL-1.1 (see README.md for details)
+
+#include <stdio.h>
+#include <stdlib.h>
+
+#include "epoll_manager.h"
+#include "cluster_client.h"
+
+void send_read(cluster_client_t *cli, uint64_t inode, std::function<void(int, uint64_t)> cb)
+{
+    cluster_op_t *op = new cluster_op_t();
+    op->opcode = OSD_OP_READ;
+    op->inode = inode;
+    op->offset = 0;
+    op->len = 4096;
+    op->iov.push_back(malloc_or_die(op->len), op->len);
+    op->callback = [cb](cluster_op_t *op)
+    {
+        uint64_t version = op->version;
+        int retval = op->retval;
+        if (retval == op->len)
+            retval = 0;
+        free(op->iov.buf[0].iov_base);
+        delete op;
+        if (cb != NULL)
+            cb(retval, version);
+    };
+    cli->execute(op);
+}
+
+void send_write(cluster_client_t *cli, uint64_t inode, int byte, uint64_t version, std::function<void(int)> cb)
+{
+    cluster_op_t *op = new cluster_op_t();
+    op->opcode = OSD_OP_WRITE;
+    op->inode = inode;
+    op->offset = 0;
+    op->len = 4096;
+    op->version = version;
+    op->iov.push_back(malloc_or_die(op->len), op->len);
+    memset(op->iov.buf[0].iov_base, byte, op->len);
+    op->callback = [cb](cluster_op_t *op)
+    {
+        int retval = op->retval;
+        if (retval == op->len)
+            retval = 0;
+        free(op->iov.buf[0].iov_base);
+        delete op;
+        if (cb != NULL)
+            cb(retval);
+    };
+    cli->execute(op);
+}
+
+int main(int narg, char *args[])
+{
+    json11::Json::object cfgo;
+    for (int i = 1; i < narg; i++)
+    {
+        if (args[i][0] == '-' && args[i][1] == '-')
+        {
+            const char *opt = args[i]+2;
+            cfgo[opt] = i == narg-1 ? "1" : args[++i];
+        }
+    }
+    json11::Json cfg(cfgo);
+    uint64_t inode = (cfg["pool_id"].uint64_value() << (64-POOL_ID_BITS))
+        | cfg["inode_id"].uint64_value();
+    uint64_t base_ver = 0;
+    // Create client
+    auto ringloop = new ring_loop_t(512);
+    auto epmgr = new epoll_manager_t(ringloop);
+    auto cli = new cluster_client_t(ringloop, epmgr->tfd, cfg);
+    cli->on_ready([&]()
+    {
+        send_read(cli, inode, [&](int r, uint64_t v)
+        {
+            if (r < 0)
+            {
+                fprintf(stderr, "Initial read operation failed\n");
+                exit(1);
+            }
+            base_ver = v;
+            // CAS v=1 = compare with zero, non-existing object
+            send_write(cli, inode, 0x01, base_ver+1, [&](int r)
+            {
+                if (r < 0)
+                {
+                    fprintf(stderr, "CAS for non-existing object failed\n");
+                    exit(1);
+                }
+                // Check that read returns the new version
+                send_read(cli, inode, [&](int r, uint64_t v)
+                {
+                    if (r < 0)
+                    {
+                        fprintf(stderr, "Read operation failed after write\n");
+                        exit(1);
+                    }
+                    if (v != base_ver+1)
+                    {
+                        fprintf(stderr, "Read operation failed to return the new version number\n");
+                        exit(1);
+                    }
+                    // CAS v=2 = compare with v=1, existing object
+                    send_write(cli, inode, 0x02, base_ver+2, [&](int r)
+                    {
+                        if (r < 0)
+                        {
+                            fprintf(stderr, "CAS for existing object failed\n");
+                            exit(1);
+                        }
+                        // CAS v=2 again = compare with v=1, but version is 2. Must fail with -EINTR
+                        send_write(cli, inode, 0x03, base_ver+2, [&](int r)
+                        {
+                            if (r != -EINTR)
+                            {
+                                fprintf(stderr, "CAS conflict detection failed\n");
+                                exit(1);
+                            }
+                            printf("Basic CAS test succeeded\n");
+                            exit(0);
+                        });
+                    });
+                });
+            });
+        });
+    });
+    while (1)
+    {
+        ringloop->loop();
+        ringloop->wait();
+    }
+    return 0;
+}

src/vitastor_c.cpp

@@ -0,0 +1,254 @@
+// Copyright (c) Vitaliy Filippov, 2019+
+// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
+
+// Simplified C client library for QEMU, fio and other external drivers
+// Also acts as a C-C++ proxy for the QEMU driver (QEMU headers don't compile with g++)
+
+#include <sys/epoll.h>
+
+#include "ringloop.h"
+#include "epoll_manager.h"
+#include "cluster_client.h"
+
+#include "vitastor_c.h"
+
+struct vitastor_qemu_fd_t
+{
+    int fd;
+    std::function<void(int, int)> callback;
+};
+
+struct vitastor_c
+{
+    std::map<int, vitastor_qemu_fd_t> handlers;
+    ring_loop_t *ringloop = NULL;
+    epoll_manager_t *epmgr = NULL;
+    timerfd_manager_t *tfd = NULL;
+    cluster_client_t *cli = NULL;
+
+    QEMUSetFDHandler *aio_set_fd_handler = NULL;
+    void *aio_ctx = NULL;
+};
+
+extern "C" {
+
+static json11::Json vitastor_c_common_config(const char *config_path, const char *etcd_host, const char *etcd_prefix,
+    int use_rdma, const char *rdma_device, int rdma_port_num, int rdma_gid_index, int rdma_mtu, int log_level)
+{
+    json11::Json::object cfg;
+    if (config_path)
+        cfg["config_path"] = std::string(config_path);
+    if (etcd_host)
+        cfg["etcd_address"] = std::string(etcd_host);
+    if (etcd_prefix)
+        cfg["etcd_prefix"] = std::string(etcd_prefix);
+    // -1 means unspecified
+    if (use_rdma >= 0)
+        cfg["use_rdma"] = use_rdma > 0;
+    if (rdma_device)
+        cfg["rdma_device"] = std::string(rdma_device);
+    if (rdma_port_num)
+        cfg["rdma_port_num"] = rdma_port_num;
+    if (rdma_gid_index)
+        cfg["rdma_gid_index"] = rdma_gid_index;
+    if (rdma_mtu)
+        cfg["rdma_mtu"] = rdma_mtu;
+    if (log_level)
+        cfg["log_level"] = log_level;
+    return json11::Json(cfg);
+}
+
+static void vitastor_c_read_handler(void *opaque)
+{
+    vitastor_qemu_fd_t *data = (vitastor_qemu_fd_t *)opaque;
+    data->callback(data->fd, EPOLLIN);
+}
+
+static void vitastor_c_write_handler(void *opaque)
+{
+    vitastor_qemu_fd_t *data = (vitastor_qemu_fd_t *)opaque;
+    data->callback(data->fd, EPOLLOUT);
+}
+
+vitastor_c *vitastor_c_create_qemu(QEMUSetFDHandler *aio_set_fd_handler, void *aio_context,
+    const char *config_path, const char *etcd_host, const char *etcd_prefix,
+    int use_rdma, const char *rdma_device, int rdma_port_num, int rdma_gid_index, int rdma_mtu, int log_level)
+{
+    json11::Json cfg_json = vitastor_c_common_config(
+        config_path, etcd_host, etcd_prefix, use_rdma,
+        rdma_device, rdma_port_num, rdma_gid_index, rdma_mtu, log_level
+    );
+    vitastor_c *self = new vitastor_c;
+    self->aio_set_fd_handler = aio_set_fd_handler;
+    self->aio_ctx = aio_context;
+    self->tfd = new timerfd_manager_t([self](int fd, bool wr, std::function<void(int, int)> callback)
+    {
+        if (callback != NULL)
+        {
+            self->handlers[fd] = { .fd = fd, .callback = callback };
+            self->aio_set_fd_handler(self->aio_ctx, fd, false,
+                vitastor_c_read_handler, wr ? vitastor_c_write_handler : NULL, NULL, &self->handlers[fd]);
+        }
+        else
+        {
+            self->handlers.erase(fd);
+            self->aio_set_fd_handler(self->aio_ctx, fd, false, NULL, NULL, NULL, NULL);
+        }
+    });
+    self->cli = new cluster_client_t(NULL, self->tfd, cfg_json);
+    return self;
+}
+
+vitastor_c *vitastor_c_create_uring(const char *config_path, const char *etcd_host, const char *etcd_prefix,
+    int use_rdma, const char *rdma_device, int rdma_port_num, int rdma_gid_index, int rdma_mtu, int log_level)
+{
+    json11::Json cfg_json = vitastor_c_common_config(
+        config_path, etcd_host, etcd_prefix, use_rdma,
+        rdma_device, rdma_port_num, rdma_gid_index, rdma_mtu, log_level
+    );
+    vitastor_c *self = new vitastor_c;
+    self->ringloop = new ring_loop_t(512);
+    self->epmgr = new epoll_manager_t(self->ringloop);
+    self->cli = new cluster_client_t(self->ringloop, self->epmgr->tfd, cfg_json);
+    return self;
+}
+
+vitastor_c *vitastor_c_create_uring_json(const char **options, int options_len)
+{
+    json11::Json::object cfg;
+    for (int i = 0; i < options_len-1; i += 2)
+    {
+        cfg[options[i]] = std::string(options[i+1]);
+    }
+    json11::Json cfg_json(cfg);
+    vitastor_c *self = new vitastor_c;
+    self->ringloop = new ring_loop_t(512);
+    self->epmgr = new epoll_manager_t(self->ringloop);
+    self->cli = new cluster_client_t(self->ringloop, self->epmgr->tfd, cfg_json);
+    return self;
+}
+
+void vitastor_c_destroy(vitastor_c *client)
+{
+    delete client->cli;
+    if (client->epmgr)
+        delete client->epmgr;
+    else
+        delete client->tfd;
+    if (client->ringloop)
+        delete client->ringloop;
+    delete client;
+}
+
+int vitastor_c_is_ready(vitastor_c *client)
+{
+    return client->cli->is_ready();
+}
+
+void vitastor_c_uring_wait_ready(vitastor_c *client)
+{
+    while (!client->cli->is_ready())
+    {
+        client->ringloop->loop();
+        if (client->cli->is_ready())
+            break;
+        client->ringloop->wait();
+    }
+}
+
+void vitastor_c_uring_handle_events(vitastor_c *client)
+{
+    client->ringloop->loop();
+}
+
+void vitastor_c_uring_wait_events(vitastor_c *client)
+{
+    client->ringloop->wait();
+}
+
+void vitastor_c_read(vitastor_c *client, uint64_t inode, uint64_t offset, uint64_t len,
+    struct iovec *iov, int iovcnt, VitastorReadHandler cb, void *opaque)
+{
+    cluster_op_t *op = new cluster_op_t;
+    op->opcode = OSD_OP_READ;
+    op->inode = inode;
+    op->offset = offset;
+    op->len = len;
+    for (int i = 0; i < iovcnt; i++)
+    {
+        op->iov.push_back(iov[i].iov_base, iov[i].iov_len);
+    }
+    op->callback = [cb, opaque](cluster_op_t *op)
+    {
+        cb(opaque, op->retval, op->version);
+        delete op;
+    };
+    client->cli->execute(op);
+}
+
+void vitastor_c_write(vitastor_c *client, uint64_t inode, uint64_t offset, uint64_t len, uint64_t check_version,
+    struct iovec *iov, int iovcnt, VitastorIOHandler cb, void *opaque)
+{
+    cluster_op_t *op = new cluster_op_t;
+    op->opcode = OSD_OP_WRITE;
+    op->inode = inode;
+    op->offset = offset;
+    op->len = len;
+    op->version = check_version;
+    for (int i = 0; i < iovcnt; i++)
+    {
+        op->iov.push_back(iov[i].iov_base, iov[i].iov_len);
+    }
+    op->callback = [cb, opaque](cluster_op_t *op)
+    {
+        cb(opaque, op->retval);
+        delete op;
+    };
+    client->cli->execute(op);
+}
+
+void vitastor_c_sync(vitastor_c *client, VitastorIOHandler cb, void *opaque)
+{
+    cluster_op_t *op = new cluster_op_t;
+    op->opcode = OSD_OP_SYNC;
+    op->callback = [cb, opaque](cluster_op_t *op)
+    {
+        cb(opaque, op->retval);
+        delete op;
+    };
+    client->cli->execute(op);
+}
+
+void vitastor_c_watch_inode(vitastor_c *client, char *image, VitastorIOHandler cb, void *opaque)
+{
+    client->cli->on_ready([=]()
+    {
+        auto watch = client->cli->st_cli.watch_inode(std::string(image));
+        cb(opaque, (long)watch);
+    });
+}
+
+void vitastor_c_close_watch(vitastor_c *client, void *handle)
+{
+    client->cli->st_cli.close_watch((inode_watch_t*)handle);
+}
+
+uint64_t vitastor_c_inode_get_size(void *handle)
+{
+    inode_watch_t *watch = (inode_watch_t*)handle;
+    return watch->cfg.size;
+}
+
+uint64_t vitastor_c_inode_get_num(void *handle)
+{
+    inode_watch_t *watch = (inode_watch_t*)handle;
+    return watch->cfg.num;
+}
+
+int vitastor_c_inode_get_readonly(void *handle)
+{
+    inode_watch_t *watch = (inode_watch_t*)handle;
+    return watch->cfg.readonly;
+}
+
+}

src/vitastor_c.h

@@ -0,0 +1,55 @@
+// Copyright (c) Vitaliy Filippov, 2019+
+// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
+
+// Simplified C client library for QEMU, fio and other external drivers
+
+#ifndef VITASTOR_QEMU_PROXY_H
+#define VITASTOR_QEMU_PROXY_H
+
+#ifndef POOL_ID_BITS
+#define POOL_ID_BITS 16
+#endif
+#include <stdint.h>
+#include <sys/uio.h>
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+struct vitastor_c;
+typedef struct vitastor_c vitastor_c;
+
+typedef void VitastorReadHandler(void *opaque, long retval, uint64_t version);
+typedef void VitastorIOHandler(void *opaque, long retval);
+
+// QEMU
+typedef void IOHandler(void *opaque);
+typedef void QEMUSetFDHandler(void *ctx, int fd, int is_external, IOHandler *fd_read, IOHandler *fd_write, void *poll_fn, void *opaque);
+
+vitastor_c *vitastor_c_create_qemu(QEMUSetFDHandler *aio_set_fd_handler, void *aio_context,
+    const char *config_path, const char *etcd_host, const char *etcd_prefix,
+    int use_rdma, const char *rdma_device, int rdma_port_num, int rdma_gid_index, int rdma_mtu, int log_level);
+vitastor_c *vitastor_c_create_uring(const char *config_path, const char *etcd_host, const char *etcd_prefix,
+    int use_rdma, const char *rdma_device, int rdma_port_num, int rdma_gid_index, int rdma_mtu, int log_level);
+vitastor_c *vitastor_c_create_uring_json(const char **options, int options_len);
+void vitastor_c_destroy(vitastor_c *client);
+int vitastor_c_is_ready(vitastor_c *client);
+void vitastor_c_uring_wait_ready(vitastor_c *client);
+void vitastor_c_uring_handle_events(vitastor_c *client);
+void vitastor_c_uring_wait_events(vitastor_c *client);
+void vitastor_c_read(vitastor_c *client, uint64_t inode, uint64_t offset, uint64_t len,
+    struct iovec *iov, int iovcnt, VitastorReadHandler cb, void *opaque);
+void vitastor_c_write(vitastor_c *client, uint64_t inode, uint64_t offset, uint64_t len, uint64_t check_version,
+    struct iovec *iov, int iovcnt, VitastorIOHandler cb, void *opaque);
+void vitastor_c_sync(vitastor_c *client, VitastorIOHandler cb, void *opaque);
+void vitastor_c_watch_inode(vitastor_c *client, char *image, VitastorIOHandler cb, void *opaque);
+void vitastor_c_close_watch(vitastor_c *client, void *handle);
+uint64_t vitastor_c_inode_get_size(void *handle);
+uint64_t vitastor_c_inode_get_num(void *handle);
+int vitastor_c_inode_get_readonly(void *handle);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif

tests/run_3osds.sh

@@ -0,0 +1,44 @@
+#!/bin/bash -ex
+
+. `dirname $0`/common.sh
+
+OSD_SIZE=${OSD_SIZE:-1024}
+PG_COUNT=${PG_COUNT:-1}
+
+dd if=/dev/zero of=./testdata/test_osd1.bin bs=1024 count=1 seek=$((OSD_SIZE*1024-1))
+dd if=/dev/zero of=./testdata/test_osd2.bin bs=1024 count=1 seek=$((OSD_SIZE*1024-1))
+dd if=/dev/zero of=./testdata/test_osd3.bin bs=1024 count=1 seek=$((OSD_SIZE*1024-1))
+
+build/src/vitastor-osd --osd_num 1 --bind_address 127.0.0.1 $OSD_ARGS --etcd_address $ETCD_URL $(node mon/simple-offsets.js --format options --device ./testdata/test_osd1.bin 2>/dev/null) &>./testdata/osd1.log &
+OSD1_PID=$!
+build/src/vitastor-osd --osd_num 2 --bind_address 127.0.0.1 $OSD_ARGS --etcd_address $ETCD_URL $(node mon/simple-offsets.js --format options --device ./testdata/test_osd2.bin 2>/dev/null) &>./testdata/osd2.log &
+OSD2_PID=$!
+build/src/vitastor-osd --osd_num 3 --bind_address 127.0.0.1 $OSD_ARGS --etcd_address $ETCD_URL $(node mon/simple-offsets.js --format options --device ./testdata/test_osd3.bin 2>/dev/null) &>./testdata/osd3.log &
+OSD3_PID=$!
+
+cd mon
+npm install
+cd ..
+node mon/mon-main.js --etcd_url http://$ETCD_URL --etcd_prefix "/vitastor" &>./testdata/mon.log &
+MON_PID=$!
+
+if [ -n "$GLOBAL_CONF" ]; then
+    $ETCDCTL put /vitastor/config/global "$GLOBAL_CONF"
+fi
+
+$ETCDCTL put /vitastor/config/pools '{"1":{"name":"testpool","scheme":"xor","pg_size":3,"pg_minsize":2,"parity_chunks":1,"pg_count":'$PG_COUNT',"failure_domain":"osd"}}'
+
+sleep 2
+
+if ! ($ETCDCTL get /vitastor/config/pgs --print-value-only | jq -s -e '(. | length) != 0 and ([ .[0].items["1"][] | select((.osd_set | sort) == ["1","2","3"]) ] | length) == '$PG_COUNT); then
+    format_error "FAILED: $PG_COUNT PG(s) NOT CONFIGURED"
+fi
+
+if ! ($ETCDCTL get /vitastor/pg/state/1/ --prefix --print-value-only | jq -s -e '[ .[] | select(.state == ["active"]) ] | length == '$PG_COUNT); then
+    format_error "FAILED: $PG_COUNT PG(s) NOT UP"
+fi
+
+if ! cmp build/src/block-vitastor.so /usr/lib/x86_64-linux-gnu/qemu/block-vitastor.so; then
+    sudo rm -f /usr/lib/x86_64-linux-gnu/qemu/block-vitastor.so
+    sudo ln -s "$(realpath .)/build/src/block-vitastor.so" /usr/lib/x86_64-linux-gnu/qemu/block-vitastor.so
+fi

tests/test_cas.sh

@@ -0,0 +1,7 @@
+#!/bin/bash -ex
+
+. `dirname $0`/run_3osds.sh
+
+build/src/test_cas --pool_id 1 --inode_id 1 --etcd_address $ETCD_URL
+
+format_green OK

tests/test_rm.sh

@@ -0,0 +1,14 @@
+#!/bin/bash -ex
+
+PG_COUNT=16
+. `dirname $0`/run_3osds.sh
+
+LD_PRELOAD=libasan.so.5 \
+    fio -thread -name=test -ioengine=build/src/libfio_vitastor.so -bs=4M -direct=1 -iodepth=1 \
+        -end_fsync=1 -fsync=1 -rw=write -etcd=$ETCD_URL -pool=1 -inode=1 -size=128M -cluster_log_level=10
+
+$ETCDCTL get --prefix '/vitastor/pg/state'
+
+build/src/vitastor-cli rm-data --etcd_address $ETCD_URL --pool 1 --inode 1
+
+format_green OK

tests/test_snapshot.sh

@@ -1,57 +1,24 @@
 #!/bin/bash -ex
 
-. `dirname $0`/common.sh
-
-dd if=/dev/zero of=./testdata/test_osd1.bin bs=1024 count=1 seek=$((1024*1024-1))
-dd if=/dev/zero of=./testdata/test_osd2.bin bs=1024 count=1 seek=$((1024*1024-1))
-dd if=/dev/zero of=./testdata/test_osd3.bin bs=1024 count=1 seek=$((1024*1024-1))
-
-build/src/vitastor-osd --osd_num 1 --bind_address 127.0.0.1 --etcd_address $ETCD_URL $(node mon/simple-offsets.js --format options --device ./testdata/test_osd1.bin 2>/dev/null) &>./testdata/osd1.log &
-OSD1_PID=$!
-build/src/vitastor-osd --osd_num 2 --bind_address 127.0.0.1 --etcd_address $ETCD_URL $(node mon/simple-offsets.js --format options --device ./testdata/test_osd2.bin 2>/dev/null) &>./testdata/osd2.log &
-OSD2_PID=$!
-build/src/vitastor-osd --osd_num 3 --bind_address 127.0.0.1 --etcd_address $ETCD_URL $(node mon/simple-offsets.js --format options --device ./testdata/test_osd3.bin 2>/dev/null) &>./testdata/osd3.log &
-OSD3_PID=$!
-
-cd mon
-npm install
-cd ..
-node mon/mon-main.js --etcd_url http://$ETCD_URL --etcd_prefix "/vitastor" &>./testdata/mon.log &
-MON_PID=$!
-
-$ETCDCTL put /vitastor/config/pools '{"1":{"name":"testpool","scheme":"xor","pg_size":3,"pg_minsize":2,"parity_chunks":1,"pg_count":1,"failure_domain":"osd"}}'
-
-sleep 2
-
-if ! ($ETCDCTL get /vitastor/config/pgs --print-value-only | jq -s -e '(. | length) != 0 and (.[0].items["1"]["1"].osd_set | sort) == ["1","2","3"]'); then
-    format_error "FAILED: 1 PG NOT CONFIGURED"
-fi
-
-if ! ($ETCDCTL get /vitastor/pg/state/1/1 --print-value-only | jq -s -e '(. | length) != 0 and .[0].state == ["active"]'); then
-    format_error "FAILED: 1 PG NOT UP"
-fi
-
-if ! cmp build/src/block-vitastor.so /usr/lib/x86_64-linux-gnu/qemu/block-vitastor.so; then
-    sudo rm -f /usr/lib/x86_64-linux-gnu/qemu/block-vitastor.so
-    sudo ln -s "$(realpath .)/build/src/block-vitastor.so" /usr/lib/x86_64-linux-gnu/qemu/block-vitastor.so
-fi
+. `dirname $0`/run_3osds.sh
 
 # Test basic write and snapshot
 
 $ETCDCTL put /vitastor/config/inode/1/2 '{"name":"testimg","size":'$((32*1024*1024))'}'
 
-LD_PRELOAD=libasan.so.5 \
+LD_PRELOAD="libasan.so.5 build/src/libfio_vitastor.so" \
     fio -thread -name=test -ioengine=build/src/libfio_vitastor.so -bs=4M -direct=1 -iodepth=1 -fsync=1 -rw=write \
         -etcd=$ETCD_URL -pool=1 -inode=2 -size=32M -cluster_log_level=10
 
 $ETCDCTL put /vitastor/config/inode/1/2 '{"name":"testimg@0","size":'$((32*1024*1024))'}'
 $ETCDCTL put /vitastor/config/inode/1/3 '{"parent_id":2,"name":"testimg","size":'$((32*1024*1024))'}'
 
-LD_PRELOAD=libasan.so.5 \
+# Preload build/src/libfio_vitastor.so so libasan detects all symbols
+LD_PRELOAD="libasan.so.5 build/src/libfio_vitastor.so" \
     fio -thread -name=test -ioengine=build/src/libfio_vitastor.so -bs=4k -direct=1 -iodepth=1 -fsync=32 -buffer_pattern=0xdeadface \
         -rw=randwrite -etcd=$ETCD_URL -image=testimg -number_ios=1024
 
-LD_PRELOAD=libasan.so.5 \
+LD_PRELOAD="libasan.so.5 build/src/libfio_vitastor.so" \
     fio -thread -name=test -ioengine=build/src/libfio_vitastor.so -bs=4M -direct=1 -iodepth=1 -rw=read -etcd=$ETCD_URL -pool=1 -inode=3 -size=32M
 
 qemu-img convert -S 4096 -p \
@@ -72,4 +39,16 @@ node mon/merge.js ./testdata/layer0.bin ./testdata/layer1.bin ./testdata/check.b
 
 cmp ./testdata/merged.bin ./testdata/check.bin
 
+# Test merge
+
+$ETCDCTL put /vitastor/config/inode/1/3 '{"parent_id":2,"name":"testimg","size":'$((32*1024*1024))'}'
+
+build/src/vitastor-cli rm --etcd_address $ETCD_URL testimg@0
+
+qemu-img convert -S 4096 -p \
+    -f raw "vitastor:etcd_host=127.0.0.1\:$ETCD_PORT/v3:image=testimg" \
+    -O raw ./testdata/merged-by-tool.bin
+
+cmp ./testdata/merged.bin ./testdata/merged-by-tool.bin
+
 format_green OK