Radically simplify create-pool pg_size check

Could probably lead to PGs hung in peering states on OSD restart in EC pools,
fixable by primary OSD restart

.gitea/workflows/test.yml

@@ -288,6 +288,24 @@ jobs:
           echo ""
         done
 
+  test_create_halfhost:
+    runs-on: ubuntu-latest
+    needs: build
+    container: ${{env.TEST_IMAGE}}:${{github.sha}}
+    steps:
+    - name: Run test
+      id: test
+      timeout-minutes: 3
+      run: /root/vitastor/tests/test_create_halfhost.sh
+    - name: Print logs
+      if: always() && steps.test.outcome == 'failure'
+      run: |
+        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
+          echo "-------- $i --------"
+          cat $i
+          echo ""
+        done
+
   test_failure_domain:
     runs-on: ubuntu-latest
     needs: build

docs/config/network.en.md

@@ -68,11 +68,17 @@ but they are not connected to the cluster.
 - Type: string
 
 RDMA device name to use for Vitastor OSD communications (for example,
-"rocep5s0f0"). Now Vitastor supports all adapters, even ones without
+"rocep5s0f0"). If not specified, Vitastor will try to find an RoCE
-ODP support, like Mellanox ConnectX-3 and non-Mellanox cards.
+device matching [osd_network](osd.en.md#osd_network), preferring RoCEv2,
+or choose the first available RDMA device if no RoCE devices are
+found or if `osd_network` is not specified. Auto-selection is also
+unsupported with old libibverbs < v32, like in Debian 10 Buster or
+CentOS 7.
 
-Versions up to Vitastor 1.2.0 required ODP which is only present in
+Vitastor supports all adapters, even ones without ODP support, like
-Mellanox ConnectX >= 4. See also [rdma_odp](#rdma_odp).
+Mellanox ConnectX-3 and non-Mellanox cards. Versions up to Vitastor
+1.2.0 required ODP which is only present in Mellanox ConnectX >= 4.
+See also [rdma_odp](#rdma_odp).
 
 Run `ibv_devinfo -v` as root to list available RDMA devices and their
 features.
@@ -95,15 +101,17 @@ your device has.
 ## rdma_gid_index
 
 - Type: integer
-- Default: 0
 
 Global address identifier index of the RDMA device to use. Different GID
 indexes may correspond to different protocols like RoCEv1, RoCEv2 and iWARP.
 Search for "GID" in `ibv_devinfo -v` output to determine which GID index
 you need.
 
-**IMPORTANT:** If you want to use RoCEv2 (as recommended) then the correct
+If not specified, Vitastor will try to auto-select a RoCEv2 IPv4 GID, then
-rdma_gid_index is usually 1 (IPv6) or 3 (IPv4).
+RoCEv2 IPv6 GID, then RoCEv1 IPv4 GID, then RoCEv1 IPv6 GID, then IB GID.
+GID auto-selection is unsupported with libibverbs < v32.
+
+A correct rdma_gid_index for RoCEv2 is usually 1 (IPv6) or 3 (IPv4).
 
 ## rdma_mtu
 

docs/config/network.ru.md

@@ -71,12 +71,17 @@ RDMA может быть нужно только если у клиентов е
 - Тип: строка
 
 Название RDMA-устройства для связи с Vitastor OSD (например, "rocep5s0f0").
-Сейчас Vitastor поддерживает все модели адаптеров, включая те, у которых
+Если не указано, Vitastor попробует найти RoCE-устройство, соответствующее
-нет поддержки ODP, то есть вы можете использовать RDMA с ConnectX-3 и
+[osd_network](osd.en.md#osd_network), предпочитая RoCEv2, или выбрать первое
-картами производства не Mellanox.
+попавшееся RDMA-устройство, если RoCE-устройств нет или если сеть `osd_network`
+не задана. Также автовыбор не поддерживается со старыми версиями библиотеки
+libibverbs < v32, например в Debian 10 Buster или CentOS 7.
 
-Версии Vitastor до 1.2.0 включительно требовали ODP, который есть только
+Vitastor поддерживает все модели адаптеров, включая те, у которых
-на Mellanox ConnectX 4 и более новых. См. также [rdma_odp](#rdma_odp).
+нет поддержки ODP, то есть вы можете использовать RDMA с ConnectX-3 и
+картами производства не Mellanox. Версии Vitastor до 1.2.0 включительно
+требовали ODP, который есть только на Mellanox ConnectX 4 и более новых.
+См. также [rdma_odp](#rdma_odp).
 
 Запустите `ibv_devinfo -v` от имени суперпользователя, чтобы посмотреть
 список доступных RDMA-устройств, их параметры и возможности.
@@ -101,15 +106,18 @@ Control) и ECN (Explicit Congestion Notification).
 ## rdma_gid_index
 
 - Тип: целое число
-- Значение по умолчанию: 0
 
 Номер глобального идентификатора адреса RDMA-устройства, который следует
 использовать. Разным gid_index могут соответствовать разные протоколы связи:
 RoCEv1, RoCEv2, iWARP. Чтобы понять, какой нужен вам - смотрите строчки со
 словом "GID" в выводе команды `ibv_devinfo -v`.
 
-**ВАЖНО:** Если вы хотите использовать RoCEv2 (как мы и рекомендуем), то
+Если не указан, Vitastor попробует автоматически выбрать сначала GID,
-правильный rdma_gid_index, как правило, 1 (IPv6) или 3 (IPv4).
+соответствующий RoCEv2 IPv4, потом RoCEv2 IPv6, потом RoCEv1 IPv4, потом
+RoCEv1 IPv6, потом IB. Авто-выбор GID не поддерживается со старыми версиями
+libibverbs < v32.
+
+Правильный rdma_gid_index для RoCEv2, как правило, 1 (IPv6) или 3 (IPv4).
 
 ## rdma_mtu
 

docs/config/src/network.yml

@@ -48,11 +48,17 @@
   type: string
   info: |
     RDMA device name to use for Vitastor OSD communications (for example,
-    "rocep5s0f0"). Now Vitastor supports all adapters, even ones without
+    "rocep5s0f0"). If not specified, Vitastor will try to find an RoCE
-    ODP support, like Mellanox ConnectX-3 and non-Mellanox cards.
+    device matching [osd_network](osd.en.md#osd_network), preferring RoCEv2,
+    or choose the first available RDMA device if no RoCE devices are
+    found or if `osd_network` is not specified. Auto-selection is also
+    unsupported with old libibverbs < v32, like in Debian 10 Buster or
+    CentOS 7.
 
-    Versions up to Vitastor 1.2.0 required ODP which is only present in
+    Vitastor supports all adapters, even ones without ODP support, like
-    Mellanox ConnectX >= 4. See also [rdma_odp](#rdma_odp).
+    Mellanox ConnectX-3 and non-Mellanox cards. Versions up to Vitastor
+    1.2.0 required ODP which is only present in Mellanox ConnectX >= 4.
+    See also [rdma_odp](#rdma_odp).
 
     Run `ibv_devinfo -v` as root to list available RDMA devices and their
     features.
@@ -64,12 +70,17 @@
     PFC (Priority Flow Control) and ECN (Explicit Congestion Notification).
   info_ru: |
     Название RDMA-устройства для связи с Vitastor OSD (например, "rocep5s0f0").
-    Сейчас Vitastor поддерживает все модели адаптеров, включая те, у которых
+    Если не указано, Vitastor попробует найти RoCE-устройство, соответствующее
-    нет поддержки ODP, то есть вы можете использовать RDMA с ConnectX-3 и
+    [osd_network](osd.en.md#osd_network), предпочитая RoCEv2, или выбрать первое
-    картами производства не Mellanox.
+    попавшееся RDMA-устройство, если RoCE-устройств нет или если сеть `osd_network`
+    не задана. Также автовыбор не поддерживается со старыми версиями библиотеки
+    libibverbs < v32, например в Debian 10 Buster или CentOS 7.
 
-    Версии Vitastor до 1.2.0 включительно требовали ODP, который есть только
+    Vitastor поддерживает все модели адаптеров, включая те, у которых
-    на Mellanox ConnectX 4 и более новых. См. также [rdma_odp](#rdma_odp).
+    нет поддержки ODP, то есть вы можете использовать RDMA с ConnectX-3 и
+    картами производства не Mellanox. Версии Vitastor до 1.2.0 включительно
+    требовали ODP, который есть только на Mellanox ConnectX 4 и более новых.
+    См. также [rdma_odp](#rdma_odp).
 
     Запустите `ibv_devinfo -v` от имени суперпользователя, чтобы посмотреть
     список доступных RDMA-устройств, их параметры и возможности.
@@ -94,23 +105,29 @@
     `ibv_devinfo -v`.
 - name: rdma_gid_index
   type: int
-  default: 0
   info: |
     Global address identifier index of the RDMA device to use. Different GID
     indexes may correspond to different protocols like RoCEv1, RoCEv2 and iWARP.
     Search for "GID" in `ibv_devinfo -v` output to determine which GID index
     you need.
 
-    **IMPORTANT:** If you want to use RoCEv2 (as recommended) then the correct
+    If not specified, Vitastor will try to auto-select a RoCEv2 IPv4 GID, then
-    rdma_gid_index is usually 1 (IPv6) or 3 (IPv4).
+    RoCEv2 IPv6 GID, then RoCEv1 IPv4 GID, then RoCEv1 IPv6 GID, then IB GID.
+    GID auto-selection is unsupported with libibverbs < v32.
+
+    A correct rdma_gid_index for RoCEv2 is usually 1 (IPv6) or 3 (IPv4).
   info_ru: |
     Номер глобального идентификатора адреса RDMA-устройства, который следует
     использовать. Разным gid_index могут соответствовать разные протоколы связи:
     RoCEv1, RoCEv2, iWARP. Чтобы понять, какой нужен вам - смотрите строчки со
     словом "GID" в выводе команды `ibv_devinfo -v`.
 
-    **ВАЖНО:** Если вы хотите использовать RoCEv2 (как мы и рекомендуем), то
+    Если не указан, Vitastor попробует автоматически выбрать сначала GID,
-    правильный rdma_gid_index, как правило, 1 (IPv6) или 3 (IPv4).
+    соответствующий RoCEv2 IPv4, потом RoCEv2 IPv6, потом RoCEv1 IPv4, потом
+    RoCEv1 IPv6, потом IB. Авто-выбор GID не поддерживается со старыми версиями
+    libibverbs < v32.
+
+    Правильный rdma_gid_index для RoCEv2, как правило, 1 (IPv6) или 3 (IPv4).
 - name: rdma_mtu
   type: int
   default: 4096

docs/intro/architecture.en.md

@@ -6,7 +6,12 @@
 
 # Architecture
 
+- [Server-side components](#server-side-components)
 - [Basic concepts](#basic-concepts)
+- [Client-side components](#client-side-components)
+- [Additional utilities](#additional-utilities)
+- [Overall read/write process](#overall-read-write-process)
+  - [Nuances of request handling](#nuances-of-request-handling)
 - [Similarities to Ceph](#similarities-to-ceph)
 - [Differences from Ceph](#differences-from-ceph)
 - [Implementation Principles](#implementation-principles)
@@ -39,10 +44,10 @@
 
 ## Client-side components
 
-- **Client library** incapsulates client I/O logic. Client library connects to etcd and to all OSDs,
+- **Client library** encapsulates client I/O logic. Client library connects to etcd and to all OSDs,
   receives cluster state from etcd, sends read and write requests directly to all OSDs. Due
   to the symmetric distributed architecture, all data blocks (each 128 KB by default) are placed
-  to different OSDs, but clients always knows where each data block is stored and connects directly
+  to different OSDs, but clients always know where each data block is stored and connect directly
   to the right OSD.
 
 All other client-side components are based on the client library:
@@ -77,8 +82,8 @@ All other client-side components are based on the client library:
 
 ## Additional utilities
 
-- **vitastor-disk** — утилита для разметки дисков под Vitastor OSD. С её помощью можно
+- **vitastor-disk** — a Vitastor OSD disk management tool. You can create, remove,
-  создавать, удалять, менять размеры или перемещать разделы OSD.
+  resize and move OSD partitions with it.
 
 ## Overall read/write process
 

docs/usage/admin.en.md

@@ -171,7 +171,14 @@ to make them use the new version of the client library.
 
 ### 1.7.x to 1.8.0
 
-After upgrading version <= 1.7.x to version >= 1.8.0, BUT <= 1.9.0: restart all clients
+It's recommended to upgrade from version <= 1.7.x to version >= 1.8.0 with full downtime,
+i.e. you should first stop clients and then the cluster (OSDs and monitor), because 1.8.0
+includes a fix for etcd event stream inconsistency which could lead to "incomplete" objects
+appearing in EC pools, and in rare cases, probably, even to data corruption during mass OSD
+restarts. It doesn't mean that you WILL hit this problem if you upgrade without full downtime,
+but it's better to secure yourself against it.
+
+Also, if you upgrade version from <= 1.7.x to version >= 1.8.0, BUT <= 1.9.0: restart all clients
 (VMs and so on), otherwise they will hang when monitor clears old PG configuration key,
 which happens 24 hours after upgrade.
 

docs/usage/admin.ru.md

@@ -168,7 +168,14 @@ done
 
 ### 1.7.x -> 1.8.0
 
-После обновления с версий <= 1.7.x до версий >= 1.8.0, НО <= 1.9.0: перезапустите всех
+Обновляться с версий <= 1.7.x до версий >= 1.8.0 рекомендуется с полной остановкой
+сначала клиентов, а затем кластера, так как в 1.8.0 исправлена проблема (неконсистентность
+потоков событий от etcd), способная приводить к появлению incomplete объектов в EC-пулах
+и, хоть и редко, но даже к повреждению данных при массовых перезапусках OSD. Если вы
+обновляетесь без полной остановки - это не значит, что вы обязательно столкнётесь с этой
+проблемой, но лучше подстраховаться.
+
+Также, если вы обновляетесь с версии <= 1.7.x до версии >= 1.8.0, НО <= 1.9.0: перезапустите всех
 клиентов (процессы виртуальных машин можно перезапустить путём миграции на другой сервер),
 иначе они зависнут, когда монитор удалит старый ключ конфигурации PG, что происходит через
 24 часа после обновления.

mon/etcd_adapter.js

@@ -232,6 +232,7 @@ class EtcdAdapter
     async become_master()
     {
         const state = { ...this.mon.get_mon_state(), id: ''+this.mon.etcd_lease_id };
+        console.log('Waiting to become master');
         // eslint-disable-next-line no-constant-condition
         while (1)
         {
@@ -243,7 +244,6 @@ class EtcdAdapter
             {
                 break;
             }
-            console.log('Waiting to become master');
             await new Promise(ok => setTimeout(ok, this.mon.config.etcd_start_timeout));
         }
         console.log('Became master');

mon/etcd_schema.js

@@ -56,6 +56,7 @@ const etcd_tree = {
             osd_out_time: 600, // seconds. min: 0
             placement_levels: { datacenter: 1, rack: 2, host: 3, osd: 4, ... },
             use_old_pg_combinator: false,
+            osd_backfillfull_ratio: 0.99,
             // client and osd
             tcp_header_buffer_size: 65536,
             use_sync_send_recv: false,

mon/mon.js

@@ -74,6 +74,7 @@ class Mon
         this.state = JSON.parse(JSON.stringify(etcd_tree));
         this.prev_stats = { osd_stats: {}, osd_diff: {} };
         this.recheck_pgs_active = false;
+        this.updating_total_stats = false;
         this.watcher_active = false;
         this.old_pg_config = false;
         this.old_pg_stats_seen = false;
@@ -658,7 +659,13 @@ class Mon
                 this.etcd_watch_revision, pool_id, up_osds, osd_tree, real_prev_pgs, pool_res.pgs, pg_history);
         }
         new_pg_config.hash = tree_hash;
-        return await this.save_pg_config(new_pg_config, etcd_request);
+        const { backfillfull_pools } = sum_object_counts({ ...this.state, pg: { ...this.state.pg, config: new_pg_config } }, this.config);
+        new_pg_config.backfillfull_pools = backfillfull_pools.length ? backfillfull_pools : undefined;
+        if (!await this.save_pg_config(new_pg_config, etcd_request))
+        {
+            return false;
+        }
+        return true;
     }
 
     async save_pg_config(new_pg_config, etcd_request = { compare: [], success: [] })
@@ -730,7 +737,7 @@ class Mon
     async update_total_stats()
     {
         const txn = [];
-        const { object_counts, object_bytes } = sum_object_counts(this.state, this.config);
+        const { object_counts, object_bytes, backfillfull_pools } = sum_object_counts(this.state, this.config);
         let stats = sum_op_stats(this.state.osd, this.prev_stats);
         let { inode_stats, seen_pools } = sum_inode_stats(this.state, this.prev_stats);
         stats.object_counts = object_counts;
@@ -783,6 +790,16 @@ class Mon
         {
             await this.etcd.etcd_call('/kv/txn', { success: txn }, this.config.etcd_mon_timeout, 0);
         }
+        if (!this.recheck_pgs_active &&
+            backfillfull_pools.join(',') != ((this.state.pg.config||{}).no_rebalance_pools||[]).join(','))
+        {
+            console.log(
+                (backfillfull_pools.length ? 'Pool(s) '+backfillfull_pools.join(', ') : 'No pools')+
+                ' are backfillfull, applying rebalance configuration'
+            );
+            const new_pg_config = { ...this.state.pg.config, backfillfull_pools: backfillfull_pools.length ? backfillfull_pools : undefined };
+            await this.save_pg_config(new_pg_config);
+        }
     }
 
     schedule_update_stats()
@@ -794,7 +811,21 @@ class Mon
         this.stats_timer = setTimeout(() =>
         {
             this.stats_timer = null;
-            this.update_total_stats().catch(console.error);
+            if (this.updating_total_stats)
+            {
+                this.schedule_update_stats();
+                return;
+            }
+            this.updating_total_stats = true;
+            try
+            {
+                this.update_total_stats().catch(console.error);
+            }
+            catch (e)
+            {
+                console.error(e);
+            }
+            this.updating_total_stats = false;
         }, this.config.mon_stats_timeout);
     }
 

mon/stats.js

@@ -109,6 +109,8 @@ function sum_object_counts(state, global_config)
             pgstats[pool_id] = { ...(state.pg.stats[pool_id] || {}), ...(pgstats[pool_id] || {}) };
         }
     }
+    const pool_per_osd = {};
+    const clean_per_osd = {};
     for (const pool_id in pgstats)
     {
         let object_size = 0;
@@ -143,10 +145,38 @@ function sum_object_counts(state, global_config)
                         object_bytes[k] += BigInt(st[k+'_count']) * object_size;
                     }
                 }
+                if (st.object_count)
+                {
+                    for (const pg_osd in (((state.pg.config.items||{})[pool_id]||{})[pg_num]||{}).osd_set||[])
+                    {
+                        if (!(pg_osd in clean_per_osd))
+                        {
+                            clean_per_osd[pg_osd] = 0n;
+                        }
+                        clean_per_osd[pg_osd] += BigInt(st.object_count);
+                        pool_per_osd[pg_osd] = pool_per_osd[pg_osd]||{};
+                        pool_per_osd[pg_osd][pool_id] = true;
+                    }
+                }
             }
         }
     }
-    return { object_counts, object_bytes };
+    // If clean_per_osd[osd] is larger than osd capacity then it will fill up during rebalance
+    let backfillfull_pools = {};
+    for (const osd in clean_per_osd)
+    {
+        const st = state.osd.stats[osd];
+        if (st && st.size && st.data_block_size && (BigInt(st.size)/BigInt(st.data_block_size)*
+            BigInt((global_config.osd_backfillfull_ratio||0.99)*1000000)/1000000n) < clean_per_osd[osd])
+        {
+            for (const pool_id in pool_per_osd[osd])
+            {
+                backfillfull_pools[pool_id] = true;
+            }
+        }
+    }
+    backfillfull_pools = Object.keys(backfillfull_pools).sort();
+    return { object_counts, object_bytes, backfillfull_pools };
 }
 
 // sum_inode_stats(this.state, this.prev_stats)

patches/libvirt-8.0-vitastor.diff

@@ -306,12 +306,12 @@ index e5ff653a60..884ecc79ea 100644
 +        etcd = virBufferContentAndReset(&buf);
 +    }
 +
-+    if (virJSONValueObjectCreate(&ret,
++    if (virJSONValueObjectAdd(&ret,
-+                                 "S:etcd-host", etcd,
++                              "S:etcd-host", etcd,
-+                                 "S:etcd-prefix", src->query,
++                              "S:etcd-prefix", src->query,
-+                                 "S:config-path", src->configFile,
++                              "S:config-path", src->configFile,
-+                                 "s:image", src->path,
++                              "s:image", src->path,
-+                                 NULL) < 0)
++                              NULL) < 0)
 +        return NULL;
 +
 +    return ret;

patches/pve-qemu-9.0-vitastor.patch

@@ -0,0 +1,193 @@
+Index: pve-qemu-kvm-9.0.0/block/meson.build
+===================================================================
+--- pve-qemu-kvm-9.0.0.orig/block/meson.build
++++ pve-qemu-kvm-9.0.0/block/meson.build
+@@ -126,6 +126,7 @@ foreach m : [
+   [libnfs, 'nfs', files('nfs.c')],
+   [libssh, 'ssh', files('ssh.c')],
+   [rbd, 'rbd', files('rbd.c')],
++  [vitastor, 'vitastor', files('vitastor.c')],
+ ]
+   if m[0].found()
+     module_ss = ss.source_set()
+Index: pve-qemu-kvm-9.0.0/meson.build
+===================================================================
+--- pve-qemu-kvm-9.0.0.orig/meson.build
++++ pve-qemu-kvm-9.0.0/meson.build
+@@ -1452,6 +1452,26 @@ if not get_option('rbd').auto() or have_
+   endif
+ endif
+ 
++vitastor = not_found
++if not get_option('vitastor').auto() or have_block
++  libvitastor_client = cc.find_library('vitastor_client', has_headers: ['vitastor_c.h'],
++    required: get_option('vitastor'))
++  if libvitastor_client.found()
++    if cc.links('''
++      #include <vitastor_c.h>
++      int main(void) {
++        vitastor_c_create_qemu(0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
++        return 0;
++      }''', dependencies: libvitastor_client)
++      vitastor = declare_dependency(dependencies: libvitastor_client)
++    elif get_option('vitastor').enabled()
++      error('could not link libvitastor_client')
++    else
++      warning('could not link libvitastor_client, disabling')
++    endif
++  endif
++endif
++
+ glusterfs = not_found
+ glusterfs_ftruncate_has_stat = false
+ glusterfs_iocb_has_stat = false
+@@ -2254,6 +2274,7 @@ endif
+ config_host_data.set('CONFIG_OPENGL', opengl.found())
+ config_host_data.set('CONFIG_PLUGIN', get_option('plugins'))
+ config_host_data.set('CONFIG_RBD', rbd.found())
++config_host_data.set('CONFIG_VITASTOR', vitastor.found())
+ config_host_data.set('CONFIG_RDMA', rdma.found())
+ config_host_data.set('CONFIG_RELOCATABLE', get_option('relocatable'))
+ config_host_data.set('CONFIG_SAFESTACK', get_option('safe_stack'))
+@@ -4454,6 +4475,7 @@ summary_info += {'fdt support':       fd
+ summary_info += {'libcap-ng support': libcap_ng}
+ summary_info += {'bpf support':       libbpf}
+ summary_info += {'rbd support':       rbd}
++summary_info += {'vitastor support':  vitastor}
+ summary_info += {'smartcard support': cacard}
+ summary_info += {'U2F support':       u2f}
+ summary_info += {'libusb':            libusb}
+Index: pve-qemu-kvm-9.0.0/meson_options.txt
+===================================================================
+--- pve-qemu-kvm-9.0.0.orig/meson_options.txt
++++ pve-qemu-kvm-9.0.0/meson_options.txt
+@@ -194,6 +194,8 @@ option('lzo', type : 'feature', value :
+        description: 'lzo compression support')
+ option('rbd', type : 'feature', value : 'auto',
+        description: 'Ceph block device driver')
++option('vitastor', type : 'feature', value : 'auto',
++       description: 'Vitastor block device driver')
+ option('opengl', type : 'feature', value : 'auto',
+        description: 'OpenGL support')
+ option('rdma', type : 'feature', value : 'auto',
+Index: pve-qemu-kvm-9.0.0/qapi/block-core.json
+===================================================================
+--- pve-qemu-kvm-9.0.0.orig/qapi/block-core.json
++++ pve-qemu-kvm-9.0.0/qapi/block-core.json
+@@ -3481,7 +3481,7 @@
+             'raw', 'rbd',
+             { 'name': 'replication', 'if': 'CONFIG_REPLICATION' },
+             'pbs',
+-            'ssh', 'throttle', 'vdi', 'vhdx',
++            'ssh', 'throttle', 'vdi', 'vhdx', 'vitastor',
+             { 'name': 'virtio-blk-vfio-pci', 'if': 'CONFIG_BLKIO' },
+             { 'name': 'virtio-blk-vhost-user', 'if': 'CONFIG_BLKIO' },
+             { 'name': 'virtio-blk-vhost-vdpa', 'if': 'CONFIG_BLKIO' },
+@@ -4591,6 +4591,28 @@
+             '*server': ['InetSocketAddressBase'] } }
+ 
+ ##
++# @BlockdevOptionsVitastor:
++#
++# Driver specific block device options for vitastor
++#
++# @image:       Image name
++# @inode:       Inode number
++# @pool:        Pool ID
++# @size:        Desired image size in bytes
++# @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',
++            '*image': 'str',
++            '*config-path': 'str',
++            '*etcd-host': 'str',
++            '*etcd-prefix': 'str' } }
++
++##
+ # @ReplicationMode:
+ #
+ # An enumeration of replication modes.
+@@ -5053,6 +5075,7 @@
+       'throttle':   'BlockdevOptionsThrottle',
+       'vdi':        'BlockdevOptionsGenericFormat',
+       'vhdx':       'BlockdevOptionsGenericFormat',
++      'vitastor':   'BlockdevOptionsVitastor',
+       'virtio-blk-vfio-pci':
+                     { 'type': 'BlockdevOptionsVirtioBlkVfioPci',
+                       'if': 'CONFIG_BLKIO' },
+@@ -5498,6 +5521,20 @@
+             '*encrypt' :        'RbdEncryptionCreateOptions' } }
+ 
+ ##
++# @BlockdevCreateOptionsVitastor:
++#
++# Driver specific image creation options for Vitastor.
++#
++# @location: Where to store the new image file.  This location cannot
++#     point to a snapshot.
++#
++# @size: Size of the virtual disk in bytes
++##
++{ 'struct': 'BlockdevCreateOptionsVitastor',
++  'data': { 'location':         'BlockdevOptionsVitastor',
++            'size':             'size' } }
++
++##
+ # @BlockdevVmdkSubformat:
+ #
+ # Subformat options for VMDK images
+@@ -5719,6 +5753,7 @@
+       'ssh':            'BlockdevCreateOptionsSsh',
+       'vdi':            'BlockdevCreateOptionsVdi',
+       'vhdx':           'BlockdevCreateOptionsVhdx',
++      'vitastor':       'BlockdevCreateOptionsVitastor',
+       'vmdk':           'BlockdevCreateOptionsVmdk',
+       'vpc':            'BlockdevCreateOptionsVpc'
+   } }
+Index: pve-qemu-kvm-9.0.0/scripts/ci/org.centos/stream/8/x86_64/configure
+===================================================================
+--- pve-qemu-kvm-9.0.0.orig/scripts/ci/org.centos/stream/8/x86_64/configure
++++ pve-qemu-kvm-9.0.0/scripts/ci/org.centos/stream/8/x86_64/configure
+@@ -30,7 +30,7 @@
+ --with-suffix="qemu-kvm" \
+ --firmwarepath=/usr/share/qemu-firmware \
+ --target-list="x86_64-softmmu" \
+---block-drv-rw-whitelist="qcow2,raw,file,host_device,nbd,iscsi,rbd,blkdebug,luks,null-co,nvme,copy-on-read,throttle,gluster" \
++--block-drv-rw-whitelist="qcow2,raw,file,host_device,nbd,iscsi,rbd,vitastor,blkdebug,luks,null-co,nvme,copy-on-read,throttle,gluster" \
+ --audio-drv-list="" \
+ --block-drv-ro-whitelist="vmdk,vhdx,vpc,https,ssh" \
+ --with-coroutine=ucontext \
+@@ -176,6 +176,7 @@
+ --enable-opengl \
+ --enable-pie \
+ --enable-rbd \
++--enable-vitastor \
+ --enable-rdma \
+ --enable-seccomp \
+ --enable-snappy \
+Index: pve-qemu-kvm-9.0.0/scripts/meson-buildoptions.sh
+===================================================================
+--- pve-qemu-kvm-9.0.0.orig/scripts/meson-buildoptions.sh
++++ pve-qemu-kvm-9.0.0/scripts/meson-buildoptions.sh
+@@ -168,6 +168,7 @@ meson_options_help() {
+   printf "%s\n" '  qed             qed image format support'
+   printf "%s\n" '  qga-vss         build QGA VSS support (broken with MinGW)'
+   printf "%s\n" '  rbd             Ceph block device driver'
++  printf "%s\n" '  vitastor        Vitastor block device driver'
+   printf "%s\n" '  rdma            Enable RDMA-based migration'
+   printf "%s\n" '  replication     replication support'
+   printf "%s\n" '  rutabaga-gfx    rutabaga_gfx support'
+@@ -445,6 +446,8 @@ _meson_option_parse() {
+     --disable-qom-cast-debug) printf "%s" -Dqom_cast_debug=false ;;
+     --enable-rbd) printf "%s" -Drbd=enabled ;;
+     --disable-rbd) printf "%s" -Drbd=disabled ;;
++    --enable-vitastor) printf "%s" -Dvitastor=enabled ;;
++    --disable-vitastor) printf "%s" -Dvitastor=disabled ;;
+     --enable-rdma) printf "%s" -Drdma=enabled ;;
+     --disable-rdma) printf "%s" -Drdma=disabled ;;
+     --enable-relocatable) printf "%s" -Drelocatable=true ;;

patches/qemu-9.1-vitastor.patch

@@ -0,0 +1,172 @@
+diff --git a/block/meson.build b/block/meson.build
+index f1262ec2ba..3cf3e23f16 100644
+--- a/block/meson.build
++++ b/block/meson.build
+@@ -114,6 +114,7 @@ foreach m : [
+   [libnfs, 'nfs', files('nfs.c')],
+   [libssh, 'ssh', files('ssh.c')],
+   [rbd, 'rbd', files('rbd.c')],
++  [vitastor, 'vitastor', files('vitastor.c')],
+ ]
+   if m[0].found()
+     module_ss = ss.source_set()
+diff --git a/meson.build b/meson.build
+index fbda17c987..3edac22aff 100644
+--- a/meson.build
++++ b/meson.build
+@@ -1510,6 +1510,26 @@ if not get_option('rbd').auto() or have_block
+   endif
+ endif
+ 
++vitastor = not_found
++if not get_option('vitastor').auto() or have_block
++  libvitastor_client = cc.find_library('vitastor_client', has_headers: ['vitastor_c.h'],
++    required: get_option('vitastor'))
++  if libvitastor_client.found()
++    if cc.links('''
++      #include <vitastor_c.h>
++      int main(void) {
++        vitastor_c_create_qemu(0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
++        return 0;
++      }''', dependencies: libvitastor_client)
++      vitastor = declare_dependency(dependencies: libvitastor_client)
++    elif get_option('vitastor').enabled()
++      error('could not link libvitastor_client')
++    else
++      warning('could not link libvitastor_client, disabling')
++    endif
++  endif
++endif
++
+ glusterfs = not_found
+ glusterfs_ftruncate_has_stat = false
+ glusterfs_iocb_has_stat = false
+@@ -2351,6 +2371,7 @@ endif
+ config_host_data.set('CONFIG_OPENGL', opengl.found())
+ config_host_data.set('CONFIG_PLUGIN', get_option('plugins'))
+ config_host_data.set('CONFIG_RBD', rbd.found())
++config_host_data.set('CONFIG_VITASTOR', vitastor.found())
+ config_host_data.set('CONFIG_RDMA', rdma.found())
+ config_host_data.set('CONFIG_RELOCATABLE', get_option('relocatable'))
+ config_host_data.set('CONFIG_SAFESTACK', get_option('safe_stack'))
+@@ -4510,6 +4531,7 @@ summary_info += {'fdt support':       fdt_opt == 'internal' ? 'internal' : fdt}
+ summary_info += {'libcap-ng support': libcap_ng}
+ summary_info += {'bpf support':       libbpf}
+ summary_info += {'rbd support':       rbd}
++summary_info += {'vitastor support':  vitastor}
+ summary_info += {'smartcard support': cacard}
+ summary_info += {'U2F support':       u2f}
+ summary_info += {'libusb':            libusb}
+diff --git a/meson_options.txt b/meson_options.txt
+index 0269fa0f16..4740ffdc27 100644
+--- a/meson_options.txt
++++ b/meson_options.txt
+@@ -194,6 +194,8 @@ option('lzo', type : 'feature', value : 'auto',
+        description: 'lzo compression support')
+ option('rbd', type : 'feature', value : 'auto',
+        description: 'Ceph block device driver')
++option('vitastor', type : 'feature', value : 'auto',
++       description: 'Vitastor block device driver')
+ option('opengl', type : 'feature', value : 'auto',
+        description: 'OpenGL support')
+ option('rdma', type : 'feature', value : 'auto',
+diff --git a/qapi/block-core.json b/qapi/block-core.json
+index aa40d44f1d..bbee6a0e9c 100644
+--- a/qapi/block-core.json
++++ b/qapi/block-core.json
+@@ -3203,7 +3203,7 @@
+             'parallels', 'preallocate', 'qcow', 'qcow2', 'qed', 'quorum',
+             'raw', 'rbd',
+             { 'name': 'replication', 'if': 'CONFIG_REPLICATION' },
+-            'ssh', 'throttle', 'vdi', 'vhdx',
++            'ssh', 'throttle', 'vdi', 'vhdx', 'vitastor',
+             { 'name': 'virtio-blk-vfio-pci', 'if': 'CONFIG_BLKIO' },
+             { 'name': 'virtio-blk-vhost-user', 'if': 'CONFIG_BLKIO' },
+             { 'name': 'virtio-blk-vhost-vdpa', 'if': 'CONFIG_BLKIO' },
+@@ -4286,6 +4286,28 @@
+             '*key-secret': 'str',
+             '*server': ['InetSocketAddressBase'] } }
+ 
++##
++# @BlockdevOptionsVitastor:
++#
++# Driver specific block device options for vitastor
++#
++# @image:       Image name
++# @inode:       Inode number
++# @pool:        Pool ID
++# @size:        Desired image size in bytes
++# @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',
++            '*image': 'str',
++            '*config-path': 'str',
++            '*etcd-host': 'str',
++            '*etcd-prefix': 'str' } }
++
+ ##
+ # @ReplicationMode:
+ #
+@@ -4742,6 +4764,7 @@
+       'throttle':   'BlockdevOptionsThrottle',
+       'vdi':        'BlockdevOptionsGenericFormat',
+       'vhdx':       'BlockdevOptionsGenericFormat',
++      'vitastor':   'BlockdevOptionsVitastor',
+       'virtio-blk-vfio-pci':
+                     { 'type': 'BlockdevOptionsVirtioBlkVfioPci',
+                       'if': 'CONFIG_BLKIO' },
+@@ -5183,6 +5206,20 @@
+             '*cluster-size' :   'size',
+             '*encrypt' :        'RbdEncryptionCreateOptions' } }
+ 
++##
++# @BlockdevCreateOptionsVitastor:
++#
++# Driver specific image creation options for Vitastor.
++#
++# @location: Where to store the new image file.  This location cannot
++#     point to a snapshot.
++#
++# @size: Size of the virtual disk in bytes
++##
++{ 'struct': 'BlockdevCreateOptionsVitastor',
++  'data': { 'location':         'BlockdevOptionsVitastor',
++            'size':             'size' } }
++
+ ##
+ # @BlockdevVmdkSubformat:
+ #
+@@ -5405,6 +5442,7 @@
+       'ssh':            'BlockdevCreateOptionsSsh',
+       'vdi':            'BlockdevCreateOptionsVdi',
+       'vhdx':           'BlockdevCreateOptionsVhdx',
++      'vitastor':       'BlockdevCreateOptionsVitastor',
+       'vmdk':           'BlockdevCreateOptionsVmdk',
+       'vpc':            'BlockdevCreateOptionsVpc'
+   } }
+diff --git a/scripts/meson-buildoptions.sh b/scripts/meson-buildoptions.sh
+index c97079a38c..4623f552ec 100644
+--- a/scripts/meson-buildoptions.sh
++++ b/scripts/meson-buildoptions.sh
+@@ -168,6 +168,7 @@ meson_options_help() {
+   printf "%s\n" '  qga-vss         build QGA VSS support (broken with MinGW)'
+   printf "%s\n" '  qpl             Query Processing Library support'
+   printf "%s\n" '  rbd             Ceph block device driver'
++  printf "%s\n" '  vitastor        Vitastor block device driver'
+   printf "%s\n" '  rdma            Enable RDMA-based migration'
+   printf "%s\n" '  replication     replication support'
+   printf "%s\n" '  rutabaga-gfx    rutabaga_gfx support'
+@@ -444,6 +445,8 @@ _meson_option_parse() {
+     --disable-qpl) printf "%s" -Dqpl=disabled ;;
+     --enable-rbd) printf "%s" -Drbd=enabled ;;
+     --disable-rbd) printf "%s" -Drbd=disabled ;;
++    --enable-vitastor) printf "%s" -Dvitastor=enabled ;;
++    --disable-vitastor) printf "%s" -Dvitastor=disabled ;;
+     --enable-rdma) printf "%s" -Drdma=enabled ;;
+     --disable-rdma) printf "%s" -Drdma=disabled ;;
+     --enable-relocatable) printf "%s" -Drelocatable=true ;;

src/client/etcd_state_client.cpp

@@ -176,7 +176,7 @@ void etcd_state_client_t::add_etcd_url(std::string addr)
             exit(1);
         }
         if (!local_ips.size())
-            local_ips = getifaddr_list();
+            local_ips = getifaddr_list(std::vector<std::string>(), true);
         std::string check_addr;
         int pos = addr.find('/');
         int pos2 = addr.find(':');
@@ -785,7 +785,7 @@ void etcd_state_client_t::parse_state(const etcd_kv_t & kv)
         }
         for (auto & pool_item: value.object_items())
         {
-            pool_config_t pc;
+            pool_config_t pc = {};
             // ID
             pool_id_t pool_id;
             char null_byte = 0;
@@ -931,12 +931,28 @@ void etcd_state_client_t::parse_state(const etcd_kv_t & kv)
             // Ignore old key if the new one is present
             return;
         }
+        for (auto & pool_id_json: value["backfillfull_pools"].array_items())
+        {
+            auto pool_id = pool_id_json.uint64_value();
+            auto pool_it = this->pool_config.find(pool_id);
+            if (pool_it != this->pool_config.end())
+            {
+                pool_it->second.backfillfull |= 2;
+            }
+        }
         for (auto & pool_item: this->pool_config)
         {
             for (auto & pg_item: pool_item.second.pg_config)
             {
                 pg_item.second.config_exists = false;
             }
+            // 3 = was 1 and became 1, 0 = was 0 and became 0
+            if (pool_item.second.backfillfull == 2 || pool_item.second.backfillfull == 1)
+            {
+                if (on_change_backfillfull_hook)
+                    on_change_backfillfull_hook(pool_item.first);
+            }
+            pool_item.second.backfillfull = pool_item.second.backfillfull >> 1;
         }
         for (auto & pool_item: value["items"].object_items())
         {

src/client/etcd_state_client.h

@@ -62,6 +62,7 @@ struct pool_config_t
     std::map<pg_num_t, pg_config_t> pg_config;
     uint64_t scrub_interval;
     std::string used_for_fs;
+    int backfillfull;
 };
 
 struct inode_config_t
@@ -131,6 +132,7 @@ public:
     std::function<json11::Json()> load_pgs_checks_hook;
     std::function<void(bool)> on_load_pgs_hook;
     std::function<void()> on_change_pool_config_hook;
+    std::function<void(pool_id_t)> on_change_backfillfull_hook;
     std::function<void(pool_id_t, pg_num_t, osd_num_t)> on_change_pg_state_hook;
     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;

src/client/http_client.cpp

@@ -62,6 +62,7 @@ struct http_co_t
     inline void end() { ended = true; if (!onstack) { delete this; } }
     void run_cb_and_clear();
     void start_connection();
+    void start_ws_connection();
     void close_connection();
     void next_request();
     void handle_events();
@@ -112,7 +113,7 @@ http_co_t* open_websocket(timerfd_manager_t *tfd, const std::string & host, cons
     handler->keepalive = false;
     handler->request = request;
     handler->response_callback = response_callback;
-    handler->start_connection();
+    handler->start_ws_connection();
     return handler;
 }
 
@@ -282,6 +283,27 @@ void http_co_t::close_connection()
     epoll_events = 0;
 }
 
+void http_co_t::start_ws_connection()
+{
+    stackin();
+    start_connection();
+    if (request_timeout > 0)
+    {
+        timeout_id = tfd->set_timer(request_timeout, false, [this](int timer_id)
+        {
+            stackin();
+            if (state != HTTP_CO_WEBSOCKET)
+            {
+                close_connection();
+                parsed = { .error = "Websocket connection timed out" };
+                run_cb_and_clear();
+            }
+            stackout();
+        });
+    }
+    stackout();
+}
+
 void http_co_t::start_connection()
 {
     stackin();

src/client/messenger.cpp

@@ -121,7 +121,7 @@ void osd_messenger_t::init()
     if (use_rdma)
     {
         rdma_context = msgr_rdma_context_t::create(
-            rdma_device != "" ? rdma_device.c_str() : NULL,
+            osd_networks, rdma_device != "" ? rdma_device.c_str() : NULL,
             rdma_port_num, rdma_gid_index, rdma_mtu, rdma_odp, log_level
         );
         if (!rdma_context)
@@ -266,7 +266,8 @@ void osd_messenger_t::parse_config(const json11::Json & config)
     this->rdma_port_num = (uint8_t)config["rdma_port_num"].uint64_value();
     if (!this->rdma_port_num)
         this->rdma_port_num = 1;
-    this->rdma_gid_index = (uint8_t)config["rdma_gid_index"].uint64_value();
+    if (!config["rdma_gid_index"].is_null())
+        this->rdma_gid_index = (uint8_t)config["rdma_gid_index"].uint64_value();
     this->rdma_mtu = (uint32_t)config["rdma_mtu"].uint64_value();
     this->rdma_max_sge = config["rdma_max_sge"].uint64_value();
     if (!this->rdma_max_sge)
@@ -281,6 +282,15 @@ void osd_messenger_t::parse_config(const json11::Json & config)
     if (!this->rdma_max_msg || this->rdma_max_msg > 128*1024*1024)
         this->rdma_max_msg = 129*1024;
     this->rdma_odp = config["rdma_odp"].bool_value();
+    std::vector<std::string> mask;
+    if (config["bind_address"].is_string())
+        mask.push_back(config["bind_address"].string_value());
+    else if (config["osd_network"].is_string())
+        mask.push_back(config["osd_network"].string_value());
+    else
+        for (auto v: config["osd_network"].array_items())
+            mask.push_back(v.string_value());
+    this->osd_networks = mask;
 #endif
     if (!osd_num)
         this->iothread_count = (uint32_t)config["client_iothread_count"].uint64_value();

src/client/messenger.h

@@ -165,8 +165,9 @@ protected:
 
 #ifdef WITH_RDMA
     bool use_rdma = true;
+    std::vector<std::string> osd_networks;
     std::string rdma_device;
-    uint64_t rdma_port_num = 1, rdma_gid_index = 0, rdma_mtu = 0;
+    uint64_t rdma_port_num = 1, rdma_gid_index = -1, rdma_mtu = 0;
     msgr_rdma_context_t *rdma_context = NULL;
     uint64_t rdma_max_sge = 0, rdma_max_send = 0, rdma_max_recv = 0;
     uint64_t rdma_max_msg = 0;
@@ -177,7 +178,7 @@ protected:
     std::vector<int> read_ready_clients;
     std::vector<int> write_ready_clients;
     // We don't use ringloop->set_immediate here because we may have no ringloop in client :)
-    std::vector<std::function<void()>> set_immediate;
+    std::vector<osd_op_t*> set_immediate_ops;
 
 public:
     timerfd_manager_t *tfd;
@@ -237,6 +238,8 @@ protected:
     void handle_op_hdr(osd_client_t *cl);
     bool handle_reply_hdr(osd_client_t *cl);
     void handle_reply_ready(osd_op_t *op);
+    void handle_immediate_ops();
+    void clear_immediate_ops(int peer_fd);
 
 #ifdef WITH_RDMA
     void try_send_rdma(osd_client_t *cl);

src/client/msgr_rdma.cpp

@@ -3,6 +3,7 @@
 
 #include <stdio.h>
 #include <stdlib.h>
+#include "addr_util.h"
 #include "msgr_rdma.h"
 #include "messenger.h"
 
@@ -69,7 +70,138 @@ msgr_rdma_connection_t::~msgr_rdma_connection_t()
     send_out_size = 0;
 }
 
-msgr_rdma_context_t *msgr_rdma_context_t::create(const char *ib_devname, uint8_t ib_port, uint8_t gid_index, uint32_t mtu, bool odp, int log_level)
+#ifdef IBV_ADVISE_MR_ADVICE_PREFETCH_NO_FAULT
+static bool is_ipv4_gid(ibv_gid_entry *gidx)
+{
+    return (((uint64_t*)gidx->gid.raw)[0] == 0 &&
+        ((uint32_t*)gidx->gid.raw)[2] == 0xffff0000);
+}
+
+static bool match_gid(ibv_gid_entry *gidx, addr_mask_t *networks, int nnet)
+{
+    if (gidx->gid_type != IBV_GID_TYPE_ROCE_V1 &&
+        gidx->gid_type != IBV_GID_TYPE_ROCE_V2 ||
+        ((uint64_t*)gidx->gid.raw)[0] == 0 &&
+        ((uint64_t*)gidx->gid.raw)[1] == 0)
+    {
+        return false;
+    }
+    if (is_ipv4_gid(gidx))
+    {
+        for (int i = 0; i < nnet; i++)
+        {
+            if (networks[i].family == AF_INET && cidr_match(*(in_addr*)(gidx->gid.raw+12), networks[i].ipv4, networks[i].bits))
+                return true;
+        }
+    }
+    else
+    {
+        for (int i = 0; i < nnet; i++)
+        {
+            if (networks[i].family == AF_INET6 && cidr6_match(*(in6_addr*)gidx->gid.raw, networks[i].ipv6, networks[i].bits))
+                return true;
+        }
+    }
+    return false;
+}
+
+struct matched_dev
+{
+    int dev = -1;
+    int port = -1;
+    int gid = -1;
+    bool rocev2 = false;
+};
+
+static void log_rdma_dev_port_gid(ibv_device *dev, int ib_port, int gid_index, ibv_gid_entry & gidx)
+{
+    bool is4 = ((uint64_t*)gidx.gid.raw)[0] == 0 && ((uint32_t*)gidx.gid.raw)[2] == 0xffff0000;
+    char buf[256];
+    inet_ntop(is4 ? AF_INET : AF_INET6, is4 ? gidx.gid.raw+12 : gidx.gid.raw, buf, sizeof(buf));
+    fprintf(
+        stderr, "Auto-selected RDMA device %s port %d GID %d - ROCEv%d IPv%d %s\n",
+        ibv_get_device_name(dev), ib_port, gid_index,
+        gidx.gid_type == IBV_GID_TYPE_ROCE_V2 ? 2 : 1, is4 ? 4 : 6, buf
+    );
+}
+
+static matched_dev match_device(ibv_device **dev_list, addr_mask_t *networks, int nnet, int log_level)
+{
+    matched_dev best;
+    ibv_device_attr attr;
+    ibv_port_attr portinfo;
+    ibv_gid_entry best_gidx;
+    int res;
+    bool have_non_roce = false, have_roce = false;
+    for (int i = 0; dev_list[i]; ++i)
+    {
+        auto dev = dev_list[i];
+        ibv_context *context = ibv_open_device(dev_list[i]);
+        if ((res = ibv_query_device(context, &attr)) != 0)
+        {
+            fprintf(stderr, "Couldn't query RDMA device %s for its features: %s\n", ibv_get_device_name(dev_list[i]), strerror(res));
+            goto cleanup;
+        }
+        for (int j = 1; j <= attr.phys_port_cnt; j++)
+        {
+            // Try to find a port with matching address
+            if ((res = ibv_query_port(context, j, &portinfo)) != 0)
+            {
+                fprintf(stderr, "Couldn't get RDMA device %s port %d info: %s\n", ibv_get_device_name(dev), j, strerror(res));
+                goto cleanup;
+            }
+            for (int k = 0; k < portinfo.gid_tbl_len; k++)
+            {
+                ibv_gid_entry gidx;
+                if ((res = ibv_query_gid_ex(context, j, k, &gidx, 0)) != 0)
+                {
+                    if (res != ENODATA)
+                    {
+                        fprintf(stderr, "Couldn't read RDMA device %s GID index %d: %s\n", ibv_get_device_name(dev), k, strerror(res));
+                        goto cleanup;
+                    }
+                    else
+                        break;
+                }
+                if (gidx.gid_type != IBV_GID_TYPE_ROCE_V1 &&
+                    gidx.gid_type != IBV_GID_TYPE_ROCE_V2)
+                    have_non_roce = true;
+                else
+                    have_roce = true;
+                if (match_gid(&gidx, networks, nnet))
+                {
+                    // Prefer RoCEv2
+                    if (!best.rocev2)
+                    {
+                        best.dev = i;
+                        best.port = j;
+                        best.gid = k;
+                        best.rocev2 = (gidx.gid_type == IBV_GID_TYPE_ROCE_V2);
+                        best_gidx = gidx;
+                    }
+                }
+            }
+        }
+cleanup:
+        ibv_close_device(context);
+        if (best.rocev2)
+        {
+            break;
+        }
+    }
+    if (best.dev >= 0 && log_level > 0)
+    {
+        log_rdma_dev_port_gid(dev_list[best.dev], best.port, best.gid, best_gidx);
+    }
+    if (best.dev < 0 && have_non_roce && !have_roce)
+    {
+        best.dev = -2;
+    }
+    return best;
+}
+#endif
+
+msgr_rdma_context_t *msgr_rdma_context_t::create(std::vector<std::string> osd_networks, const char *ib_devname, uint8_t ib_port, uint8_t gid_index, uint32_t mtu, bool odp, int log_level)
 {
     int res;
     ibv_device **dev_list = NULL;
@@ -80,28 +212,23 @@ msgr_rdma_context_t *msgr_rdma_context_t::create(const char *ib_devname, uint8_t
     clock_gettime(CLOCK_REALTIME, &tv);
     srand48(tv.tv_sec*1000000000 + tv.tv_nsec);
     dev_list = ibv_get_device_list(NULL);
-    if (!dev_list)
+    if (!dev_list || !*dev_list)
     {
         if (errno == -ENOSYS || errno == ENOSYS)
         {
             if (log_level > 0)
                 fprintf(stderr, "No RDMA devices found (RDMA device list returned ENOSYS)\n");
         }
+        else if (!*dev_list)
+        {
+            if (log_level > 0)
+                fprintf(stderr, "No RDMA devices found\n");
+        }
         else
             fprintf(stderr, "Failed to get RDMA device list: %s\n", strerror(errno));
         goto cleanup;
     }
-    if (!ib_devname)
+    if (ib_devname)
-    {
-        ctx->dev = *dev_list;
-        if (!ctx->dev)
-        {
-            if (log_level > 0)
-                fprintf(stderr, "No RDMA devices found\n");
-            goto cleanup;
-        }
-    }
-    else
     {
         int i;
         for (i = 0; dev_list[i]; ++i)
@@ -114,6 +241,39 @@ msgr_rdma_context_t *msgr_rdma_context_t::create(const char *ib_devname, uint8_t
             goto cleanup;
         }
     }
+#ifdef IBV_ADVISE_MR_ADVICE_PREFETCH_NO_FAULT
+    else if (osd_networks.size())
+    {
+        std::vector<addr_mask_t> nets;
+        for (auto & netstr: osd_networks)
+        {
+            nets.push_back(cidr_parse(netstr));
+        }
+        auto best = match_device(dev_list, nets.data(), nets.size(), log_level);
+        if (best.dev == -2)
+        {
+            best.dev = 0;
+            if (log_level > 0)
+                fprintf(stderr, "No RoCE devices found, using first available RDMA device %s\n", ibv_get_device_name(*dev_list));
+        }
+        else if (best.dev < 0)
+        {
+            if (log_level > 0)
+                fprintf(stderr, "RDMA device matching osd_network is not found, disabling RDMA\n");
+            goto cleanup;
+        }
+        else
+        {
+            ib_port = best.port;
+            gid_index = best.gid;
+        }
+        ctx->dev = dev_list[best.dev];
+    }
+#endif
+    else
+    {
+        ctx->dev = *dev_list;
+    }
 
     ctx->context = ibv_open_device(ctx->dev);
     if (!ctx->context)
@@ -123,7 +283,6 @@ msgr_rdma_context_t *msgr_rdma_context_t::create(const char *ib_devname, uint8_t
     }
 
     ctx->ib_port = ib_port;
-    ctx->gid_index = gid_index;
     if ((res = ibv_query_port(ctx->context, ib_port, &ctx->portinfo)) != 0)
     {
         fprintf(stderr, "Couldn't get RDMA device %s port %d info: %s\n", ibv_get_device_name(ctx->dev), ib_port, strerror(res));
@@ -135,11 +294,55 @@ msgr_rdma_context_t *msgr_rdma_context_t::create(const char *ib_devname, uint8_t
         fprintf(stderr, "RDMA device %s must have local LID because it's not Ethernet, but LID is zero\n", ibv_get_device_name(ctx->dev));
         goto cleanup;
     }
-    if (ibv_query_gid(ctx->context, ib_port, gid_index, &ctx->my_gid))
+
+#ifdef IBV_ADVISE_MR_ADVICE_PREFETCH_NO_FAULT
+    if (gid_index != -1)
+#endif
     {
-        fprintf(stderr, "Couldn't read RDMA device %s GID index %d\n", ibv_get_device_name(ctx->dev), gid_index);
+        ctx->gid_index = gid_index < 0 ? 0 : gid_index;
-        goto cleanup;
+        if (ibv_query_gid(ctx->context, ib_port, gid_index, &ctx->my_gid))
+        {
+            fprintf(stderr, "Couldn't read RDMA device %s GID index %d\n", ibv_get_device_name(ctx->dev), gid_index);
+            goto cleanup;
+        }
     }
+#ifdef IBV_ADVISE_MR_ADVICE_PREFETCH_NO_FAULT
+    else
+    {
+        // Auto-guess GID
+        ibv_gid_entry best_gidx;
+        for (int k = 0; k < ctx->portinfo.gid_tbl_len; k++)
+        {
+            ibv_gid_entry gidx;
+            if (ibv_query_gid_ex(ctx->context, ib_port, k, &gidx, 0) != 0)
+            {
+                fprintf(stderr, "Couldn't read RDMA device %s GID index %d\n", ibv_get_device_name(ctx->dev), k);
+                goto cleanup;
+            }
+            // Skip empty GID
+            if (((uint64_t*)gidx.gid.raw)[0] == 0 &&
+                ((uint64_t*)gidx.gid.raw)[1] == 0)
+            {
+                continue;
+            }
+            // Prefer IPv4 RoCEv2 -> IPv6 RoCEv2 -> IPv4 RoCEv1 -> IPv6 RoCEv1 -> IB
+            if (gid_index == -1 ||
+                gidx.gid_type == IBV_GID_TYPE_ROCE_V2 && best_gidx.gid_type != IBV_GID_TYPE_ROCE_V2 ||
+                gidx.gid_type == IBV_GID_TYPE_ROCE_V1 && best_gidx.gid_type == IBV_GID_TYPE_IB ||
+                gidx.gid_type == best_gidx.gid_type && is_ipv4_gid(&gidx))
+            {
+                gid_index = k;
+                best_gidx = gidx;
+            }
+        }
+        ctx->gid_index = gid_index = (gid_index == -1 ? 0 : gid_index);
+        if (log_level > 0)
+        {
+            log_rdma_dev_port_gid(ctx->dev, ctx->ib_port, ctx->gid_index, best_gidx);
+        }
+        ctx->my_gid = best_gidx.gid;
+    }
+#endif
 
     ctx->pd = ibv_alloc_pd(ctx->context);
     if (!ctx->pd)
@@ -598,6 +801,7 @@ void osd_messenger_t::handle_rdma_events()
                 }
                 fprintf(stderr, " with status: %s, stopping client\n", ibv_wc_status_str(wc[i].status));
                 stop_client(client_id);
+                clear_immediate_ops(client_id);
                 continue;
             }
             if (!is_send)
@@ -606,6 +810,7 @@ void osd_messenger_t::handle_rdma_events()
                 if (!handle_read_buffer(cl, rc->recv_buffers[rc->next_recv_buf].buf, wc[i].byte_len))
                 {
                     // handle_read_buffer may stop the client
+                    clear_immediate_ops(client_id);
                     continue;
                 }
                 try_recv_rdma_wr(cl, rc->recv_buffers[rc->next_recv_buf]);
@@ -666,9 +871,5 @@ void osd_messenger_t::handle_rdma_events()
             }
         }
     } while (event_count > 0);
-    for (auto cb: set_immediate)
+    handle_immediate_ops();
-    {
-        cb();
-    }
-    set_immediate.clear();
 }

src/client/msgr_rdma.h

@@ -36,7 +36,7 @@ struct msgr_rdma_context_t
     int max_cqe = 0;
     int used_max_cqe = 0;
 
-    static msgr_rdma_context_t *create(const char *ib_devname, uint8_t ib_port, uint8_t gid_index, uint32_t mtu, bool odp, int log_level);
+    static msgr_rdma_context_t *create(std::vector<std::string> osd_networks, const char *ib_devname, uint8_t ib_port, uint8_t gid_index, uint32_t mtu, bool odp, int log_level);
     ~msgr_rdma_context_t();
 };
 

src/client/msgr_receive.cpp

@@ -65,6 +65,7 @@ void osd_messenger_t::read_requests()
 bool osd_messenger_t::handle_read(int result, osd_client_t *cl)
 {
     bool ret = false;
+    int peer_fd = cl->peer_fd;
     cl->read_msg.msg_iovlen = 0;
     cl->refs--;
     if (cl->peer_state == PEER_STOPPED)
@@ -101,7 +102,8 @@ bool osd_messenger_t::handle_read(int result, osd_client_t *cl)
         {
             if (!handle_read_buffer(cl, cl->in_buf, result))
             {
-                goto fin;
+                clear_immediate_ops(peer_fd);
+                return false;
             }
         }
         else
@@ -113,7 +115,8 @@ bool osd_messenger_t::handle_read(int result, osd_client_t *cl)
             {
                 if (!handle_finished_read(cl))
                 {
-                    goto fin;
+                    clear_immediate_ops(peer_fd);
+                    return false;
                 }
             }
         }
@@ -122,15 +125,47 @@ bool osd_messenger_t::handle_read(int result, osd_client_t *cl)
             ret = true;
         }
     }
-fin:
+    handle_immediate_ops();
-    for (auto cb: set_immediate)
-    {
-        cb();
-    }
-    set_immediate.clear();
     return ret;
 }
 
+void osd_messenger_t::clear_immediate_ops(int peer_fd)
+{
+    size_t i = 0, j = 0;
+    while (i < set_immediate_ops.size())
+    {
+        if (set_immediate_ops[i]->peer_fd == peer_fd)
+        {
+            delete set_immediate_ops[i];
+        }
+        else
+        {
+            if (i != j)
+                set_immediate_ops[j] = set_immediate_ops[i];
+            j++;
+        }
+        i++;
+    }
+    set_immediate_ops.resize(j);
+}
+
+void osd_messenger_t::handle_immediate_ops()
+{
+    for (auto op: set_immediate_ops)
+    {
+        if (op->op_type == OSD_OP_IN)
+        {
+            exec_op(op);
+        }
+        else
+        {
+            // Copy lambda to be unaffected by `delete op`
+            std::function<void(osd_op_t*)>(op->callback)(op);
+        }
+    }
+    set_immediate_ops.clear();
+}
+
 bool osd_messenger_t::handle_read_buffer(osd_client_t *cl, void *curbuf, int remain)
 {
     // Compose operation(s) from the buffer
@@ -199,7 +234,7 @@ bool osd_messenger_t::handle_finished_read(osd_client_t *cl)
     {
         // Operation is ready
         cl->received_ops.push_back(cl->read_op);
-        set_immediate.push_back([this, op = cl->read_op]() { exec_op(op); });
+        set_immediate_ops.push_back(cl->read_op);
         cl->read_op = NULL;
         cl->read_state = 0;
     }
@@ -295,7 +330,7 @@ void osd_messenger_t::handle_op_hdr(osd_client_t *cl)
     {
         // Operation is ready
         cl->received_ops.push_back(cur_op);
-        set_immediate.push_back([this, cur_op]() { exec_op(cur_op); });
+        set_immediate_ops.push_back(cur_op);
         cl->read_op = NULL;
         cl->read_state = 0;
     }
@@ -416,9 +451,5 @@ void osd_messenger_t::handle_reply_ready(osd_op_t *op)
         (tv_end.tv_sec - op->tv_begin.tv_sec)*1000000 +
         (tv_end.tv_nsec - op->tv_begin.tv_nsec)/1000
     );
-    set_immediate.push_back([op]()
+    set_immediate_ops.push_back(op);
-    {
-        // Copy lambda to be unaffected by `delete op`
-        std::function<void(osd_op_t*)>(op->callback)(op);
-    });
 }

src/client/qemu_driver.c

@@ -16,7 +16,6 @@
 #include "qapi/error.h"
 #include "qapi/qmp/qdict.h"
 #include "qapi/qmp/qerror.h"
-#include "qemu/uri.h"
 #include "qemu/error-report.h"
 #include "qemu/module.h"
 #include "qemu/option.h"
@@ -1021,7 +1020,11 @@ static BlockDriver bdrv_vitastor = {
     // FIXME: Implement it along with per-inode statistics
     //.bdrv_get_allocated_file_size   = vitastor_get_allocated_file_size,
 
+#if QEMU_VERSION_MAJOR > 9 || QEMU_VERSION_MAJOR == 9 && QEMU_VERSION_MINOR > 0
+    .bdrv_open                      = vitastor_file_open,
+#else
     .bdrv_file_open                 = vitastor_file_open,
+#endif
     .bdrv_close                     = vitastor_close,
 
     // Option list for the create operation

src/cmd/cli_dd.cpp

@@ -261,6 +261,7 @@ struct dd_out_info_t
             else
             {
                 // ok
+                out_size = owatch->cfg.size;
                 return true;
             }
             // Wait for sub-command
@@ -882,6 +883,8 @@ resume_2:
         oinfo.end_fsync = oinfo.end_fsync && oinfo.out_seekable;
         read_offset = 0;
         read_end = iinfo.in_seekable ? iinfo.in_size-iseek : 0;
+        if (oinfo.out_size && (!read_end || read_end > oinfo.out_size-oseek))
+            read_end = oinfo.out_size-oseek;
         if (bytelimit && (!read_end || read_end > bytelimit))
             read_end = bytelimit;
         clock_gettime(CLOCK_REALTIME, &tv_begin);

src/cmd/cli_osd_tree.cpp

@@ -216,7 +216,7 @@ resume_1:
                 for (uint64_t osd_num: node.child_osds)
                 {
                     auto & osd = placement_tree->osds.at(osd_num);
-                    fmt_items.push_back(json11::Json::object{
+                    auto json_osd = json11::Json::object{
                         { "type", "osd" },
                         { "name", osd.num },
                         { "parent", node.name },
@@ -230,7 +230,16 @@ resume_1:
                         { "bitmap", (uint64_t)osd.bitmap_granularity },
                         { "commit", osd.immediate_commit == IMMEDIATE_NONE ? "none" : (osd.immediate_commit == IMMEDIATE_ALL ? "all" : "small") },
                         { "op_stats", osd_stats[osd_num]["op_stats"] },
-                    });
+                    };
+                    if (osd_stats[osd_num]["slow_ops_primary"].uint64_value() > 0)
+                    {
+                        json_osd["slow_ops_primary"] = osd_stats[osd_num]["slow_ops_primary"];
+                    }
+                    if (osd_stats[osd_num]["slow_ops_secondary"].uint64_value() > 0)
+                    {
+                        json_osd["slow_ops_secondary"] = osd_stats[osd_num]["slow_ops_secondary"];
+                    }
+                    fmt_items.push_back(json_osd);
                 }
             }
             result.data = fmt_items;

src/cmd/cli_pool_create.cpp

@@ -35,6 +35,7 @@ struct pool_creator_t
     uint64_t new_pools_mod_rev;
     json11::Json state_node_tree;
     json11::Json new_pools;
+    std::map<osd_num_t, json11::Json> osd_stats;
 
     bool is_done() { return state == 100; }
 
@@ -46,8 +47,6 @@ struct pool_creator_t
             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)
@@ -90,13 +89,19 @@ resume_1:
         // If not forced, check that we have enough osds for pg_size
         if (!force)
         {
-            // Get node_placement configuration from etcd
+            // Get node_placement configuration from etcd and OSD stats
             parent->etcd_txn(json11::Json::object {
                 { "success", json11::Json::array {
                     json11::Json::object {
                         { "request_range", json11::Json::object {
                             { "key", base64_encode(parent->cli->st_cli.etcd_prefix+"/config/node_placement") },
-                        } }
+                        } },
+                    },
+                    json11::Json::object {
+                        { "request_range", json11::Json::object {
+                            { "key", base64_encode(parent->cli->st_cli.etcd_prefix+"/osd/stats/") },
+                            { "range_end", base64_encode(parent->cli->st_cli.etcd_prefix+"/osd/stats0") },
+                        } },
                     },
                 } },
             });
@@ -112,10 +117,21 @@ resume_2:
                 return;
             }
 
-            // Get state_node_tree based on node_placement and osd peer states
+            // Get state_node_tree based on node_placement and osd stats
             {
-                auto kv = parent->cli->st_cli.parse_etcd_kv(parent->etcd_result["responses"][0]["response_range"]["kvs"][0]);
+                auto node_placement_kv = parent->cli->st_cli.parse_etcd_kv(parent->etcd_result["responses"][0]["response_range"]["kvs"][0]);
-                state_node_tree = get_state_node_tree(kv.value.object_items());
+                timespec tv_now;
+                clock_gettime(CLOCK_REALTIME, &tv_now);
+                uint64_t osd_out_time = parent->cli->config["osd_out_time"].uint64_value();
+                if (!osd_out_time)
+                    osd_out_time = 600;
+                osd_stats.clear();
+                parent->iterate_kvs_1(parent->etcd_result["responses"][1]["response_range"]["kvs"], "/osd/stats/", [&](uint64_t cur_osd, json11::Json value)
+                {
+                    if ((uint64_t)value["time"].number_value()+osd_out_time >= tv_now.tv_sec)
+                        osd_stats[cur_osd] = value;
+                });
+                state_node_tree = get_state_node_tree(node_placement_kv.value.object_items(), osd_stats);
             }
 
             // Skip tag checks, if pool has none
@@ -158,42 +174,18 @@ resume_3:
                 }
             }
 
-            // Get stats (for block_size, bitmap_granularity, ...) of osds in state_node_tree
-            {
-                json11::Json::array osd_stats;
-
-                for (auto osd_num: state_node_tree["osds"].array_items())
-                {
-                    osd_stats.push_back(json11::Json::object {
-                        { "request_range", json11::Json::object {
-                            { "key", base64_encode(parent->cli->st_cli.etcd_prefix+"/osd/stats/"+osd_num.as_string()) },
-                        } }
-                    });
-                }
-
-                parent->etcd_txn(json11::Json::object{ { "success", osd_stats } });
-            }
-
-            state = 4;
-resume_4:
-            if (parent->waiting > 0)
-                return;
-            if (parent->etcd_err.err)
-            {
-                result = parent->etcd_err;
-                state = 100;
-                return;
-            }
-
             // Filter osds from state_node_tree based on pool parameters and osd stats
             {
-                std::vector<json11::Json> osd_stats;
+                std::vector<json11::Json> filtered_osd_stats;
-                for (auto & ocr: parent->etcd_result["responses"].array_items())
+                for (auto & osd_num: state_node_tree["osds"].array_items())
                 {
-                    auto kv = parent->cli->st_cli.parse_etcd_kv(ocr["response_range"]["kvs"][0]);
+                    auto st_it = osd_stats.find(osd_num.uint64_value());
-                    osd_stats.push_back(kv.value);
+                    if (st_it != osd_stats.end())
+                    {
+                        filtered_osd_stats.push_back(st_it->second);
+                    }
                 }
-                guess_block_size(osd_stats);
+                guess_block_size(filtered_osd_stats);
                 state_node_tree = filter_state_node_tree_by_stats(state_node_tree, osd_stats);
             }
 
@@ -201,8 +193,7 @@ resume_4:
             {
                 auto failure_domain = cfg["failure_domain"].string_value() == ""
                     ? "host" : cfg["failure_domain"].string_value();
-                uint64_t max_pg_size = get_max_pg_size(state_node_tree["nodes"].object_items(),
+                uint64_t max_pg_size = get_max_pg_size(state_node_tree, failure_domain, cfg["root_node"].string_value());
-                    failure_domain, cfg["root_node"].string_value());
 
                 if (cfg["pg_size"].uint64_value() > max_pg_size)
                 {
@@ -358,56 +349,50 @@ resume_8:
 
     // Returns a JSON object of form {"nodes": {...}, "osds": [...]} that
     // contains: all nodes (osds, hosts, ...) based on node_placement config
-    // and current peer state, and a list of active peer osds.
+    // and current osd stats.
-    json11::Json get_state_node_tree(json11::Json::object node_placement)
+    json11::Json get_state_node_tree(json11::Json::object node_placement, std::map<osd_num_t, json11::Json> & osd_stats)
     {
-        // Erase non-peer osd nodes from node_placement
+        // Erase non-existing osd nodes from node_placement
         for (auto np_it = node_placement.begin(); np_it != node_placement.end();)
         {
             // Numeric nodes are osds
             osd_num_t osd_num = stoull_full(np_it->first);
 
-            // If node is osd and it is not in peer states, erase it
+            // If node is osd and its stats do not exist, erase it
-            if (osd_num > 0 &&
+            if (osd_num > 0 && osd_stats.find(osd_num) == osd_stats.end())
-                parent->cli->st_cli.peer_states.find(osd_num) == parent->cli->st_cli.peer_states.end())
-            {
                 node_placement.erase(np_it++);
-            }
             else
                 np_it++;
         }
 
-        // List of peer osds
+        // List of osds
-        std::vector<std::string> peer_osds;
+        std::vector<std::string> existing_osds;
 
-        // Record peer osds and add missing osds/hosts to np
+        // Record osds and add missing osds/hosts to np
-        for (auto & ps: parent->cli->st_cli.peer_states)
+        for (auto & ps: osd_stats)
         {
             std::string osd_num = std::to_string(ps.first);
 
-            // Record peer osd
+            // Record osd
-            peer_osds.push_back(osd_num);
+            existing_osds.push_back(osd_num);
 
-            // Add osd, if necessary
+            // Add host if necessary
-            if (node_placement.find(osd_num) == node_placement.end())
+            std::string osd_host = ps.second["host"].as_string();
+            if (node_placement.find(osd_host) == node_placement.end())
             {
-                std::string osd_host = ps.second["host"].as_string();
+                node_placement[osd_host] = json11::Json::object {
-
+                    { "level", "host" }
-                // Add host, if necessary
-                if (node_placement.find(osd_host) == node_placement.end())
-                {
-                    node_placement[osd_host] = json11::Json::object {
-                        { "level", "host" }
-                    };
-                }
-
-                node_placement[osd_num] = json11::Json::object {
-                    { "parent", osd_host }
                 };
             }
+
+            // Add osd
+            node_placement[osd_num] = json11::Json::object {
+                { "parent", node_placement[osd_num]["parent"].is_null() ? osd_host : node_placement[osd_num]["parent"] },
+                { "level", "osd" },
+            };
         }
 
-        return json11::Json::object { { "osds", peer_osds }, { "nodes", node_placement } };
+        return json11::Json::object { { "osds", existing_osds }, { "nodes", node_placement } };
     }
 
     // Returns new state_node_tree based on given state_node_tree with osds
@@ -534,15 +519,13 @@ resume_8:
     // filtered out by stats parameters (block_size, bitmap_granularity) in
     // given osd_stats and current pool config.
     // Requires: state_node_tree["osds"] must match osd_stats 1-1
-    json11::Json filter_state_node_tree_by_stats(const json11::Json & state_node_tree, std::vector<json11::Json> & osd_stats)
+    json11::Json filter_state_node_tree_by_stats(const json11::Json & state_node_tree, std::map<osd_num_t, json11::Json> & osd_stats)
     {
-        auto & osds = state_node_tree["osds"].array_items();
-
         // Accepted state_node_tree nodes
         auto accepted_nodes = state_node_tree["nodes"].object_items();
 
         // List of accepted osds
-        std::vector<std::string> accepted_osds;
+        json11::Json::array accepted_osds;
 
         block_size = cfg["block_size"].uint64_value()
             ? cfg["block_size"].uint64_value()
@@ -554,21 +537,25 @@ resume_8:
             ? etcd_state_client_t::parse_immediate_commit(cfg["immediate_commit"].string_value(), IMMEDIATE_ALL)
             : parent->cli->st_cli.global_immediate_commit;
 
-        for (size_t i = 0; i < osd_stats.size(); i++)
+        for (auto osd_num_json: state_node_tree["osds"].array_items())
         {
-            auto & os = osd_stats[i];
+            auto osd_num = osd_num_json.uint64_value();
-            // Get osd number
+            auto os_it = osd_stats.find(osd_num);
-            auto osd_num = osds[i].as_string();
+            if (os_it == osd_stats.end())
+            {
+                continue;
+            }
+            auto & os = os_it->second;
             if (!os["data_block_size"].is_null() && os["data_block_size"] != block_size ||
                 !os["bitmap_granularity"].is_null() && os["bitmap_granularity"] != bitmap_granularity ||
                 !os["immediate_commit"].is_null() &&
                 etcd_state_client_t::parse_immediate_commit(os["immediate_commit"].string_value(), IMMEDIATE_NONE) < immediate_commit)
             {
-                accepted_nodes.erase(osd_num);
+                accepted_nodes.erase(osd_num_json.as_string());
             }
             else
             {
-                accepted_osds.push_back(osd_num);
+                accepted_osds.push_back(osd_num_json);
             }
         }
 
@@ -576,87 +563,28 @@ resume_8:
     }
 
     // Returns maximum pg_size possible for given node_tree and failure_domain, starting at parent_node
-    uint64_t get_max_pg_size(json11::Json::object node_tree, const std::string & level, const std::string & parent_node)
+    uint64_t get_max_pg_size(json11::Json state_node_tree, const std::string & level, const std::string & root_node)
     {
-        uint64_t max_pg_sz = 0;
+        std::set<std::string> level_seen;
-
+        for (auto & osd: state_node_tree["osds"].array_items())
-        std::vector<std::string> nodes;
-
-        // Check if parent node is an osd (numeric)
-        if (parent_node != "" && stoull_full(parent_node))
         {
-            // Add it to node list if osd is in node tree
+            // find OSD parent at <level>, but stop at <root_node>
-            if (node_tree.find(parent_node) != node_tree.end())
+            auto cur_id = osd.string_value();
-                nodes.push_back(parent_node);
+            auto cur = state_node_tree["nodes"][cur_id];
-        }
+            while (!cur.is_null())
-        // If parent node given, ...
-        else if (parent_node != "")
-        {
-            // ... look for children nodes of this parent
-            for (auto & sn: node_tree)
             {
-                auto & props = sn.second.object_items();
+                if (cur["level"] == level)
-
-                auto parent_prop = props.find("parent");
-                if (parent_prop != props.end() && (parent_prop->second.as_string() == parent_node))
                 {
-                    nodes.push_back(sn.first);
+                    level_seen.insert(cur_id);
-
+                    break;
-                    // If we're not looking for all osds, we only need a single
-                    // child osd node
-                    if (level != "osd" && stoull_full(sn.first))
-                        break;
                 }
+                if (cur_id == root_node)
+                    break;
+                cur_id = cur["parent"].string_value();
+                cur = state_node_tree["nodes"][cur_id];
             }
         }
-        // No parent node given, and we're not looking for all osds
+        return level_seen.size();
-        else if (level != "osd")
-        {
-            // ... look for all level nodes
-            for (auto & sn: node_tree)
-            {
-                auto & props = sn.second.object_items();
-
-                auto level_prop = props.find("level");
-                if (level_prop != props.end() && (level_prop->second.as_string() == level))
-                {
-                    nodes.push_back(sn.first);
-                }
-            }
-        }
-        // Otherwise, ...
-        else
-        {
-            // ... we're looking for osd nodes only
-            for (auto & sn: node_tree)
-            {
-                if (stoull_full(sn.first))
-                {
-                    nodes.push_back(sn.first);
-                }
-            }
-        }
-
-        // Process gathered nodes
-        for (auto & node: nodes)
-        {
-            // Check for osd node, return constant max size
-            if (stoull_full(node))
-            {
-                max_pg_sz += 1;
-            }
-            // Otherwise, ...
-            else
-            {
-                // ... exclude parent node from tree, and ...
-                node_tree.erase(parent_node);
-
-                // ... descend onto the resulting tree
-                max_pg_sz += get_max_pg_size(node_tree, level, node);
-            }
-        }
-
-        return max_pg_sz;
     }
 
     json11::Json create_pool(const etcd_kv_t & kv)

src/cmd/cli_status.cpp

@@ -134,6 +134,7 @@ resume_2:
         }
         int osd_count = 0, osd_up = 0;
         uint64_t total_raw = 0, free_raw = 0, free_down_raw = 0, down_raw = 0;
+        std::vector<uint64_t> slow_op_primary_osds, slow_op_secondary_osds;
         parent->iterate_kvs_1(osd_stats, "/osd/stats/", [&](uint64_t stat_osd_num, json11::Json value)
         {
             osd_count++;
@@ -141,18 +142,29 @@ resume_2:
             auto osd_free = value["free"].uint64_value();
             total_raw += osd_size;
             free_raw += osd_free;
-            if (!osd_free)
+            if (osd_size)
             {
-                osds_full++;
+                if (!osd_free)
-            }
+                {
-            else if (osd_free < (uint64_t)(osd_size*(1-osd_nearfull_ratio)))
+                    osds_full++;
-            {
+                }
-                osds_nearfull++;
+                else if (osd_free < (uint64_t)(osd_size*(1-osd_nearfull_ratio)))
+                {
+                    osds_nearfull++;
+                }
             }
             auto peer_it = parent->cli->st_cli.peer_states.find(stat_osd_num);
             if (peer_it != parent->cli->st_cli.peer_states.end())
             {
                 osd_up++;
+                if (value["slow_ops_primary"].uint64_value() > 0)
+                {
+                    slow_op_primary_osds.push_back(stat_osd_num);
+                }
+                if (value["slow_ops_secondary"].uint64_value() > 0)
+                {
+                    slow_op_secondary_osds.push_back(stat_osd_num);
+                }
             }
             else
             {
@@ -216,6 +228,10 @@ resume_2:
                 { "mon_master", mon_master },
                 { "osd_up", osd_up },
                 { "osd_count", osd_count },
+                { "osds_full", osds_full },
+                { "osds_nearfull", osds_nearfull },
+                { "osds_primary_slow_ops", slow_op_primary_osds },
+                { "osds_secondary_slow_ops", slow_op_secondary_osds },
                 { "total_raw", total_raw },
                 { "free_raw", free_raw },
                 { "down_raw", down_raw },
@@ -300,6 +316,26 @@ resume_2:
             warning_str += "    "+std::to_string(osds_nearfull)+
                 (osds_nearfull > 1 ? " osds are almost full\n" : " osd is almost full\n");
         }
+        if (slow_op_primary_osds.size() > 0)
+        {
+            warning_str += "    "+std::to_string(slow_op_primary_osds.size());
+            warning_str += (slow_op_primary_osds.size() > 1 ? " osds have" : " osd has");
+            warning_str += " slow client ops: ";
+            for (int i = 0; i < slow_op_primary_osds.size(); i++)
+            {
+                warning_str += (i > 0 ? ", " : "")+std::to_string(slow_op_primary_osds[i])+"\n";
+            }
+        }
+        if (slow_op_secondary_osds.size() > 0)
+        {
+            warning_str += "    "+std::to_string(slow_op_secondary_osds.size());
+            warning_str += (slow_op_secondary_osds.size() > 1 ? " osds have" : " osd has");
+            warning_str += " slow replication ops: ";
+            for (int i = 0; i < slow_op_secondary_osds.size(); i++)
+            {
+                warning_str += (i > 0 ? ", " : "")+std::to_string(slow_op_secondary_osds[i])+"\n";
+            }
+        }
         if (warning_str != "")
         {
             warning_str = "\n  warning:\n"+warning_str;

src/disk_tool/disk_tool_prepare.cpp

@@ -108,6 +108,10 @@ int disk_tool_t::prepare_one(std::map<std::string, std::string> options, int is_
     try
     {
         dsk.parse_config(options);
+        // Set all offsets to 4096 to calculate metadata size with excess
+        dsk.journal_offset = 4096;
+        dsk.meta_offset = 4096;
+        dsk.data_offset = 4096;
         dsk.data_io = dsk.meta_io = dsk.journal_io = (options["io"] == "cached" ? "cached" : "direct");
         dsk.open_data();
         dsk.open_meta();
@@ -171,8 +175,8 @@ int disk_tool_t::prepare_one(std::map<std::string, std::string> options, int is_
     }
     sb["osd_num"] = osd_num;
     // Zero out metadata and journal
-    if (write_zero(dsk.meta_fd, dsk.meta_offset, dsk.meta_len) != 0 ||
+    if (write_zero(dsk.meta_fd, sb["meta_offset"].uint64_value(), dsk.meta_len) != 0 ||
-        write_zero(dsk.journal_fd, dsk.journal_offset, dsk.journal_len) != 0)
+        write_zero(dsk.journal_fd, sb["journal_offset"].uint64_value(), dsk.journal_len) != 0)
     {
         fprintf(stderr, "Failed to zero out metadata or journal: %s\n", strerror(errno));
         dsk.close_all();
@@ -498,6 +502,9 @@ int disk_tool_t::get_meta_partition(std::vector<vitastor_dev_info_t> & ssds, std
     {
         blockstore_disk_t dsk;
         dsk.parse_config(options);
+        dsk.journal_offset = 4096;
+        dsk.meta_offset = 4096;
+        dsk.data_offset = 4096;
         dsk.data_io = dsk.meta_io = dsk.journal_io = "cached";
         dsk.open_data();
         dsk.open_meta();

src/osd/osd.cpp

@@ -535,10 +535,12 @@ void osd_t::print_stats()
 
 void osd_t::print_slow()
 {
-    bool has_slow = false;
+    cur_slow_op_primary = 0;
+    cur_slow_op_secondary = 0;
     char alloc[1024];
     timespec now;
     clock_gettime(CLOCK_REALTIME, &now);
+    // FIXME: Also track slow local blockstore ops and recovery/flush/scrub ops
     for (auto & kv: msgr.clients)
     {
         for (auto op: kv.second->received_ops)
@@ -608,6 +610,7 @@ void osd_t::print_slow()
                     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)
                 {
+                    cur_slow_op_secondary++;
                     bufprintf(" state=%d", op->bs_op ? PRIV(op->bs_op)->op_state : -1);
                     int wait_for = op->bs_op ? PRIV(op->bs_op)->wait_for : 0;
                     if (wait_for)
@@ -618,15 +621,19 @@ void osd_t::print_slow()
                 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)
                 {
+                    cur_slow_op_primary++;
                     bufprintf(" state=%d", !op->op_data ? -1 : op->op_data->st);
                 }
+                else
+                {
+                    cur_slow_op_primary++;
+                }
 #undef bufprintf
                 printf("%s\n", alloc);
-                has_slow = true;
             }
         }
     }
-    if (has_slow && bs)
+    if ((cur_slow_op_primary+cur_slow_op_secondary) > 0 && bs)
     {
         bs->dump_diagnostics();
     }

src/osd/osd.h

@@ -150,7 +150,9 @@ class osd_t
     bool pg_config_applied = false;
     bool etcd_reporting_pg_state = false;
     bool etcd_reporting_stats = false;
-    int autosync_timer_id = -1, print_stats_timer_id = -1, slow_log_timer_id = -1;
+    int print_stats_timer_id = -1, slow_log_timer_id = -1;
+    uint64_t cur_slow_op_primary = 0;
+    uint64_t cur_slow_op_secondary = 0;
 
     // peers and PGs
 
@@ -168,6 +170,8 @@ class osd_t
     object_id recovery_last_oid;
     int recovery_pg_done = 0, recovery_done = 0;
     osd_op_t *autosync_op = NULL;
+    int autosync_copies_to_delete = 0;
+    int autosync_timer_id = -1;
 
     // Scrubbing
     uint64_t scrub_nearest_ts = 0;
@@ -222,6 +226,7 @@ class osd_t
     void parse_config(bool init);
     void init_cluster();
     void on_change_osd_state_hook(osd_num_t peer_osd);
+    void on_change_backfillfull_hook(pool_id_t pool_id);
     void on_change_pg_history_hook(pool_id_t pool_id, pg_num_t pg_num);
     void on_change_etcd_state_hook(std::map<std::string, etcd_kv_t> & changes);
     void on_load_config_hook(json11::Json::object & changes);

src/osd/osd_cluster.cpp

@@ -65,6 +65,7 @@ void osd_t::init_cluster()
         st_cli.tfd = tfd;
         st_cli.log_level = log_level;
         st_cli.on_change_osd_state_hook = [this](osd_num_t peer_osd) { on_change_osd_state_hook(peer_osd); };
+        st_cli.on_change_backfillfull_hook = [this](pool_id_t pool_id) { on_change_backfillfull_hook(pool_id); };
         st_cli.on_change_pg_history_hook = [this](pool_id_t pool_id, pg_num_t pg_num) { on_change_pg_history_hook(pool_id, pg_num); };
         st_cli.on_change_hook = [this](std::map<std::string, etcd_kv_t> & changes) { on_change_etcd_state_hook(changes); };
         st_cli.on_load_config_hook = [this](json11::Json::object & cfg) { on_load_config_hook(cfg); };
@@ -201,6 +202,14 @@ json11::Json osd_t::get_statistics()
     st["immediate_commit"] = immediate_commit == IMMEDIATE_ALL ? "all" : (immediate_commit == IMMEDIATE_SMALL ? "small" : "none");
     st["host"] = self_state["host"];
     st["version"] = VITASTOR_VERSION;
+    if (cur_slow_op_primary > 0)
+    {
+        st["slow_ops_primary"] = cur_slow_op_primary;
+    }
+    if (cur_slow_op_secondary > 0)
+    {
+        st["slow_ops_secondary"] = cur_slow_op_secondary;
+    }
     json11::Json::object op_stats, subop_stats;
     for (int i = OSD_OP_MIN; i <= OSD_OP_MAX; i++)
     {
@@ -406,6 +415,14 @@ void osd_t::on_change_osd_state_hook(osd_num_t peer_osd)
     }
 }
 
+void osd_t::on_change_backfillfull_hook(pool_id_t pool_id)
+{
+    if (!(peering_state & (OSD_RECOVERING | OSD_FLUSHING_PGS)))
+    {
+        peering_state = peering_state | OSD_RECOVERING;
+    }
+}
+
 void osd_t::on_change_etcd_state_hook(std::map<std::string, etcd_kv_t> & changes)
 {
     if (changes.find(st_cli.etcd_prefix+"/config/global") != changes.end())

src/osd/osd_flush.cpp

@@ -13,10 +13,11 @@ void osd_t::submit_pg_flush_ops(pg_t & pg)
     bool first = true;
     while (it != pg.flush_actions.end())
     {
-        if (!first && (it->first.oid.inode != prev_it->first.oid.inode ||
+        if (!first &&
-            (it->first.oid.stripe & ~STRIPE_MASK) != (prev_it->first.oid.stripe & ~STRIPE_MASK)) &&
+            (it->first.oid.inode != prev_it->first.oid.inode ||
-            fb->rollback_lists[it->first.osd_num].size() >= FLUSH_BATCH ||
+                (it->first.oid.stripe & ~STRIPE_MASK) != (prev_it->first.oid.stripe & ~STRIPE_MASK)) &&
-            fb->stable_lists[it->first.osd_num].size() >= FLUSH_BATCH)
+            (fb->rollback_lists[it->first.osd_num].size() >= FLUSH_BATCH ||
+                fb->stable_lists[it->first.osd_num].size() >= FLUSH_BATCH))
         {
             // Stop only at the object boundary
             break;
@@ -75,6 +76,7 @@ void osd_t::handle_flush_op(bool rollback, pool_id_t pool_id, pg_num_t pg_num, p
         // Throw the result away
         return;
     }
+    fb->flush_done++;
     if (retval != 0)
     {
         if (peer_osd == this->osd_num)
@@ -92,12 +94,11 @@ void osd_t::handle_flush_op(bool rollback, pool_id_t pool_id, pg_num_t pg_num, p
             auto fd_it = msgr.osd_peer_fds.find(peer_osd);
             if (fd_it != msgr.osd_peer_fds.end())
             {
+                // Will repeer/stop this PG
                 msgr.stop_client(fd_it->second);
             }
-            return;
         }
     }
-    fb->flush_done++;
     if (fb->flush_done == fb->flush_ops)
     {
         // This flush batch is done
@@ -251,10 +252,18 @@ bool osd_t::pick_next_recovery(osd_recovery_op_t &op)
             auto mask = recovery_last_degraded ? (PG_ACTIVE | PG_HAS_DEGRADED) : (PG_ACTIVE | PG_DEGRADED | PG_HAS_MISPLACED);
             auto check = recovery_last_degraded ? (PG_ACTIVE | PG_HAS_DEGRADED) : (PG_ACTIVE | PG_HAS_MISPLACED);
             // Restart scanning from the same PG as the last time
+        restart:
             for (auto pg_it = pgs.lower_bound(recovery_last_pg); pg_it != pgs.end(); pg_it++)
             {
                 if ((pg_it->second.state & mask) == check)
                 {
+                    auto pool_it = st_cli.pool_config.find(pg_it->first.pool_id);
+                    if (pool_it != st_cli.pool_config.end() && pool_it->second.backfillfull)
+                    {
+                        // Skip the pool
+                        recovery_last_pg.pool_id++;
+                        goto restart;
+                    }
                     auto & src = recovery_last_degraded ? pg_it->second.degraded_objects : pg_it->second.misplaced_objects;
                     assert(src.size() > 0);
                     // Restart scanning from the next object

src/osd/osd_primary.cpp

@@ -645,6 +645,18 @@ void osd_t::remove_object_from_state(object_id & oid, pg_osd_set_state_t **objec
     {
         throw std::runtime_error("BUG: Invalid object state: "+std::to_string((*object_state)->state));
     }
+    if (changed && immediate_commit != IMMEDIATE_ALL)
+    {
+        // Trigger double automatic sync after changing PG state when we're running with fsyncs.
+        // First autosync commits all written objects and applies copies_to_delete_after_sync;
+        // Second autosync commits all deletions run by the first sync.
+        // Without it, rebalancing in a cluster without load may result in some small amount of
+        // garbage left on "extra" OSDs of the PG, because last deletions are not synced at all.
+        // FIXME: 1000% correct way is to switch PG state only after copies_to_delete_after_sync.
+        // But it's much more complicated.
+        unstable_write_count += autosync_writes;
+        autosync_copies_to_delete = 2;
+    }
     if (changed && report)
     {
         report_pg_state(pg);

src/osd/osd_primary_subops.cpp

@@ -9,6 +9,10 @@ void osd_t::autosync()
 {
     if (immediate_commit != IMMEDIATE_ALL && !autosync_op)
     {
+        if (autosync_copies_to_delete > 0)
+        {
+            autosync_copies_to_delete--;
+        }
         autosync_op = new osd_op_t();
         autosync_op->op_type = OSD_OP_IN;
         autosync_op->peer_fd = SELF_FD;
@@ -29,6 +33,11 @@ void osd_t::autosync()
             }
             delete autosync_op;
             autosync_op = NULL;
+            if (autosync_copies_to_delete > 0)
+            {
+                // Trigger the second "copies_to_delete" autosync
+                autosync();
+            }
         };
         exec_op(autosync_op);
     }

src/osd/osd_primary_sync.cpp

@@ -213,6 +213,15 @@ resume_8:
         {
             goto resume_6;
         }
+        if (immediate_commit == IMMEDIATE_NONE)
+        {
+            // Mark OSDs as dirty because deletions have to be synced too!
+            for (int i = 0; i < op_data->copies_to_delete_count; i++)
+            {
+                auto & chunk = op_data->copies_to_delete[i];
+                this->dirty_osds.insert(chunk.osd_num);
+            }
+        }
     }
     for (int i = 0; i < op_data->dirty_pg_count; i++)
     {
@@ -227,7 +236,7 @@ resume_8:
             start_pg_peering(pg);
         }
     }
-    // FIXME: Free those in the destructor?
+    // FIXME: Free those in the destructor (not here)?
     free(op_data->dirty_pgs);
     op_data->dirty_pgs = NULL;
     op_data->dirty_osds = NULL;

src/osd/osd_primary_write.cpp

@@ -7,6 +7,12 @@
 bool osd_t::check_write_queue(osd_op_t *cur_op, pg_t & pg)
 {
     osd_primary_op_data_t *op_data = cur_op->op_data;
+    // First check if PG is not active anymore
+    if (!(pg.state & PG_ACTIVE))
+    {
+        pg_cancel_write_queue(pg, cur_op, op_data->oid, -EPIPE);
+        return false;
+    }
     // Check if actions are pending for this object
     auto act_it = pg.flush_actions.lower_bound((obj_piece_id_t){
         .oid = op_data->oid,

src/util/addr_util.cpp

@@ -65,7 +65,7 @@ std::string addr_to_string(const sockaddr_storage &addr)
     return std::string(peer_str)+":"+std::to_string(port);
 }
 
-static bool cidr_match(const in_addr &addr, const in_addr &net, uint8_t bits)
+bool cidr_match(const in_addr &addr, const in_addr &net, uint8_t bits)
 {
     if (bits == 0)
     {
@@ -75,7 +75,7 @@ static bool cidr_match(const in_addr &addr, const in_addr &net, uint8_t bits)
     return !((addr.s_addr ^ net.s_addr) & htonl(0xFFFFFFFFu << (32 - bits)));
 }
 
-static bool cidr6_match(const in6_addr &address, const in6_addr &network, uint8_t bits)
+bool cidr6_match(const in6_addr &address, const in6_addr &network, uint8_t bits)
 {
     const uint32_t *a = address.s6_addr32;
     const uint32_t *n = network.s6_addr32;
@@ -93,47 +93,49 @@ static bool cidr6_match(const in6_addr &address, const in6_addr &network, uint8_
     return true;
 }
 
-struct addr_mask_t
+addr_mask_t cidr_parse(std::string mask)
 {
-    sa_family_t family;
+    unsigned bits = 255;
+    int p = mask.find('/');
+    if (p != std::string::npos)
+    {
+        char null_byte = 0;
+        if (sscanf(mask.c_str()+p+1, "%u%c", &bits, &null_byte) != 1 || bits > 128)
+            throw std::runtime_error("Invalid IP address mask: " + mask);
+        mask = mask.substr(0, p);
+    }
     in_addr ipv4;
     in6_addr ipv6;
-    uint8_t bits;
+    if (inet_pton(AF_INET, mask.c_str(), &ipv4) == 1)
-};
+    {
+        if (bits == 255)
+            bits = 32;
+        if (bits > 32)
+            throw std::runtime_error("Invalid IP address mask: " + mask);
+        return (addr_mask_t){ .family = AF_INET, .ipv4 = ipv4, .bits = (uint8_t)(bits ? bits : 32) };
+    }
+    else if (inet_pton(AF_INET6, mask.c_str(), &ipv6) == 1)
+    {
+        if (bits == 255)
+            bits = 128;
+        return (addr_mask_t){ .family = AF_INET6, .ipv6 = ipv6, .bits = (uint8_t)bits };
+    }
+    else
+    {
+        throw std::runtime_error("Invalid IP address mask: " + mask);
+    }
+}
 
 std::vector<std::string> getifaddr_list(std::vector<std::string> mask_cfg, bool include_v6)
 {
     std::vector<addr_mask_t> masks;
     for (auto mask: mask_cfg)
     {
-        unsigned bits = 0;
+        masks.push_back(cidr_parse(mask));
-        int p = mask.find('/');
+        if (masks[masks.size()-1].family == AF_INET6)
-        if (p != std::string::npos)
         {
-            char null_byte = 0;
+            // Auto-enable IPv6 addresses
-            if (sscanf(mask.c_str()+p+1, "%u%c", &bits, &null_byte) != 1 || bits > 128)
+            include_v6 = true;
-            {
-                throw std::runtime_error((include_v6 ? "Invalid IPv4 address mask: " : "Invalid IP address mask: ") + mask);
-            }
-            mask = mask.substr(0, p);
-        }
-        in_addr ipv4;
-        in6_addr ipv6;
-        if (inet_pton(AF_INET, mask.c_str(), &ipv4) == 1)
-        {
-            if (bits > 32)
-            {
-                throw std::runtime_error((include_v6 ? "Invalid IPv4 address mask: " : "Invalid IP address mask: ") + mask);
-            }
-            masks.push_back((addr_mask_t){ .family = AF_INET, .ipv4 = ipv4, .bits = (uint8_t)bits });
-        }
-        else if (include_v6 && inet_pton(AF_INET6, mask.c_str(), &ipv6) == 1)
-        {
-            masks.push_back((addr_mask_t){ .family = AF_INET6, .ipv6 = ipv6, .bits = (uint8_t)bits });
-        }
-        else
-        {
-            throw std::runtime_error((include_v6 ? "Invalid IPv4 address mask: " : "Invalid IP address mask: ") + mask);
         }
     }
     std::set<std::string> addresses;

src/util/addr_util.h

@@ -1,10 +1,22 @@
 #pragma once
 
+#include <netinet/in.h>
 #include <sys/socket.h>
 #include <string>
 #include <vector>
 
+struct addr_mask_t
+{
+    sa_family_t family;
+    in_addr ipv4;
+    in6_addr ipv6;
+    uint8_t bits;
+};
+
 bool string_to_addr(std::string str, bool parse_port, int default_port, struct sockaddr_storage *addr);
 std::string addr_to_string(const sockaddr_storage &addr);
+addr_mask_t cidr_parse(std::string mask);
+bool cidr_match(const in_addr &address, const in_addr &network, uint8_t bits);
+bool cidr6_match(const in6_addr &address, const in6_addr &network, uint8_t bits);
 std::vector<std::string> getifaddr_list(std::vector<std::string> mask_cfg = std::vector<std::string>(), bool include_v6 = false);
 int create_and_bind_socket(std::string bind_address, int bind_port, int listen_backlog, int *listening_port);

src/util/timerfd_manager.cpp

@@ -62,7 +62,7 @@ int timerfd_manager_t::set_timer_us(uint64_t micros, bool repeat, std::function<
         .callback = callback,
     });
     inc_timer(timers[timers.size()-1]);
-    set_nearest();
+    set_nearest(false);
     return timer_id;
 }
 
@@ -82,13 +82,13 @@ void timerfd_manager_t::clear_timer(int timer_id)
             {
                 nearest--;
             }
-            set_nearest();
+            set_nearest(false);
             break;
         }
     }
 }
 
-void timerfd_manager_t::set_nearest()
+void timerfd_manager_t::set_nearest(bool trigger_inline)
 {
     if (onstack > 0)
     {
@@ -134,10 +134,13 @@ again:
         }
         if (exp.it_value.tv_sec < 0 || exp.it_value.tv_sec == 0 && exp.it_value.tv_nsec <= 0)
         {
-            // It already happened
+            // It already happened - set minimal timeout
-            // FIXME: Postpone to setImmediate/BH to avoid reenterability problems
+            if (trigger_inline)
-            trigger_nearest();
+            {
-            goto again;
+                trigger_nearest();
+                goto again;
+            }
+            exp.it_value = { .tv_sec = 0, .tv_nsec = 1 };
         }
         if (timerfd_settime(timerfd, 0, &exp, NULL))
         {
@@ -157,7 +160,7 @@ void timerfd_manager_t::handle_readable()
         trigger_nearest();
     }
     wait_state = 0;
-    set_nearest();
+    set_nearest(true);
 }
 
 void timerfd_manager_t::trigger_nearest()

src/util/timerfd_manager.h

@@ -26,7 +26,7 @@ class timerfd_manager_t
     std::vector<timerfd_timer_t> timers;
 
     void inc_timer(timerfd_timer_t & t);
-    void set_nearest();
+    void set_nearest(bool trigger_inline);
     void trigger_nearest();
     void handle_readable();
 public:

tests/run_tests.sh

@@ -19,9 +19,12 @@ ANTIETCD=1 ./test_etcd_fail.sh
 
 ./test_interrupted_rebalance.sh
 IMMEDIATE_COMMIT=1 ./test_interrupted_rebalance.sh
+
 SCHEME=ec ./test_interrupted_rebalance.sh
 SCHEME=ec IMMEDIATE_COMMIT=1 ./test_interrupted_rebalance.sh
 
+./test_create_halfhost.sh
+
 ./test_failure_domain.sh
 
 ./test_snapshot.sh

tests/test_create_halfhost.sh

@@ -0,0 +1,35 @@
+#!/bin/bash -ex
+
+. `dirname $0`/common.sh
+
+node mon/mon-main.js $MON_PARAMS --etcd_address $ETCD_URL --etcd_prefix "/vitastor" >>./testdata/mon.log 2>&1 &
+MON_PID=$!
+wait_etcd
+
+TIME=$(date '+%s')
+$ETCDCTL put /vitastor/config/global '{"placement_levels":{"dc":10,"host":100,"half":105,"osd":110}}'
+$ETCDCTL put /vitastor/config/node_placement '{
+  "h11":{"level":"half","parent":"host1"},
+  "h12":{"level":"half","parent":"host1"},
+  "h21":{"level":"half","parent":"host2"},
+  "h22":{"level":"half","parent":"host2"},
+  "h31":{"level":"half","parent":"host3"},
+  "h32":{"level":"half","parent":"host3"},
+  "1":{"parent":"h11"},
+  "2":{"parent":"h12"},
+  "3":{"parent":"h21"},
+  "4":{"parent":"h22"},
+  "5":{"parent":"h31"},
+  "6":{"parent":"h32"}
+}'
+$ETCDCTL put /vitastor/osd/stats/1 '{"host":"host1","size":1073741824,"time":"'$TIME'"}'
+$ETCDCTL put /vitastor/osd/stats/2 '{"host":"host1","size":1073741824,"time":"'$TIME'"}'
+$ETCDCTL put /vitastor/osd/stats/3 '{"host":"host2","size":1073741824,"time":"'$TIME'"}'
+$ETCDCTL put /vitastor/osd/stats/4 '{"host":"host2","size":1073741824,"time":"'$TIME'"}'
+$ETCDCTL put /vitastor/osd/stats/5 '{"host":"host3","size":1073741824,"time":"'$TIME'"}'
+$ETCDCTL put /vitastor/osd/stats/6 '{"host":"host3","size":1073741824,"time":"'$TIME'"}'
+build/src/cmd/vitastor-cli --etcd_address $ETCD_URL osd-tree
+# check that it doesn't fail
+build/src/cmd/vitastor-cli --etcd_address $ETCD_URL create-pool testpool --ec 2+1 -n 32
+
+format_green OK

tests/test_heal.sh

@@ -53,7 +53,7 @@ kill_osds &
 
 LD_PRELOAD="build/src/client/libfio_vitastor.so" \
     fio -thread -name=test -ioengine=build/src/client/libfio_vitastor.so -bsrange=4k-128k -blockalign=4k -direct=1 -iodepth=32 -fsync=256 -rw=randrw \
-        -randrepeat=0 -refill_buffers=1 -mirror_file=./testdata/bin/mirror.bin -etcd=$ETCD_URL -image=testimg -loops=10 -runtime=120
+        -serialize_overlap=1 -randrepeat=0 -refill_buffers=1 -mirror_file=./testdata/bin/mirror.bin -etcd=$ETCD_URL -image=testimg -loops=10 -runtime=120
 
 qemu-img convert -S 4096 -p \
     -f raw "vitastor:etcd_host=127.0.0.1\:$ETCD_PORT/v3:image=testimg" \