Small fix due to the new functionality in cinder "Skip sparse copy during volume reimage" 8d7e292bcd

SYNC operation fsyncs only completed operations, so treating writes as "eligible
for fsync" before actually completing them is incorrect

It affected SCHEME=ec test_heal.sh (with immediate_commit=none) test - it was
flapping with lost writes - some non-fsynced writes were legitimately lost by
the OSD, but weren't repeated by the client

.gitignore

@@ -3,16 +3,3 @@
 package-lock.json
 fio
 qemu
-osd
-stub_osd
-stub_uring_osd
-stub_bench
-osd_test
-osd_peering_pg_test
-dump_journal
-nbd_proxy
-rm_inode
-test_allocator
-test_blockstore
-test_shit
-osd_rmw_test

copy-fio-includes.sh

@@ -1,6 +1,6 @@
 #!/bin/bash
 
-gcc -I. -E -o fio_headers.i src/fio_headers.h
+gcc -I. -E -o fio_headers.i src/util/fio_headers.h
 
 rm -rf fio-copy
 for i in `grep -Po 'fio/[^"]+' fio_headers.i | sort | uniq`; do

copy-qemu-includes.sh

@@ -5,7 +5,7 @@
 #cd b/qemu; make qapi
 
 gcc -I qemu/b/qemu `pkg-config glib-2.0 --cflags` \
-    -I qemu/include -E -o qemu_driver.i src/qemu_driver.c
+    -I qemu/include -E -o qemu_driver.i src/client/qemu_driver.c
 
 rm -rf qemu-copy
 for i in `grep -Po 'qemu/[^"]+' qemu_driver.i | sort | uniq`; do

debian/patched-qemu.Dockerfile

@@ -27,7 +27,7 @@ RUN apt-get -y build-dep qemu
 RUN apt-get --download-only source qemu
 
 ADD patches /root/vitastor/patches
-ADD src/qemu_driver.c /root/vitastor/src/qemu_driver.c
+ADD src/client/qemu_driver.c /root/qemu_driver.c
 
 #RUN set -e; \
 #    apt-get install -y wget; \
@@ -52,7 +52,7 @@ RUN set -e; \
     cd /root/packages/qemu-$REL/qemu-*/; \
     quilt push -a; \
     quilt add block/vitastor.c; \
-    cp /root/vitastor/src/qemu_driver.c block/vitastor.c; \
+    cp /root/qemu_driver.c block/vitastor.c; \
     quilt refresh; \
     V=$(head -n1 debian/changelog | perl -pe 's/5\.2\+dfsg-9/5.2+dfsg-11/; s/^.*\((.*?)(~bpo[\d\+]*)?\).*$/$1/')+vitastor4; \
     if [ "$REL" = bullseye ]; then V=${V}bullseye; fi; \

debian/vitastor-mon.install

@@ -1,2 +1,3 @@
-mon usr/lib/vitastor
-mon/vitastor-mon.service /lib/systemd/system
+mon usr/lib/vitastor/mon
+mon/scripts/make-etcd usr/lib/vitastor/mon
+mon/scripts/vitastor-mon.service /lib/systemd/system

debian/vitastor-osd.install

@@ -1,6 +1,6 @@
 usr/bin/vitastor-osd
 usr/bin/vitastor-disk
 usr/bin/vitastor-dump-journal
-mon/vitastor-osd@.service /lib/systemd/system
-mon/vitastor.target /lib/systemd/system
-mon/90-vitastor.rules /lib/udev/rules.d
+mon/scripts/vitastor-osd@.service /lib/systemd/system
+mon/scripts/vitastor.target /lib/systemd/system
+mon/scripts/90-vitastor.rules /lib/udev/rules.d

docs/config/network.en.md

@@ -248,7 +248,7 @@ etcd_report_interval to guarantee that keepalive actually works.
 ## etcd_ws_keepalive_interval
 
 - Type: seconds
-- Default: 30
+- Default: 5
 - Can be changed online: yes
 
 etcd websocket ping interval required to keep the connection alive and

docs/config/network.ru.md

@@ -259,7 +259,7 @@ etcd_report_interval, чтобы keepalive гарантированно рабо
 ## etcd_ws_keepalive_interval
 
 - Тип: секунды
-- Значение по умолчанию: 30
+- Значение по умолчанию: 5
 - Можно менять на лету: да
 
 Интервал проверки живости вебсокет-подключений к etcd.

docs/config/src/network.yml

@@ -282,7 +282,7 @@
     etcd_report_interval, чтобы keepalive гарантированно работал.
 - name: etcd_ws_keepalive_interval
   type: sec
-  default: 30
+  default: 5
   online: true
   info: |
     etcd websocket ping interval required to keep the connection alive and

docs/installation/openstack.en.md

@@ -11,7 +11,8 @@ To enable Vitastor support in an OpenStack installation:
 - Install vitastor-client, patched QEMU and libvirt packages from Vitastor DEB or RPM repository
 - Use `patches/nova-21.diff` or `patches/nova-23.diff` to patch your Nova installation.
   Patch 21 fits Nova 21-22, patch 23 fits Nova 23-24.
-- Install `patches/cinder-vitastor.py` as `..../cinder/volume/drivers/vitastor.py`
+- Install `patches/cinder-vitastor-21.py` or `pathces/cinder-vitastor-22.py` as `..../cinder/volume/drivers/vitastor.py`
+  Patch 21 fits Cinder up 21 (zed), Patch 22 fits Cinder after 22 (2023.1)
 - Define a volume type in cinder.conf (see below)
 - Block network access from VMs to Vitastor network (to OSDs and etcd),
   because Vitastor doesn't support authentication

docs/installation/openstack.ru.md

@@ -11,7 +11,8 @@
 - Установите пакеты vitastor-client, libvirt и QEMU из DEB или RPM репозитория Vitastor
 - Примените патч `patches/nova-21.diff` или `patches/nova-23.diff` к вашей инсталляции Nova.
   nova-21.diff подходит для Nova 21-22, nova-23.diff подходит для Nova 23-24.
-- Скопируйте `patches/cinder-vitastor.py` в инсталляцию Cinder как `cinder/volume/drivers/vitastor.py`
+- Скопируйте `patches/cinder-vitastor-21.py` или `pathces/cinder-vitastor-22.py` в инсталляцию Cinder как `cinder/volume/drivers/vitastor.py`.
+  `cinder-vitastor-21.py` подходит для Cinder 21 (zed) и младше, `cinder-vitastor-22.py` подходит для Cinder 22 (2023.1) и старше.
 - Создайте тип томов в cinder.conf (см. ниже)
 - Обязательно заблокируйте доступ от виртуальных машин к сети Vitastor (OSD и etcd), т.к. Vitastor (пока) не поддерживает аутентификацию
 - Перезапустите Cinder и Nova

docs/installation/source.en.md

@@ -41,7 +41,7 @@ It's recommended to build the QEMU driver (qemu_driver.c) in-tree, as a part of
 QEMU build process. To do that:
 - Install vitastor client library headers (from source or from vitastor-client-dev package)
 - Take a corresponding patch from `patches/qemu-*-vitastor.patch` and apply it to QEMU source
-- Copy `src/qemu_driver.c` to QEMU source directory as `block/vitastor.c`
+- Copy `src/client/qemu_driver.c` to QEMU source directory as `block/vitastor.c`
 - Build QEMU as usual
 
 But it is also possible to build it out-of-tree. To do that:

docs/installation/source.ru.md

@@ -41,7 +41,7 @@ cmake .. && make -j8 install
 Драйвер QEMU (qemu_driver.c) рекомендуется собирать вместе с самим QEMU. Для этого:
 - Установите заголовки клиентской библиотеки Vitastor (из исходников или из пакета vitastor-client-dev)
 - Возьмите соответствующий патч из `patches/qemu-*-vitastor.patch` и примените его к исходникам QEMU
-- Скопируйте [src/qemu_driver.c](../../src/qemu_driver.c) в директорию исходников QEMU как `block/vitastor.c`
+- Скопируйте [src/client/qemu_driver.c](../../src/client/qemu_driver.c) в директорию исходников QEMU как `block/vitastor.c`
 - Соберите QEMU как обычно
 
 Однако в целях отладки драйвер также можно собирать отдельно от QEMU. Для этого:

docs/intro/quickstart.en.md

@@ -22,7 +22,7 @@
   with lazy fsync, but prepare for inferior single-thread latency. Read more about capacitors
   [here](../config/layout-cluster.en.md#immediate_commit).
 - If you want to use HDDs, get modern HDDs with Media Cache or SSD Cache: HGST Ultrastar,
-  Toshiba MG08, Seagate EXOS or something similar. If your drives don't have such cache then
+  Toshiba MG, Seagate EXOS or something similar. If your drives don't have such cache then
   you also need small SSDs for journal and metadata (even 2 GB per 1 TB of HDD space is enough).
 - Get a fast network (at least 10 Gbit/s). Something like Mellanox ConnectX-4 with RoCEv2 is ideal.
 - Disable CPU powersaving: `cpupower idle-set -D 0 && cpupower frequency-set -g performance`.
@@ -33,7 +33,7 @@
 - SATA SSD: Micron 5100/5200/5300/5400, Samsung PM863/PM883/PM893, Intel D3-S4510/4520/4610/4620, Kingston DC500M
 - NVMe: Micron 9100/9200/9300/9400, Micron 7300/7450, Samsung PM983/PM9A3, Samsung PM1723/1735/1743,
   Intel DC-P3700/P4500/P4600, Intel D7-P5500/P5600, Intel Optane, Kingston DC1000B/DC1500M
-- HDD: HGST Ultrastar, Toshiba MG06/MG07/MG08, Seagate EXOS
+- HDD: HGST Ultrastar, Toshiba MG, Seagate EXOS
 
 ## Configure monitors
 

docs/intro/quickstart.ru.md

@@ -123,4 +123,4 @@ vitastor-cli create -s 10G testimg
 Если вы хотите использовать не только блочные образы виртуальных машин или контейнеров,
 а также кластерную файловую систему, то:
 
-- [Следуйте инструкциям](../usage/nfs.en.md#vitastorfs)
+- [Следуйте инструкциям](../usage/nfs.ru.md#vitastorfs)

mon/.eslintrc.js

@@ -11,6 +11,7 @@ module.exports = {
         "ecmaVersion": 2020
     },
     "plugins": [
+        "import"
     ],
     "rules": {
         "indent": [
@@ -44,6 +45,10 @@ module.exports = {
         ],
         "node/shebang": [
             "off"
+        ],
+        "import/no-unresolved": [
+            2,
+            { "commonjs": true }
         ]
     }
 };

mon/etcd_adapter.js

@@ -0,0 +1,356 @@
+// Copyright (c) Vitaliy Filippov, 2019+
+// License: VNPL-1.1 (see README.md for details)
+
+const http = require('http');
+const WebSocket = require('ws');
+
+const MON_STOPPED = 'Monitor instance is stopped';
+
+class EtcdAdapter
+{
+    constructor(mon)
+    {
+        this.mon = mon;
+        this.ws = null;
+        this.ws_alive = false;
+        this.ws_keepalive_timer = null;
+    }
+
+    parse_config(config)
+    {
+        this.parse_etcd_addresses(config.etcd_address||config.etcd_url);
+    }
+
+    parse_etcd_addresses(addrs)
+    {
+        const is_local_ip = this.mon.local_ips(true).reduce((a, c) => { a[c] = true; return a; }, {});
+        this.etcd_local = [];
+        this.etcd_urls = [];
+        this.selected_etcd_url = null;
+        this.etcd_urls_to_try = [];
+        if (!(addrs instanceof Array))
+            addrs = addrs ? (''+(addrs||'')).split(/,/) : [];
+        if (!addrs.length)
+        {
+            console.error('Vitastor etcd address(es) not specified. Please set on the command line or in the config file');
+            process.exit(1);
+        }
+        for (let url of addrs)
+        {
+            let scheme = 'http';
+            url = url.trim().replace(/^(https?):\/\//, (m, m1) => { scheme = m1; return ''; });
+            const slash = url.indexOf('/');
+            const colon = url.indexOf(':');
+            const is_local = is_local_ip[colon >= 0 ? url.substr(0, colon) : (slash >= 0 ? url.substr(0, slash) : url)];
+            url = scheme+'://'+(slash >= 0 ? url : url+'/v3');
+            if (is_local)
+                this.etcd_local.push(url);
+            else
+                this.etcd_urls.push(url);
+        }
+    }
+
+    pick_next_etcd()
+    {
+        if (this.selected_etcd_url)
+            return this.selected_etcd_url;
+        if (!this.etcd_urls_to_try || !this.etcd_urls_to_try.length)
+        {
+            this.etcd_urls_to_try = [ ...this.etcd_local ];
+            const others = [ ...this.etcd_urls ];
+            while (others.length)
+            {
+                const url = others.splice(0|(others.length*Math.random()), 1);
+                this.etcd_urls_to_try.push(url[0]);
+            }
+        }
+        this.selected_etcd_url = this.etcd_urls_to_try.shift();
+        return this.selected_etcd_url;
+    }
+
+    stop_watcher(cur_addr)
+    {
+        cur_addr = cur_addr || this.selected_etcd_url;
+        if (this.ws)
+        {
+            console.log('Disconnected from etcd at '+this.ws_used_url);
+            this.ws.close();
+            this.ws = null;
+        }
+        if (this.ws_keepalive_timer)
+        {
+            clearInterval(this.ws_keepalive_timer);
+            this.ws_keepalive_timer = null;
+        }
+        if (this.selected_etcd_url == cur_addr)
+        {
+            this.selected_etcd_url = null;
+        }
+    }
+
+    restart_watcher(cur_addr)
+    {
+        this.stop_watcher(cur_addr);
+        this.start_watcher(this.mon.config.etcd_mon_retries).catch(this.mon.die);
+    }
+
+    async start_watcher(retries)
+    {
+        let retry = 0;
+        if (!retries || retries < 1)
+        {
+            retries = 1;
+        }
+        const tried = {};
+        while (retries < 0 || retry < retries)
+        {
+            const cur_addr = this.pick_next_etcd();
+            const base = 'ws'+cur_addr.substr(4);
+            let now = Date.now();
+            if (tried[base] && now-tried[base] < this.mon.config.etcd_start_timeout)
+            {
+                await new Promise(ok => setTimeout(ok, this.mon.config.etcd_start_timeout-(now-tried[base])));
+                now = Date.now();
+            }
+            tried[base] = now;
+            if (this.mon.stopped)
+            {
+                return;
+            }
+            const ok = await new Promise(ok =>
+            {
+                const timer_id = setTimeout(() =>
+                {
+                    if (this.ws)
+                    {
+                        console.log('Disconnected from etcd at '+this.ws_used_url);
+                        this.ws.close();
+                        this.ws = null;
+                    }
+                    ok(false);
+                }, this.mon.config.etcd_mon_timeout);
+                this.ws = new WebSocket(base+'/watch');
+                this.ws_used_url = cur_addr;
+                const fail = () =>
+                {
+                    ok(false);
+                };
+                this.ws.on('error', fail);
+                this.ws.on('open', () =>
+                {
+                    this.ws.removeListener('error', fail);
+                    if (timer_id)
+                        clearTimeout(timer_id);
+                    ok(true);
+                });
+            });
+            if (ok)
+                break;
+            if (this.selected_etcd_url == cur_addr)
+                this.selected_etcd_url = null;
+            this.ws = null;
+            retry++;
+        }
+        if (!this.ws)
+        {
+            this.mon.die('Failed to open etcd watch websocket');
+            return;
+        }
+        if (this.mon.stopped)
+        {
+            this.stop_watcher();
+            return;
+        }
+        const cur_addr = this.selected_etcd_url;
+        this.ws_alive = true;
+        this.ws_keepalive_timer = setInterval(() =>
+        {
+            if (this.ws_alive && this.ws)
+            {
+                this.ws_alive = false;
+                this.ws.send(JSON.stringify({ progress_request: {} }));
+            }
+            else
+            {
+                console.log('etcd websocket timed out, restarting it');
+                this.restart_watcher(cur_addr);
+            }
+        }, (Number(this.mon.config.etcd_ws_keepalive_interval) || 5)*1000);
+        this.ws.on('error', () => this.restart_watcher(cur_addr));
+        this.ws.send(JSON.stringify({
+            create_request: {
+                key: b64(this.mon.config.etcd_prefix+'/'),
+                range_end: b64(this.mon.config.etcd_prefix+'0'),
+                start_revision: ''+this.mon.etcd_watch_revision,
+                watch_id: 1,
+                progress_notify: true,
+            },
+        }));
+        this.ws.on('message', (msg) =>
+        {
+            if (this.mon.stopped)
+            {
+                this.stop_watcher();
+                return;
+            }
+            this.ws_alive = true;
+            let data;
+            try
+            {
+                data = JSON.parse(msg);
+            }
+            catch (e)
+            {
+            }
+            if (!data || !data.result)
+            {
+                console.error('Unknown message received from watch websocket: '+msg);
+            }
+            else if (data.result.canceled)
+            {
+                // etcd watch canceled
+                if (data.result.compact_revision)
+                {
+                    // we may miss events if we proceed
+                    this.mon.die('Revisions before '+data.result.compact_revision+' were compacted by etcd, exiting');
+                }
+                this.mon.die('Watch canceled by etcd, reason: '+data.result.cancel_reason+', exiting');
+            }
+            else if (data.result.created)
+            {
+                // etcd watch created
+                console.log('Successfully subscribed to etcd at '+this.selected_etcd_url+', revision '+data.result.header.revision);
+            }
+            else
+            {
+                this.mon.on_message(data.result);
+            }
+        });
+    }
+
+    async become_master()
+    {
+        const state = { ...this.mon.get_mon_state(), id: ''+this.mon.etcd_lease_id };
+        // eslint-disable-next-line no-constant-condition
+        while (1)
+        {
+            const res = await this.etcd_call('/kv/txn', {
+                compare: [ { target: 'CREATE', create_revision: 0, key: b64(this.mon.config.etcd_prefix+'/mon/master') } ],
+                success: [ { requestPut: { key: b64(this.mon.config.etcd_prefix+'/mon/master'), value: b64(JSON.stringify(state)), lease: ''+this.mon.etcd_lease_id } } ],
+            }, this.mon.config.etcd_start_timeout, 0);
+            if (res.succeeded)
+            {
+                break;
+            }
+            console.log('Waiting to become master');
+            await new Promise(ok => setTimeout(ok, this.mon.config.etcd_start_timeout));
+        }
+        console.log('Became master');
+    }
+
+    async etcd_call(path, body, timeout, retries)
+    {
+        let retry = 0;
+        if (retries >= 0 && retries < 1)
+        {
+            retries = 1;
+        }
+        const tried = {};
+        while (retries < 0 || retry < retries)
+        {
+            retry++;
+            const base = this.pick_next_etcd();
+            let now = Date.now();
+            if (tried[base] && now-tried[base] < timeout)
+            {
+                await new Promise(ok => setTimeout(ok, timeout-(now-tried[base])));
+                now = Date.now();
+            }
+            tried[base] = now;
+            if (this.mon.stopped)
+            {
+                throw new Error(MON_STOPPED);
+            }
+            const res = await POST(base+path, body, timeout);
+            if (this.mon.stopped)
+            {
+                throw new Error(MON_STOPPED);
+            }
+            if (res.error)
+            {
+                if (this.selected_etcd_url == base)
+                    this.selected_etcd_url = null;
+                console.error('Failed to query etcd '+path+' (retry '+retry+'/'+retries+'): '+res.error);
+                continue;
+            }
+            if (res.json)
+            {
+                if (res.json.error)
+                {
+                    console.error(path+': etcd returned error: '+res.json.error);
+                    break;
+                }
+                return res.json;
+            }
+        }
+        throw new Error('Failed to query etcd ('+retries+' retries)');
+    }
+}
+
+function POST(url, body, timeout)
+{
+    return new Promise(ok =>
+    {
+        const body_text = Buffer.from(JSON.stringify(body));
+        let timer_id = timeout > 0 ? setTimeout(() =>
+        {
+            if (req)
+                req.abort();
+            req = null;
+            ok({ error: 'timeout' });
+        }, timeout) : null;
+        let req = http.request(url, { method: 'POST', headers: {
+            'Content-Type': 'application/json',
+            'Content-Length': body_text.length,
+        } }, (res) =>
+        {
+            if (!req)
+            {
+                return;
+            }
+            clearTimeout(timer_id);
+            let res_body = '';
+            res.setEncoding('utf8');
+            res.on('error', (error) => ok({ error }));
+            res.on('data', chunk => { res_body += chunk; });
+            res.on('end', () =>
+            {
+                if (res.statusCode != 200)
+                {
+                    ok({ error: res_body, code: res.statusCode });
+                    return;
+                }
+                try
+                {
+                    res_body = JSON.parse(res_body);
+                    ok({ response: res, json: res_body });
+                }
+                catch (e)
+                {
+                    ok({ error: e, response: res, body: res_body });
+                }
+            });
+        });
+        req.on('error', (error) => ok({ error }));
+        req.on('close', () => ok({ error: new Error('Connection closed prematurely') }));
+        req.write(body_text);
+        req.end();
+    });
+}
+
+function b64(str)
+{
+    return Buffer.from(str).toString('base64');
+}
+
+module.exports = EtcdAdapter;

mon/etcd_schema.js

@@ -0,0 +1,391 @@
+// Copyright (c) Vitaliy Filippov, 2019+
+// License: VNPL-1.1 (see README.md for details)
+
+// FIXME document all etcd keys and config variables in the form of JSON schema or similar
+const etcd_nonempty_keys = {
+    'config/global': 1,
+    'config/node_placement': 1,
+    'config/pools': 1,
+    'config/pgs': 1,
+    'history/last_clean_pgs': 1,
+    'stats': 1,
+};
+const etcd_allow = new RegExp('^'+[
+    'config/global',
+    'config/node_placement',
+    'config/pools',
+    'config/osd/[1-9]\\d*',
+    'config/pgs',
+    'config/inode/[1-9]\\d*/[1-9]\\d*',
+    'osd/state/[1-9]\\d*',
+    'osd/stats/[1-9]\\d*',
+    'osd/inodestats/[1-9]\\d*',
+    'osd/space/[1-9]\\d*',
+    'mon/master',
+    'mon/member/[a-f0-9]+',
+    'pg/state/[1-9]\\d*/[1-9]\\d*',
+    'pg/stats/[1-9]\\d*/[1-9]\\d*',
+    'pg/history/[1-9]\\d*/[1-9]\\d*',
+    'history/last_clean_pgs',
+    'inode/stats/[1-9]\\d*/\\d+',
+    'pool/stats/[1-9]\\d*',
+    'stats',
+    'index/image/.*',
+    'index/maxid/[1-9]\\d*',
+].join('$|^')+'$');
+
+const etcd_tree = {
+    config: {
+        /* global: {
+            // WARNING: NOT ALL OF THESE ARE ACTUALLY CONFIGURABLE HERE
+            // THIS IS JUST A POOR MAN'S CONFIG DOCUMENTATION
+            // etcd connection
+            config_path: "/etc/vitastor/vitastor.conf",
+            etcd_prefix: "/vitastor",
+            // etcd connection - configurable online
+            etcd_address: "10.0.115.10:2379/v3",
+            // mon
+            etcd_mon_ttl: 5, // min: 1
+            etcd_mon_timeout: 1000, // ms. min: 0
+            etcd_mon_retries: 5, // min: 0
+            mon_change_timeout: 1000, // ms. min: 100
+            mon_retry_change_timeout: 50, // ms. min: 10
+            mon_stats_timeout: 1000, // ms. min: 100
+            osd_out_time: 600, // seconds. min: 0
+            placement_levels: { datacenter: 1, rack: 2, host: 3, osd: 4, ... },
+            use_old_pg_combinator: false,
+            // client and osd
+            tcp_header_buffer_size: 65536,
+            use_sync_send_recv: false,
+            use_rdma: true,
+            rdma_device: null, // for example, "rocep5s0f0"
+            rdma_port_num: 1,
+            rdma_gid_index: 0,
+            rdma_mtu: 4096,
+            rdma_max_sge: 128,
+            rdma_max_send: 8,
+            rdma_max_recv: 16,
+            rdma_max_msg: 132096,
+            block_size: 131072,
+            disk_alignment: 4096,
+            bitmap_granularity: 4096,
+            immediate_commit: false, // 'all' or 'small'
+            // client - configurable online
+            client_max_dirty_bytes: 33554432,
+            client_max_dirty_ops: 1024,
+            client_enable_writeback: false,
+            client_max_buffered_bytes: 33554432,
+            client_max_buffered_ops: 1024,
+            client_max_writeback_iodepth: 256,
+            client_retry_interval: 50, // ms. min: 10
+            client_eio_retry_interval: 1000, // ms
+            client_retry_enospc: true,
+            osd_nearfull_ratio: 0.95,
+            // client and osd - configurable online
+            log_level: 0,
+            peer_connect_interval: 5, // seconds. min: 1
+            peer_connect_timeout: 5, // seconds. min: 1
+            osd_idle_timeout: 5, // seconds. min: 1
+            osd_ping_timeout: 5, // seconds. min: 1
+            max_etcd_attempts: 5,
+            etcd_quick_timeout: 1000, // ms
+            etcd_slow_timeout: 5000, // ms
+            etcd_keepalive_timeout: 30, // seconds, default is max(30, etcd_report_interval*2)
+            etcd_ws_keepalive_interval: 5, // seconds
+            // osd
+            etcd_report_interval: 5, // seconds
+            etcd_stats_interval: 30, // seconds
+            run_primary: true,
+            osd_network: null, // "192.168.7.0/24" or an array of masks
+            bind_address: "0.0.0.0",
+            bind_port: 0,
+            readonly: false,
+            osd_memlock: false,
+            // osd - configurable online
+            autosync_interval: 5,
+            autosync_writes: 128,
+            client_queue_depth: 128, // unused
+            recovery_queue_depth: 1,
+            recovery_sleep_us: 0,
+            recovery_tune_util_low: 0.1,
+            recovery_tune_client_util_low: 0,
+            recovery_tune_util_high: 1.0,
+            recovery_tune_client_util_high: 0.5,
+            recovery_tune_interval: 1,
+            recovery_tune_agg_interval: 10, // 10 times recovery_tune_interval
+            recovery_tune_sleep_min_us: 10, // 10 microseconds
+            recovery_pg_switch: 128,
+            recovery_sync_batch: 16,
+            no_recovery: false,
+            no_rebalance: false,
+            print_stats_interval: 3,
+            slow_log_interval: 10,
+            inode_vanish_time: 60,
+            auto_scrub: false,
+            no_scrub: false,
+            scrub_interval: '30d', // 1s/1m/1h/1d
+            scrub_queue_depth: 1,
+            scrub_sleep: 0, // milliseconds
+            scrub_list_limit: 1000, // objects to list on one scrub iteration
+            scrub_find_best: true,
+            scrub_ec_max_bruteforce: 100, // maximum EC error locator brute-force iterators
+            // blockstore - fixed in superblock
+            block_size,
+            disk_alignment,
+            journal_block_size,
+            meta_block_size,
+            bitmap_granularity,
+            journal_device,
+            journal_offset,
+            journal_size,
+            disable_journal_fsync,
+            data_device,
+            data_offset,
+            data_size,
+            disable_data_fsync,
+            meta_device,
+            meta_offset,
+            disable_meta_fsync,
+            disable_device_lock,
+            // blockstore - configurable offline
+            inmemory_metadata,
+            inmemory_journal,
+            journal_sector_buffer_count,
+            journal_no_same_sector_overwrites,
+            // blockstore - configurable online
+            max_write_iodepth,
+            min_flusher_count: 1,
+            max_flusher_count: 256,
+            throttle_small_writes: false,
+            throttle_target_iops: 100,
+            throttle_target_mbs: 100,
+            throttle_target_parallelism: 1,
+            throttle_threshold_us: 50,
+        }, */
+        global: {},
+        /* node_placement: {
+            host1: { level: 'host', parent: 'rack1' },
+            ...
+        }, */
+        node_placement: {},
+        /* pools: {
+            <id>: {
+                name: 'testpool',
+                // 'ec' uses Reed-Solomon-Vandermonde codes, 'jerasure' is an alias for 'ec'
+                scheme: 'replicated' | 'xor' | 'ec' | 'jerasure',
+                pg_size: 3,
+                pg_minsize: 2,
+                // number of parity chunks, required for EC
+                parity_chunks?: 1,
+                pg_count: 100,
+                // default is failure_domain=host
+                failure_domain?: 'host',
+                // additional failure domain rules; failure_domain=x is equivalent to x=123..N
+                level_placement?: 'dc=112233 host=123456',
+                raw_placement?: 'any, dc=1 host!=1, dc=1 host!=(1,2)',
+                old_combinator: false,
+                max_osd_combinations: 10000,
+                // block_size, bitmap_granularity, immediate_commit must match all OSDs used in that pool
+                block_size: 131072,
+                bitmap_granularity: 4096,
+                // 'all'/'small'/'none', same as in OSD options
+                immediate_commit: 'none',
+                pg_stripe_size: 0,
+                root_node?: 'rack1',
+                // restrict pool to OSDs having all of these tags
+                osd_tags?: 'nvme' | [ 'nvme', ... ],
+                // prefer to put primary on OSD with these tags
+                primary_affinity_tags?: 'nvme' | [ 'nvme', ... ],
+                // scrub interval
+                scrub_interval?: '30d',
+            },
+            ...
+        }, */
+        pools: {},
+        osd: {
+            /* <id>: { reweight?: 1, tags?: [ 'nvme', ... ], noout?: true }, ... */
+        },
+        /* pgs: {
+            hash: string,
+            items: {
+                <pool_id>: {
+                    <pg_id>: {
+                        osd_set: [ 1, 2, 3 ],
+                        primary: 1,
+                        pause: false,
+                    }
+                }
+            }
+        }, */
+        pgs: {},
+        /* inode: {
+            <pool_id>: {
+                <inode_t>: {
+                    name: string,
+                    size?: uint64_t, // bytes
+                    parent_pool?: <pool_id>,
+                    parent_id?: <inode_t>,
+                    readonly?: boolean,
+                }
+            }
+        }, */
+        inode: {},
+    },
+    osd: {
+        state: {
+            /* <osd_num_t>: {
+                state: "up",
+                addresses: string[],
+                host: string,
+                port: uint16_t,
+                primary_enabled: boolean,
+                blockstore_enabled: boolean,
+            }, */
+        },
+        stats: {
+            /* <osd_num_t>: {
+                time: number, // unix time
+                blockstore_ready: boolean,
+                size: uint64_t, // bytes
+                free: uint64_t, // bytes
+                host: string,
+                op_stats: {
+                    <string>: { count: uint64_t, usec: uint64_t, bytes: uint64_t },
+                },
+                subop_stats: {
+                    <string>: { count: uint64_t, usec: uint64_t },
+                },
+                recovery_stats: {
+                    degraded: { count: uint64_t, bytes: uint64_t },
+                    misplaced: { count: uint64_t, bytes: uint64_t },
+                },
+            }, */
+        },
+        inodestats: {
+            /* <pool_id>: {
+                <inode_t>: {
+                    read: { count: uint64_t, usec: uint64_t, bytes: uint64_t },
+                    write: { count: uint64_t, usec: uint64_t, bytes: uint64_t },
+                    delete: { count: uint64_t, usec: uint64_t, bytes: uint64_t },
+                },
+            }, */
+        },
+        space: {
+            /* <osd_num_t>: {
+                <pool_id>: {
+                    <inode_t>: uint64_t, // bytes
+                },
+            }, */
+        },
+    },
+    mon: {
+        master: {
+            /* ip: [ string ], id: uint64_t */
+        },
+        standby: {
+            /* <uint64_t>: { ip: [ string ] }, */
+        },
+    },
+    pg: {
+        state: {
+            /* <pool_id>: {
+                <pg_id>: {
+                    primary: osd_num_t,
+                    state: ("starting"|"peering"|"incomplete"|"active"|"repeering"|"stopping"|"offline"|
+                        "degraded"|"has_incomplete"|"has_degraded"|"has_misplaced"|"has_unclean"|
+                        "has_invalid"|"has_inconsistent"|"has_corrupted"|"left_on_dead"|"scrubbing")[],
+                }
+            }, */
+        },
+        stats: {
+            /* <pool_id>: {
+                <pg_id>: {
+                    object_count: uint64_t,
+                    clean_count: uint64_t,
+                    misplaced_count: uint64_t,
+                    degraded_count: uint64_t,
+                    incomplete_count: uint64_t,
+                    write_osd_set: osd_num_t[],
+                },
+            }, */
+        },
+        history: {
+            /* <pool_id>: {
+                <pg_id>: {
+                    osd_sets: osd_num_t[][],
+                    all_peers: osd_num_t[],
+                    epoch: uint64_t,
+                    next_scrub: uint64_t,
+                },
+            }, */
+        },
+    },
+    inode: {
+        stats: {
+            /* <pool_id>: {
+                <inode_t>: {
+                    raw_used: uint64_t, // raw used bytes on OSDs
+                    read: { count: uint64_t, usec: uint64_t, bytes: uint64_t, bps: uint64_t, iops: uint64_t, lat: uint64_t },
+                    write: { count: uint64_t, usec: uint64_t, bytes: uint64_t, bps: uint64_t, iops: uint64_t, lat: uint64_t },
+                    delete: { count: uint64_t, usec: uint64_t, bytes: uint64_t, bps: uint64_t, iops: uint64_t, lat: uint64_t },
+                },
+            }, */
+        },
+    },
+    pool: {
+        stats: {
+            /* <pool_id>: {
+                used_raw_tb: float, // used raw space in the pool
+                total_raw_tb: float, // maximum amount of space in the pool
+                raw_to_usable: float, // raw to usable ratio
+                space_efficiency: float, // 0..1
+            } */
+        },
+    },
+    stats: {
+        /* op_stats: {
+            <string>: { count: uint64_t, usec: uint64_t, bytes: uint64_t, bps: uint64_t, iops: uint64_t, lat: uint64_t },
+        },
+        subop_stats: {
+            <string>: { count: uint64_t, usec: uint64_t, iops: uint64_t, lat: uint64_t },
+        },
+        recovery_stats: {
+            degraded: { count: uint64_t, bytes: uint64_t, bps: uint64_t, iops: uint64_t },
+            misplaced: { count: uint64_t, bytes: uint64_t, bps: uint64_t, iops: uint64_t },
+        },
+        object_counts: {
+            object: uint64_t,
+            clean: uint64_t,
+            misplaced: uint64_t,
+            degraded: uint64_t,
+            incomplete: uint64_t,
+        },
+        object_bytes: {
+            total: uint64_t,
+            clean: uint64_t,
+            misplaced: uint64_t,
+            degraded: uint64_t,
+            incomplete: uint64_t,
+        }, */
+    },
+    history: {
+        last_clean_pgs: {},
+    },
+    index: {
+        image: {
+            /* <name>: {
+                id: uint64_t,
+                pool_id: uint64_t,
+            }, */
+        },
+        maxid: {
+            /* <pool_id>: uint64_t, */
+        },
+    },
+};
+
+module.exports = {
+    etcd_nonempty_keys,
+    etcd_allow,
+    etcd_tree,
+};

mon/lp_optimizer/test-nonuniform.js

@@ -8,7 +8,7 @@
 // But we support this case with the "parity_space" parameter in optimize_initial()/optimize_change().
 
 const { SimpleCombinator } = require('./simple_pgs.js');
-const LPOptimizer = require('./lp-optimizer.js');
+const LPOptimizer = require('./lp_optimizer.js');
 
 const osd_tree = {
     ripper5: {

mon/lp_optimizer/test-optimize-simple.js

@@ -2,7 +2,7 @@
 // License: VNPL-1.1 (see README.md for details)
 
 const { compat } = require('./simple_pgs.js');
-const LPOptimizer = require('./lp-optimizer.js');
+const LPOptimizer = require('./lp_optimizer.js');
 
 async function run()
 {

mon/lp_optimizer/test-optimize-undersized.js

@@ -2,7 +2,7 @@
 // License: VNPL-1.1 (see README.md for details)
 
 const { compat, flatten_tree } = require('./simple_pgs.js');
-const LPOptimizer = require('./lp-optimizer.js');
+const LPOptimizer = require('./lp_optimizer.js');
 
 const crush_tree = [
     { level: 1, children: [

mon/lp_optimizer/test-optimize-unfeasible.js

@@ -2,7 +2,7 @@
 // License: VNPL-1.1 (see README.md for details)
 
 const { compat } = require('./simple_pgs.js');
-const LPOptimizer = require('./lp-optimizer.js');
+const LPOptimizer = require('./lp_optimizer.js');
 
 const osd_tree = {
     100: {

mon/lp_optimizer/test-optimize.js

@@ -2,7 +2,7 @@
 // License: VNPL-1.1 (see README.md for details)
 
 const { compat, flatten_tree } = require('./simple_pgs.js');
-const LPOptimizer = require('./lp-optimizer.js');
+const LPOptimizer = require('./lp_optimizer.js');
 
 const osd_tree = {
     100: {

mon/mon-main.js

@@ -23,4 +23,4 @@ for (let i = 2; i < process.argv.length; i++)
     }
 }
 
-new Mon(options).start().catch(e => { console.error(e); process.exit(1); });
+Mon.run_forever(options);

mon/osd_tree.js

@@ -0,0 +1,215 @@
+// Copyright (c) Vitaliy Filippov, 2019+
+// License: VNPL-1.1 (see README.md for details)
+
+function get_osd_tree(global_config, state)
+{
+    const levels = global_config.placement_levels||{};
+    levels.host = levels.host || 100;
+    levels.osd = levels.osd || 101;
+    const tree = {};
+    let up_osds = {};
+    // This requires monitor system time to be in sync with OSD system times (at least to some extent)
+    const down_time = Date.now()/1000 - global_config.osd_out_time;
+    for (const osd_num of Object.keys(state.osd.stats).sort((a, b) => a - b))
+    {
+        const stat = state.osd.stats[osd_num];
+        const osd_cfg = state.config.osd[osd_num];
+        let reweight = osd_cfg == null ? 1 : Number(osd_cfg.reweight);
+        if (reweight < 0 || isNaN(reweight))
+            reweight = 1;
+        if (stat && stat.size && reweight && (state.osd.state[osd_num] || Number(stat.time) >= down_time ||
+            osd_cfg && osd_cfg.noout))
+        {
+            // Numeric IDs are reserved for OSDs
+            if (state.osd.state[osd_num] && reweight > 0)
+            {
+                // React to down OSDs immediately
+                up_osds[osd_num] = true;
+            }
+            tree[osd_num] = tree[osd_num] || {};
+            tree[osd_num].id = osd_num;
+            tree[osd_num].parent = tree[osd_num].parent || stat.host;
+            tree[osd_num].level = 'osd';
+            tree[osd_num].size = reweight * stat.size / 1024 / 1024 / 1024 / 1024; // terabytes
+            if (osd_cfg && osd_cfg.tags)
+            {
+                tree[osd_num].tags = (osd_cfg.tags instanceof Array ? [ ...osd_cfg.tags ] : [ osd_cfg.tags ])
+                    .reduce((a, c) => { a[c] = true; return a; }, {});
+            }
+            delete tree[osd_num].children;
+            if (!tree[stat.host])
+            {
+                tree[stat.host] = {
+                    id: stat.host,
+                    level: 'host',
+                    parent: null,
+                    children: [],
+                };
+            }
+        }
+    }
+    for (const node_id in state.config.node_placement||{})
+    {
+        const node_cfg = state.config.node_placement[node_id];
+        if (/^\d+$/.exec(node_id))
+        {
+            node_cfg.level = 'osd';
+        }
+        if (!node_id || !node_cfg.level || !levels[node_cfg.level] ||
+            node_cfg.level === 'osd' && !tree[node_id])
+        {
+            // All nodes must have non-empty IDs and valid levels
+            // OSDs have to actually exist
+            continue;
+        }
+        tree[node_id] = tree[node_id] || {};
+        tree[node_id].id = node_id;
+        tree[node_id].level = node_cfg.level;
+        tree[node_id].parent = node_cfg.parent;
+        if (node_cfg.level !== 'osd')
+        {
+            tree[node_id].children = [];
+        }
+    }
+    return { up_osds, levels, osd_tree: tree };
+}
+
+function make_hier_tree(global_config, tree)
+{
+    const levels = global_config.placement_levels||{};
+    levels.host = levels.host || 100;
+    levels.osd = levels.osd || 101;
+    tree = { ...tree };
+    for (const node_id in tree)
+    {
+        tree[node_id] = { ...tree[node_id], children: [] };
+    }
+    tree[''] = { children: [] };
+    for (const node_id in tree)
+    {
+        if (node_id === '' || tree[node_id].level === 'osd' && (!tree[node_id].size || tree[node_id].size <= 0))
+        {
+            continue;
+        }
+        const node_cfg = tree[node_id];
+        const node_level = levels[node_cfg.level] || node_cfg.level;
+        let parent_level = node_cfg.parent && tree[node_cfg.parent] && tree[node_cfg.parent].children
+            && tree[node_cfg.parent].level;
+        parent_level = parent_level ? (levels[parent_level] || parent_level) : null;
+        // Parent's level must be less than child's; OSDs must be leaves
+        const parent = parent_level && parent_level < node_level ? node_cfg.parent : '';
+        tree[parent].children.push(tree[node_id]);
+    }
+    // Delete empty nodes
+    let deleted = 0;
+    do
+    {
+        deleted = 0;
+        for (const node_id in tree)
+        {
+            if (tree[node_id].level !== 'osd' && (!tree[node_id].children || !tree[node_id].children.length))
+            {
+                const parent = tree[node_id].parent;
+                if (parent)
+                {
+                    tree[parent].children = tree[parent].children.filter(c => c != tree[node_id]);
+                }
+                deleted++;
+                delete tree[node_id];
+            }
+        }
+    } while (deleted > 0);
+    return tree;
+}
+
+function filter_osds_by_root_node(global_config, pool_tree, root_node)
+{
+    if (!root_node)
+    {
+        return;
+    }
+    let hier_tree = make_hier_tree(global_config, pool_tree);
+    let included = [ ...(hier_tree[root_node] || {}).children||[] ];
+    for (let i = 0; i < included.length; i++)
+    {
+        if (included[i].children)
+        {
+            included.splice(i+1, 0, ...included[i].children);
+        }
+    }
+    let cur = pool_tree[root_node] || {};
+    while (cur && cur.id)
+    {
+        included.unshift(cur);
+        cur = pool_tree[cur.parent||''];
+    }
+    included = included.reduce((a, c) => { a[c.id||''] = true; return a; }, {});
+    for (const item in pool_tree)
+    {
+        if (!included[item])
+        {
+            delete pool_tree[item];
+        }
+    }
+}
+
+function filter_osds_by_tags(orig_tree, tags)
+{
+    if (!tags)
+    {
+        return;
+    }
+    for (const tag of (tags instanceof Array ? tags : [ tags ]))
+    {
+        for (const osd in orig_tree)
+        {
+            if (orig_tree[osd].level === 'osd' &&
+                (!orig_tree[osd].tags || !orig_tree[osd].tags[tag]))
+            {
+                delete orig_tree[osd];
+            }
+        }
+    }
+}
+
+function filter_osds_by_block_layout(orig_tree, osd_stats, block_size, bitmap_granularity, immediate_commit)
+{
+    for (const osd in orig_tree)
+    {
+        if (orig_tree[osd].level === 'osd')
+        {
+            const osd_stat = osd_stats[osd];
+            if (osd_stat && (osd_stat.bs_block_size && osd_stat.bs_block_size != block_size ||
+                osd_stat.bitmap_granularity && osd_stat.bitmap_granularity != bitmap_granularity ||
+                osd_stat.immediate_commit == 'small' && immediate_commit == 'all' ||
+                osd_stat.immediate_commit == 'none' && immediate_commit != 'none'))
+            {
+                delete orig_tree[osd];
+            }
+        }
+    }
+}
+
+function get_affinity_osds(pool_cfg, up_osds, osd_tree)
+{
+    let aff_osds = up_osds;
+    if (pool_cfg.primary_affinity_tags)
+    {
+        aff_osds = Object.keys(up_osds).reduce((a, c) => { a[c] = osd_tree[c]; return a; }, {});
+        filter_osds_by_tags(aff_osds, pool_cfg.primary_affinity_tags);
+        for (const osd in aff_osds)
+        {
+            aff_osds[osd] = true;
+        }
+    }
+    return aff_osds;
+}
+
+module.exports = {
+    get_osd_tree,
+    make_hier_tree,
+    filter_osds_by_root_node,
+    filter_osds_by_tags,
+    filter_osds_by_block_layout,
+    get_affinity_osds,
+};

mon/package.json

@@ -4,7 +4,7 @@
   "description": "Vitastor SDS monitor service",
   "main": "mon-main.js",
   "scripts": {
-    "test": "echo \"Error: no test specified\" && exit 1"
+    "lint": "eslint *.js lp_optimizer/*.js scripts/*.js"
   },
   "author": "Vitaliy Filippov",
   "license": "UNLICENSED",
@@ -14,12 +14,10 @@
   },
   "devDependencies": {
     "eslint": "^8.0.0",
+    "eslint-plugin-import": "^2.29.1",
     "eslint-plugin-node": "^11.1.0"
   },
   "engines": {
     "node": ">=12.0.0"
-  },
-  "scripts": {
-    "lint": "eslint *.js"
   }
 }

mon/pg_gen.js

@@ -0,0 +1,267 @@
+// Copyright (c) Vitaliy Filippov, 2019+
+// License: VNPL-1.1 (see README.md for details)
+
+const { RuleCombinator } = require('./lp_optimizer/dsl_pgs.js');
+const { SimpleCombinator, flatten_tree } = require('./lp_optimizer/simple_pgs.js');
+const { validate_pool_cfg, get_pg_rules } = require('./pool_config.js');
+const LPOptimizer = require('./lp_optimizer/lp_optimizer.js');
+const { scale_pg_count } = require('./pg_utils.js');
+const { make_hier_tree, filter_osds_by_root_node,
+    filter_osds_by_tags, filter_osds_by_block_layout, get_affinity_osds } = require('./osd_tree.js');
+
+let seed;
+
+function reset_rng()
+{
+    seed = 0x5f020e43;
+}
+
+function rng()
+{
+    seed ^= seed << 13;
+    seed ^= seed >> 17;
+    seed ^= seed << 5;
+    return seed + 2147483648;
+}
+
+function pick_primary(pool_config, osd_set, up_osds, aff_osds)
+{
+    let alive_set;
+    if (pool_config.scheme === 'replicated')
+    {
+        // Prefer "affinity" OSDs
+        alive_set = osd_set.filter(osd_num => osd_num && aff_osds[osd_num]);
+        if (!alive_set.length)
+            alive_set = osd_set.filter(osd_num => osd_num && up_osds[osd_num]);
+    }
+    else
+    {
+        // Prefer data OSDs for EC because they can actually read something without an additional network hop
+        const pg_data_size = (pool_config.pg_size||0) - (pool_config.parity_chunks||0);
+        alive_set = osd_set.slice(0, pg_data_size).filter(osd_num => osd_num && aff_osds[osd_num]);
+        if (!alive_set.length)
+            alive_set = osd_set.filter(osd_num => osd_num && aff_osds[osd_num]);
+        if (!alive_set.length)
+        {
+            alive_set = osd_set.slice(0, pg_data_size).filter(osd_num => osd_num && up_osds[osd_num]);
+            if (!alive_set.length)
+                alive_set = osd_set.filter(osd_num => osd_num && up_osds[osd_num]);
+        }
+    }
+    if (!alive_set.length)
+    {
+        return 0;
+    }
+    return alive_set[rng() % alive_set.length];
+}
+
+function recheck_primary(state, global_config, up_osds, osd_tree)
+{
+    let new_config_pgs;
+    for (const pool_id in state.config.pools)
+    {
+        const pool_cfg = state.config.pools[pool_id];
+        if (!validate_pool_cfg(pool_id, pool_cfg, global_config.placement_levels, false))
+        {
+            continue;
+        }
+        const aff_osds = get_affinity_osds(pool_cfg, up_osds, osd_tree);
+        reset_rng();
+        for (let pg_num = 1; pg_num <= pool_cfg.pg_count; pg_num++)
+        {
+            if (!state.config.pgs.items[pool_id])
+            {
+                continue;
+            }
+            const pg_cfg = state.config.pgs.items[pool_id][pg_num];
+            if (pg_cfg)
+            {
+                const new_primary = pick_primary(state.config.pools[pool_id], pg_cfg.osd_set, up_osds, aff_osds);
+                if (pg_cfg.primary != new_primary)
+                {
+                    if (!new_config_pgs)
+                    {
+                        new_config_pgs = JSON.parse(JSON.stringify(state.config.pgs));
+                    }
+                    console.log(
+                        `Moving pool ${pool_id} (${pool_cfg.name || 'unnamed'}) PG ${pg_num}`+
+                        ` primary OSD from ${pg_cfg.primary} to ${new_primary}`
+                    );
+                    new_config_pgs.items[pool_id][pg_num].primary = new_primary;
+                }
+            }
+        }
+    }
+    return new_config_pgs;
+}
+
+function save_new_pgs_txn(save_to, request, state, etcd_prefix, etcd_watch_revision, pool_id, up_osds, osd_tree, prev_pgs, new_pgs, pg_history)
+{
+    const aff_osds = get_affinity_osds(state.config.pools[pool_id] || {}, up_osds, osd_tree);
+    const pg_items = {};
+    reset_rng();
+    new_pgs.map((osd_set, i) =>
+    {
+        osd_set = osd_set.map(osd_num => osd_num === LPOptimizer.NO_OSD ? 0 : osd_num);
+        pg_items[i+1] = {
+            osd_set,
+            primary: pick_primary(state.config.pools[pool_id], osd_set, up_osds, aff_osds),
+        };
+        if (prev_pgs[i] && prev_pgs[i].join(' ') != osd_set.join(' ') &&
+            prev_pgs[i].filter(osd_num => osd_num).length > 0)
+        {
+            pg_history[i] = pg_history[i] || {};
+            pg_history[i].osd_sets = pg_history[i].osd_sets || [];
+            pg_history[i].osd_sets.push(prev_pgs[i]);
+        }
+        if (pg_history[i] && pg_history[i].osd_sets)
+        {
+            pg_history[i].osd_sets = Object.values(pg_history[i].osd_sets
+                .reduce((a, c) => { a[c.join(' ')] = c; return a; }, {}));
+        }
+    });
+    for (let i = 0; i < new_pgs.length || i < prev_pgs.length; i++)
+    {
+        // FIXME: etcd has max_txn_ops limit, and it's 128 by default
+        // Sooo we probably want to change our storage scheme for PG histories...
+        request.compare.push({
+            key: b64(etcd_prefix+'/pg/history/'+pool_id+'/'+(i+1)),
+            target: 'MOD',
+            mod_revision: ''+etcd_watch_revision,
+            result: 'LESS',
+        });
+        if (pg_history[i])
+        {
+            request.success.push({
+                requestPut: {
+                    key: b64(etcd_prefix+'/pg/history/'+pool_id+'/'+(i+1)),
+                    value: b64(JSON.stringify(pg_history[i])),
+                },
+            });
+        }
+        else
+        {
+            request.success.push({
+                requestDeleteRange: {
+                    key: b64(etcd_prefix+'/pg/history/'+pool_id+'/'+(i+1)),
+                },
+            });
+        }
+    }
+    save_to.items = save_to.items || {};
+    if (!new_pgs.length)
+    {
+        delete save_to.items[pool_id];
+    }
+    else
+    {
+        save_to.items[pool_id] = pg_items;
+    }
+}
+
+async function generate_pool_pgs(state, global_config, pool_id, osd_tree, levels)
+{
+    const pool_cfg = state.config.pools[pool_id];
+    if (!validate_pool_cfg(pool_id, pool_cfg, global_config.placement_levels, false))
+    {
+        return null;
+    }
+    let pool_tree = { ...osd_tree };
+    filter_osds_by_root_node(global_config, pool_tree, pool_cfg.root_node);
+    filter_osds_by_tags(pool_tree, pool_cfg.osd_tags);
+    filter_osds_by_block_layout(
+        pool_tree,
+        state.osd.stats,
+        pool_cfg.block_size || global_config.block_size || 131072,
+        pool_cfg.bitmap_granularity || global_config.bitmap_granularity || 4096,
+        pool_cfg.immediate_commit || global_config.immediate_commit || 'none'
+    );
+    pool_tree = make_hier_tree(global_config, pool_tree);
+    // First try last_clean_pgs to minimize data movement
+    let prev_pgs = [];
+    for (const pg in ((state.history.last_clean_pgs.items||{})[pool_id]||{}))
+    {
+        prev_pgs[pg-1] = [ ...state.history.last_clean_pgs.items[pool_id][pg].osd_set ];
+    }
+    if (!prev_pgs.length)
+    {
+        // Fall back to config/pgs if it's empty
+        for (const pg in ((state.config.pgs.items||{})[pool_id]||{}))
+        {
+            prev_pgs[pg-1] = [ ...state.config.pgs.items[pool_id][pg].osd_set ];
+        }
+    }
+    const old_pg_count = prev_pgs.length;
+    const optimize_cfg = {
+        osd_weights: Object.values(pool_tree).filter(item => item.level === 'osd').reduce((a, c) => { a[c.id] = c.size; return a; }, {}),
+        combinator: !global_config.use_old_pg_combinator || pool_cfg.level_placement || pool_cfg.raw_placement
+            // new algorithm:
+            ? new RuleCombinator(pool_tree, get_pg_rules(pool_id, pool_cfg, global_config.placement_levels), pool_cfg.max_osd_combinations)
+            // old algorithm:
+            : new SimpleCombinator(flatten_tree(pool_tree[''].children, levels, pool_cfg.failure_domain, 'osd'), pool_cfg.pg_size, pool_cfg.max_osd_combinations),
+        pg_count: pool_cfg.pg_count,
+        pg_size: pool_cfg.pg_size,
+        pg_minsize: pool_cfg.pg_minsize,
+        ordered: pool_cfg.scheme != 'replicated',
+    };
+    let optimize_result;
+    // Re-shuffle PGs if config/pgs.hash is empty
+    if (old_pg_count > 0 && state.config.pgs.hash)
+    {
+        if (prev_pgs.length != pool_cfg.pg_count)
+        {
+            // Scale PG count
+            // Do it even if old_pg_count is already equal to pool_cfg.pg_count,
+            // because last_clean_pgs may still contain the old number of PGs
+            scale_pg_count(prev_pgs, pool_cfg.pg_count);
+        }
+        for (const pg of prev_pgs)
+        {
+            while (pg.length < pool_cfg.pg_size)
+            {
+                pg.push(0);
+            }
+        }
+        optimize_result = await LPOptimizer.optimize_change({
+            prev_pgs,
+            ...optimize_cfg,
+        });
+    }
+    else
+    {
+        optimize_result = await LPOptimizer.optimize_initial(optimize_cfg);
+    }
+    console.log(`Pool ${pool_id} (${pool_cfg.name || 'unnamed'}):`);
+    LPOptimizer.print_change_stats(optimize_result);
+    let pg_effsize = pool_cfg.pg_size;
+    for (const pg of optimize_result.int_pgs)
+    {
+        const this_pg_size = pg.filter(osd => osd != LPOptimizer.NO_OSD).length;
+        if (this_pg_size && this_pg_size < pg_effsize)
+        {
+            pg_effsize = this_pg_size;
+        }
+    }
+    return {
+        pool_id,
+        pgs: optimize_result.int_pgs,
+        stats: {
+            total_raw_tb: optimize_result.space,
+            pg_real_size: pg_effsize || pool_cfg.pg_size,
+            raw_to_usable: (pg_effsize || pool_cfg.pg_size) / (pool_cfg.scheme === 'replicated'
+                ? 1 : (pool_cfg.pg_size - (pool_cfg.parity_chunks||0))),
+            space_efficiency: optimize_result.space/(optimize_result.total_space||1),
+        },
+    };
+}
+
+function b64(str)
+{
+    return Buffer.from(str).toString('base64');
+}
+
+module.exports = {
+    recheck_primary,
+    save_new_pgs_txn,
+    generate_pool_pgs,
+};

mon/pool_config.js

@@ -0,0 +1,169 @@
+// Copyright (c) Vitaliy Filippov, 2019+
+// License: VNPL-1.1 (see README.md for details)
+
+const { parse_level_indexes, parse_pg_dsl } = require('./lp_optimizer/dsl_pgs.js');
+
+function validate_pool_cfg(pool_id, pool_cfg, placement_levels, warn)
+{
+    pool_cfg.pg_size = Math.floor(pool_cfg.pg_size);
+    pool_cfg.pg_minsize = Math.floor(pool_cfg.pg_minsize);
+    pool_cfg.parity_chunks = Math.floor(pool_cfg.parity_chunks) || undefined;
+    pool_cfg.pg_count = Math.floor(pool_cfg.pg_count);
+    pool_cfg.max_osd_combinations = Math.floor(pool_cfg.max_osd_combinations) || 10000;
+    if (!/^[1-9]\d*$/.exec(''+pool_id))
+    {
+        if (warn)
+            console.log('Pool ID '+pool_id+' is invalid');
+        return false;
+    }
+    if (pool_cfg.scheme !== 'xor' && pool_cfg.scheme !== 'replicated' &&
+        pool_cfg.scheme !== 'ec' && pool_cfg.scheme !== 'jerasure')
+    {
+        if (warn)
+            console.log('Pool '+pool_id+' has invalid coding scheme (one of "xor", "replicated", "ec" and "jerasure" required)');
+        return false;
+    }
+    if (!pool_cfg.pg_size || pool_cfg.pg_size < 1 || pool_cfg.pg_size > 256 ||
+        pool_cfg.scheme !== 'replicated' && pool_cfg.pg_size < 3)
+    {
+        if (warn)
+            console.log('Pool '+pool_id+' has invalid pg_size');
+        return false;
+    }
+    if (!pool_cfg.pg_minsize || pool_cfg.pg_minsize < 1 || pool_cfg.pg_minsize > pool_cfg.pg_size ||
+        pool_cfg.scheme === 'xor' && pool_cfg.pg_minsize < (pool_cfg.pg_size - 1))
+    {
+        if (warn)
+            console.log('Pool '+pool_id+' has invalid pg_minsize');
+        return false;
+    }
+    if (pool_cfg.scheme === 'xor' && pool_cfg.parity_chunks != 0 && pool_cfg.parity_chunks != 1)
+    {
+        if (warn)
+            console.log('Pool '+pool_id+' has invalid parity_chunks (must be 1)');
+        return false;
+    }
+    if ((pool_cfg.scheme === 'ec' || pool_cfg.scheme === 'jerasure') &&
+        (pool_cfg.parity_chunks < 1 || pool_cfg.parity_chunks > pool_cfg.pg_size-2))
+    {
+        if (warn)
+            console.log('Pool '+pool_id+' has invalid parity_chunks (must be between 1 and pg_size-2)');
+        return false;
+    }
+    if (!pool_cfg.pg_count || pool_cfg.pg_count < 1)
+    {
+        if (warn)
+            console.log('Pool '+pool_id+' has invalid pg_count');
+        return false;
+    }
+    if (!pool_cfg.name)
+    {
+        if (warn)
+            console.log('Pool '+pool_id+' has empty name');
+        return false;
+    }
+    if (pool_cfg.max_osd_combinations < 100)
+    {
+        if (warn)
+            console.log('Pool '+pool_id+' has invalid max_osd_combinations (must be at least 100)');
+        return false;
+    }
+    if (pool_cfg.root_node && typeof(pool_cfg.root_node) != 'string')
+    {
+        if (warn)
+            console.log('Pool '+pool_id+' has invalid root_node (must be a string)');
+        return false;
+    }
+    if (pool_cfg.osd_tags && typeof(pool_cfg.osd_tags) != 'string' &&
+        (!(pool_cfg.osd_tags instanceof Array) || pool_cfg.osd_tags.filter(t => typeof t != 'string').length > 0))
+    {
+        if (warn)
+            console.log('Pool '+pool_id+' has invalid osd_tags (must be a string or array of strings)');
+        return false;
+    }
+    if (pool_cfg.primary_affinity_tags && typeof(pool_cfg.primary_affinity_tags) != 'string' &&
+        (!(pool_cfg.primary_affinity_tags instanceof Array) || pool_cfg.primary_affinity_tags.filter(t => typeof t != 'string').length > 0))
+    {
+        if (warn)
+            console.log('Pool '+pool_id+' has invalid primary_affinity_tags (must be a string or array of strings)');
+        return false;
+    }
+    if (!get_pg_rules(pool_id, pool_cfg, placement_levels, true))
+    {
+        return false;
+    }
+    return true;
+}
+
+function get_pg_rules(pool_id, pool_cfg, placement_levels, warn)
+{
+    if (pool_cfg.level_placement)
+    {
+        const pg_size = (0|pool_cfg.pg_size);
+        let rules = pool_cfg.level_placement;
+        if (typeof rules === 'string')
+        {
+            rules = rules.split(/\s+/).map(s => s.split(/=/, 2)).reduce((a, c) => { a[c[0]] = c[1]; return a; }, {});
+        }
+        else
+        {
+            rules = { ...rules };
+        }
+        // Always add failure_domain to prevent rules from being totally incorrect
+        const all_diff = [];
+        for (let i = 1; i <= pg_size; i++)
+        {
+            all_diff.push(i);
+        }
+        rules[pool_cfg.failure_domain || 'host'] = all_diff;
+        placement_levels = placement_levels||{};
+        placement_levels.host = placement_levels.host || 100;
+        placement_levels.osd = placement_levels.osd || 101;
+        for (const k in rules)
+        {
+            if (!placement_levels[k] || typeof rules[k] !== 'string' &&
+                (!(rules[k] instanceof Array) ||
+                rules[k].filter(s => typeof s !== 'string' && typeof s !== 'number').length > 0))
+            {
+                if (warn)
+                    console.log('Pool '+pool_id+' configuration is invalid: level_placement should be { [level]: string | (string|number)[] }');
+                return null;
+            }
+            else if (rules[k].length != pg_size)
+            {
+                if (warn)
+                    console.log('Pool '+pool_id+' configuration is invalid: values in level_placement should contain exactly pg_size ('+pg_size+') items');
+                return null;
+            }
+        }
+        return parse_level_indexes(rules);
+    }
+    else if (typeof pool_cfg.raw_placement === 'string')
+    {
+        try
+        {
+            return parse_pg_dsl(pool_cfg.raw_placement);
+        }
+        catch (e)
+        {
+            if (warn)
+                console.log('Pool '+pool_id+' configuration is invalid: invalid raw_placement: '+e.message);
+        }
+    }
+    else
+    {
+        let rules = [ [] ];
+        let prev = [ 1 ];
+        for (let i = 1; i < pool_cfg.pg_size; i++)
+        {
+            rules.push([ [ pool_cfg.failure_domain||'host', '!=', prev ] ]);
+            prev = [ ...prev, i+1 ];
+        }
+        return rules;
+    }
+}
+
+module.exports = {
+    validate_pool_cfg,
+    get_pg_rules,
+};

mon/stats.js

@@ -0,0 +1,286 @@
+// Copyright (c) Vitaliy Filippov, 2019+
+// License: VNPL-1.1 (see README.md for details)
+
+function derive_osd_stats(st, prev, prev_diff)
+{
+    const diff = { op_stats: {}, subop_stats: {}, recovery_stats: {}, inode_stats: {} };
+    if (!st || !st.time || !prev || !prev.time || prev.time >= st.time)
+    {
+        return prev_diff || diff;
+    }
+    const timediff = BigInt(st.time*1000 - prev.time*1000);
+    for (const op in st.op_stats||{})
+    {
+        const pr = prev && prev.op_stats && prev.op_stats[op];
+        let c = st.op_stats[op];
+        c = { bytes: BigInt(c.bytes||0), usec: BigInt(c.usec||0), count: BigInt(c.count||0) };
+        const b = c.bytes - BigInt(pr && pr.bytes||0);
+        const us = c.usec - BigInt(pr && pr.usec||0);
+        const n = c.count - BigInt(pr && pr.count||0);
+        if (n > 0)
+            diff.op_stats[op] = { ...c, bps: b*1000n/timediff, iops: n*1000n/timediff, lat: us/n };
+    }
+    for (const op in st.subop_stats||{})
+    {
+        const pr = prev && prev.subop_stats && prev.subop_stats[op];
+        let c = st.subop_stats[op];
+        c = { usec: BigInt(c.usec||0), count: BigInt(c.count||0) };
+        const us = c.usec - BigInt(pr && pr.usec||0);
+        const n = c.count - BigInt(pr && pr.count||0);
+        if (n > 0)
+            diff.subop_stats[op] = { ...c, iops: n*1000n/timediff, lat: us/n };
+    }
+    for (const op in st.recovery_stats||{})
+    {
+        const pr = prev && prev.recovery_stats && prev.recovery_stats[op];
+        let c = st.recovery_stats[op];
+        c = { bytes: BigInt(c.bytes||0), count: BigInt(c.count||0) };
+        const b = c.bytes - BigInt(pr && pr.bytes||0);
+        const n = c.count - BigInt(pr && pr.count||0);
+        if (n > 0)
+            diff.recovery_stats[op] = { ...c, bps: b*1000n/timediff, iops: n*1000n/timediff };
+    }
+    for (const pool_id in st.inode_stats||{})
+    {
+        diff.inode_stats[pool_id] = {};
+        for (const inode_num in st.inode_stats[pool_id])
+        {
+            const inode_diff = diff.inode_stats[pool_id][inode_num] = {};
+            for (const op of [ 'read', 'write', 'delete' ])
+            {
+                const c = st.inode_stats[pool_id][inode_num][op];
+                const pr = prev && prev.inode_stats && prev.inode_stats[pool_id] &&
+                    prev.inode_stats[pool_id][inode_num] && prev.inode_stats[pool_id][inode_num][op];
+                const n = BigInt(c.count||0) - BigInt(pr && pr.count||0);
+                inode_diff[op] = {
+                    bps: (BigInt(c.bytes||0) - BigInt(pr && pr.bytes||0))*1000n/timediff,
+                    iops: n*1000n/timediff,
+                    lat: (BigInt(c.usec||0) - BigInt(pr && pr.usec||0))/(n || 1n),
+                };
+            }
+        }
+    }
+    return diff;
+}
+
+// sum_op_stats(this.state.osd, this.prev_stats)
+function sum_op_stats(all_osd, prev_stats)
+{
+    for (const osd in all_osd.stats)
+    {
+        const cur = { ...all_osd.stats[osd], inode_stats: all_osd.inodestats[osd]||{} };
+        prev_stats.osd_diff[osd] = derive_osd_stats(
+            cur, prev_stats.osd_stats[osd], prev_stats.osd_diff[osd]
+        );
+        prev_stats.osd_stats[osd] = cur;
+    }
+    const sum_diff = { op_stats: {}, subop_stats: {}, recovery_stats: {} };
+    // Sum derived values instead of deriving summed
+    for (const osd in all_osd.state)
+    {
+        const derived = prev_stats.osd_diff[osd];
+        if (!all_osd.state[osd] || !derived)
+        {
+            continue;
+        }
+        for (const type in sum_diff)
+        {
+            for (const op in derived[type]||{})
+            {
+                for (const k in derived[type][op])
+                {
+                    sum_diff[type][op] = sum_diff[type][op] || {};
+                    sum_diff[type][op][k] = (sum_diff[type][op][k] || 0n) + derived[type][op][k];
+                }
+            }
+        }
+    }
+    return sum_diff;
+}
+
+// sum_object_counts(this.state, this.config)
+function sum_object_counts(state, global_config)
+{
+    const object_counts = { object: 0n, clean: 0n, misplaced: 0n, degraded: 0n, incomplete: 0n };
+    const object_bytes = { object: 0n, clean: 0n, misplaced: 0n, degraded: 0n, incomplete: 0n };
+    for (const pool_id in state.pg.stats)
+    {
+        let object_size = 0;
+        for (const osd_num of state.pg.stats[pool_id].write_osd_set||[])
+        {
+            if (osd_num && state.osd.stats[osd_num] && state.osd.stats[osd_num].block_size)
+            {
+                object_size = state.osd.stats[osd_num].block_size;
+                break;
+            }
+        }
+        const pool_cfg = (state.config.pools[pool_id]||{});
+        if (!object_size)
+        {
+            object_size = pool_cfg.block_size || global_config.block_size || 131072;
+        }
+        if (pool_cfg.scheme !== 'replicated')
+        {
+            object_size *= ((pool_cfg.pg_size||0) - (pool_cfg.parity_chunks||0));
+        }
+        object_size = BigInt(object_size);
+        for (const pg_num in state.pg.stats[pool_id])
+        {
+            const st = state.pg.stats[pool_id][pg_num];
+            if (st)
+            {
+                for (const k in object_counts)
+                {
+                    if (st[k+'_count'])
+                    {
+                        object_counts[k] += BigInt(st[k+'_count']);
+                        object_bytes[k] += BigInt(st[k+'_count']) * object_size;
+                    }
+                }
+            }
+        }
+    }
+    return { object_counts, object_bytes };
+}
+
+// sum_inode_stats(this.state, this.prev_stats)
+function sum_inode_stats(state, prev_stats)
+{
+    const inode_stats = {};
+    const inode_stub = () => ({
+        raw_used: 0n,
+        read: { count: 0n, usec: 0n, bytes: 0n, bps: 0n, iops: 0n, lat: 0n },
+        write: { count: 0n, usec: 0n, bytes: 0n, bps: 0n, iops: 0n, lat: 0n },
+        delete: { count: 0n, usec: 0n, bytes: 0n, bps: 0n, iops: 0n, lat: 0n },
+    });
+    const seen_pools = {};
+    for (const pool_id in state.config.pools)
+    {
+        seen_pools[pool_id] = true;
+        state.pool.stats[pool_id] = state.pool.stats[pool_id] || {};
+        state.pool.stats[pool_id].used_raw_tb = 0n;
+    }
+    for (const osd_num in state.osd.space)
+    {
+        for (const pool_id in state.osd.space[osd_num])
+        {
+            state.pool.stats[pool_id] = state.pool.stats[pool_id] || {};
+            if (!seen_pools[pool_id])
+            {
+                state.pool.stats[pool_id].used_raw_tb = 0n;
+                seen_pools[pool_id] = true;
+            }
+            inode_stats[pool_id] = inode_stats[pool_id] || {};
+            for (const inode_num in state.osd.space[osd_num][pool_id])
+            {
+                const u = BigInt(state.osd.space[osd_num][pool_id][inode_num]||0);
+                if (inode_num)
+                {
+                    inode_stats[pool_id][inode_num] = inode_stats[pool_id][inode_num] || inode_stub();
+                    inode_stats[pool_id][inode_num].raw_used += u;
+                }
+                state.pool.stats[pool_id].used_raw_tb += u;
+            }
+        }
+    }
+    for (const pool_id in seen_pools)
+    {
+        const used = state.pool.stats[pool_id].used_raw_tb;
+        state.pool.stats[pool_id].used_raw_tb = Number(used)/1024/1024/1024/1024;
+    }
+    for (const osd_num in state.osd.state)
+    {
+        const ist = state.osd.inodestats[osd_num];
+        if (!ist || !state.osd.state[osd_num])
+        {
+            continue;
+        }
+        for (const pool_id in ist)
+        {
+            inode_stats[pool_id] = inode_stats[pool_id] || {};
+            for (const inode_num in ist[pool_id])
+            {
+                inode_stats[pool_id][inode_num] = inode_stats[pool_id][inode_num] || inode_stub();
+                for (const op of [ 'read', 'write', 'delete' ])
+                {
+                    inode_stats[pool_id][inode_num][op].count += BigInt(ist[pool_id][inode_num][op].count||0);
+                    inode_stats[pool_id][inode_num][op].usec += BigInt(ist[pool_id][inode_num][op].usec||0);
+                    inode_stats[pool_id][inode_num][op].bytes += BigInt(ist[pool_id][inode_num][op].bytes||0);
+                }
+            }
+        }
+    }
+    for (const osd in state.osd.state)
+    {
+        const osd_diff = prev_stats.osd_diff[osd];
+        if (!osd_diff || !state.osd.state[osd])
+        {
+            continue;
+        }
+        for (const pool_id in osd_diff.inode_stats)
+        {
+            for (const inode_num in prev_stats.osd_diff[osd].inode_stats[pool_id])
+            {
+                inode_stats[pool_id][inode_num] = inode_stats[pool_id][inode_num] || inode_stub();
+                for (const op of [ 'read', 'write', 'delete' ])
+                {
+                    const op_diff = prev_stats.osd_diff[osd].inode_stats[pool_id][inode_num][op] || {};
+                    const op_st = inode_stats[pool_id][inode_num][op];
+                    op_st.bps += op_diff.bps;
+                    op_st.iops += op_diff.iops;
+                    op_st.lat += op_diff.lat;
+                    op_st.n_osd = (op_st.n_osd || 0) + 1;
+                }
+            }
+        }
+    }
+    for (const pool_id in inode_stats)
+    {
+        for (const inode_num in inode_stats[pool_id])
+        {
+            let nonzero = inode_stats[pool_id][inode_num].raw_used > 0;
+            for (const op of [ 'read', 'write', 'delete' ])
+            {
+                const op_st = inode_stats[pool_id][inode_num][op];
+                if (op_st.n_osd)
+                {
+                    op_st.lat /= BigInt(op_st.n_osd);
+                    delete op_st.n_osd;
+                }
+                if (op_st.bps > 0 || op_st.iops > 0)
+                    nonzero = true;
+            }
+            if (!nonzero && (!state.config.inode[pool_id] || !state.config.inode[pool_id][inode_num]))
+            {
+                // Deleted inode (no data, no I/O, no config)
+                delete inode_stats[pool_id][inode_num];
+            }
+        }
+    }
+    return { inode_stats, seen_pools };
+}
+
+function serialize_bigints(obj)
+{
+    obj = { ...obj };
+    for (const k in obj)
+    {
+        if (typeof obj[k] == 'bigint')
+        {
+            obj[k] = ''+obj[k];
+        }
+        else if (typeof obj[k] == 'object')
+        {
+            obj[k] = serialize_bigints(obj[k]);
+        }
+    }
+    return obj;
+}
+
+module.exports = {
+    derive_osd_stats,
+    sum_op_stats,
+    sum_object_counts,
+    sum_inode_stats,
+    serialize_bigints,
+};

patches/cinder-vitastor-22.py

@@ -0,0 +1,966 @@
+# Vitastor Driver for OpenStack Cinder
+#
+# --------------------------------------------
+# Install as cinder/volume/drivers/vitastor.py
+# --------------------------------------------
+#
+# Copyright 2020 Vitaliy Filippov
+#
+# Licensed under the Apache License, Version 2.0 (the "License"); you may
+# not use this file except in compliance with the License. You may obtain
+# a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
+# WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
+# License for the specific language governing permissions and limitations
+# under the License.
+"""Cinder Vitastor Driver"""
+
+import binascii
+import base64
+import errno
+import json
+import math
+import os
+import tempfile
+
+from castellan import key_manager
+from oslo_config import cfg
+from oslo_log import log as logging
+from oslo_service import loopingcall
+from oslo_concurrency import processutils
+from oslo_utils import encodeutils
+from oslo_utils import excutils
+from oslo_utils import fileutils
+from oslo_utils import units
+import six
+from six.moves.urllib import request
+
+from cinder import exception
+from cinder.i18n import _
+from cinder.image import image_utils
+from cinder import interface
+from cinder import objects
+from cinder.objects import fields
+from cinder import utils
+from cinder.volume import configuration
+from cinder.volume import driver
+from cinder.volume import volume_utils
+
+VERSION = '1.4.1'
+
+LOG = logging.getLogger(__name__)
+
+VITASTOR_OPTS = [
+    cfg.StrOpt(
+        'vitastor_config_path',
+        default='/etc/vitastor/vitastor.conf',
+        help='Vitastor configuration file path'
+    ),
+    cfg.StrOpt(
+        'vitastor_etcd_address',
+        default='',
+        help='Vitastor etcd address(es)'),
+    cfg.StrOpt(
+        'vitastor_etcd_prefix',
+        default='/vitastor',
+        help='Vitastor etcd prefix'
+    ),
+    cfg.StrOpt(
+        'vitastor_pool_id',
+        default='',
+        help='Vitastor pool ID to use for volumes'
+    ),
+    # FIXME exclusive_cinder_pool ?
+]
+
+CONF = cfg.CONF
+CONF.register_opts(VITASTOR_OPTS, group = configuration.SHARED_CONF_GROUP)
+
+class VitastorDriverException(exception.VolumeDriverException):
+    message = _("Vitastor Cinder driver failure: %(reason)s")
+
+@interface.volumedriver
+class VitastorDriver(driver.CloneableImageVD,
+    driver.ManageableVD, driver.ManageableSnapshotsVD,
+    driver.BaseVD):
+    """Implements Vitastor volume commands."""
+
+    cfg = {}
+    _etcd_urls = []
+
+    def __init__(self, active_backend_id = None, *args, **kwargs):
+        super(VitastorDriver, self).__init__(*args, **kwargs)
+        self.configuration.append_config_values(VITASTOR_OPTS)
+
+    @classmethod
+    def get_driver_options(cls):
+        additional_opts = cls._get_oslo_driver_opts(
+            'reserved_percentage',
+            'max_over_subscription_ratio',
+            'volume_dd_blocksize'
+        )
+        return VITASTOR_OPTS + additional_opts
+
+    def do_setup(self, context):
+        """Performs initialization steps that could raise exceptions."""
+        super(VitastorDriver, self).do_setup(context)
+        # Make sure configuration is in UTF-8
+        for attr in [ 'config_path', 'etcd_address', 'etcd_prefix', 'pool_id' ]:
+            val = self.configuration.safe_get('vitastor_'+attr)
+            if val is not None:
+                self.cfg[attr] = utils.convert_str(val)
+        self.cfg = self._load_config(self.cfg)
+
+    def _load_config(self, cfg):
+        # Try to load configuration file
+        try:
+            f = open(cfg['config_path'] or '/etc/vitastor/vitastor.conf')
+            conf = json.loads(f.read())
+            f.close()
+            for k in conf:
+                cfg[k] = cfg.get(k, conf[k])
+        except:
+            pass
+        if isinstance(cfg['etcd_address'], str):
+            cfg['etcd_address'] = cfg['etcd_address'].split(',')
+        # Sanitize etcd URLs
+        for i, etcd_url in enumerate(cfg['etcd_address']):
+            ssl = False
+            if etcd_url.lower().startswith('http://'):
+                etcd_url = etcd_url[7:]
+            elif etcd_url.lower().startswith('https://'):
+                etcd_url = etcd_url[8:]
+                ssl = True
+            if etcd_url.find('/') < 0:
+                etcd_url += '/v3'
+            if ssl:
+                etcd_url = 'https://'+etcd_url
+            else:
+                etcd_url = 'http://'+etcd_url
+            cfg['etcd_address'][i] = etcd_url
+        return cfg
+
+    def check_for_setup_error(self):
+        """Returns an error if prerequisites aren't met."""
+
+    def _encode_etcd_key(self, key):
+        if not isinstance(key, bytes):
+            key = str(key).encode('utf-8')
+        return base64.b64encode(self.cfg['etcd_prefix'].encode('utf-8')+b'/'+key).decode('utf-8')
+
+    def _encode_etcd_value(self, value):
+        if not isinstance(value, bytes):
+            value = str(value).encode('utf-8')
+        return base64.b64encode(value).decode('utf-8')
+
+    def _encode_etcd_requests(self, obj):
+        for v in obj:
+            for rt in v:
+                if 'key' in v[rt]:
+                    v[rt]['key'] = self._encode_etcd_key(v[rt]['key'])
+                if 'range_end' in v[rt]:
+                    v[rt]['range_end'] = self._encode_etcd_key(v[rt]['range_end'])
+                if 'value' in v[rt]:
+                    v[rt]['value'] = self._encode_etcd_value(v[rt]['value'])
+
+    def _etcd_txn(self, params):
+        if 'compare' in params:
+            for v in params['compare']:
+                if 'key' in v:
+                    v['key'] = self._encode_etcd_key(v['key'])
+        if 'failure' in params:
+            self._encode_etcd_requests(params['failure'])
+        if 'success' in params:
+            self._encode_etcd_requests(params['success'])
+        body = json.dumps(params).encode('utf-8')
+        headers = {
+            'Content-Type': 'application/json'
+        }
+        err = None
+        for etcd_url in self.cfg['etcd_address']:
+            try:
+                resp = request.urlopen(request.Request(etcd_url+'/kv/txn', body, headers), timeout = 5)
+                data = json.loads(resp.read())
+                if 'responses' not in data:
+                    data['responses'] = []
+                for i, resp in enumerate(data['responses']):
+                    if 'response_range' in resp:
+                        if 'kvs' not in resp['response_range']:
+                            resp['response_range']['kvs'] = []
+                        for kv in resp['response_range']['kvs']:
+                            kv['key'] = base64.b64decode(kv['key'].encode('utf-8')).decode('utf-8')
+                            if kv['key'].startswith(self.cfg['etcd_prefix']+'/'):
+                                kv['key'] = kv['key'][len(self.cfg['etcd_prefix'])+1 : ]
+                            kv['value'] = json.loads(base64.b64decode(kv['value'].encode('utf-8')))
+                    if len(resp.keys()) != 1:
+                        LOG.exception('unknown responses['+str(i)+'] format: '+json.dumps(resp))
+                    else:
+                        resp = data['responses'][i] = resp[list(resp.keys())[0]]
+                return data
+            except Exception as e:
+                LOG.exception('error calling etcd transaction: '+body.decode('utf-8')+'\nerror: '+str(e))
+                err = e
+        raise err
+
+    def _etcd_foreach(self, prefix, add_fn):
+        total = 0
+        batch = 1000
+        begin = prefix+'/'
+        while True:
+            resp = self._etcd_txn({ 'success': [
+                { 'request_range': {
+                    'key': begin,
+                    'range_end': prefix+'0',
+                    'limit': batch+1,
+                } },
+            ] })
+            i = 0
+            while i < batch and i < len(resp['responses'][0]['kvs']):
+                kv = resp['responses'][0]['kvs'][i]
+                add_fn(kv)
+                i += 1
+            if len(resp['responses'][0]['kvs']) <= batch:
+                break
+            begin = resp['responses'][0]['kvs'][batch]['key']
+        return total
+
+    def _update_volume_stats(self):
+        location_info = json.dumps({
+            'config': self.configuration.vitastor_config_path,
+            'etcd_address': self.configuration.vitastor_etcd_address,
+            'etcd_prefix': self.configuration.vitastor_etcd_prefix,
+            'pool_id': self.configuration.vitastor_pool_id,
+        })
+
+        stats = {
+            'vendor_name': 'Vitastor',
+            'driver_version': self.VERSION,
+            'storage_protocol': 'vitastor',
+            'total_capacity_gb': 'unknown',
+            'free_capacity_gb': 'unknown',
+            # FIXME check if safe_get is required
+            'reserved_percentage': self.configuration.safe_get('reserved_percentage'),
+            'multiattach': True,
+            'thin_provisioning_support': True,
+            'max_over_subscription_ratio': self.configuration.safe_get('max_over_subscription_ratio'),
+            'location_info': location_info,
+            'backend_state': 'down',
+            'volume_backend_name': self.configuration.safe_get('volume_backend_name') or 'vitastor',
+            'replication_enabled': False,
+        }
+
+        try:
+            pool_stats = self._etcd_txn({ 'success': [
+                { 'request_range': { 'key': 'pool/stats/'+str(self.cfg['pool_id']) } }
+            ] })
+            total_provisioned = 0
+            def add_total(kv):
+                nonlocal total_provisioned
+                if kv['key'].find('@') >= 0:
+                    total_provisioned += kv['value']['size']
+            self._etcd_foreach('config/inode/'+str(self.cfg['pool_id']), lambda kv: add_total(kv))
+            stats['provisioned_capacity_gb'] = round(total_provisioned/1024.0/1024.0/1024.0, 2)
+            pool_stats = pool_stats['responses'][0]['kvs']
+            if len(pool_stats):
+                pool_stats = pool_stats[0]['value']
+                stats['free_capacity_gb'] = round(1024.0*(pool_stats['total_raw_tb']-pool_stats['used_raw_tb'])/pool_stats['raw_to_usable'], 2)
+                stats['total_capacity_gb'] = round(1024.0*pool_stats['total_raw_tb'], 2)
+            stats['backend_state'] = 'up'
+        except Exception as e:
+            # just log and return unknown capacities
+            LOG.exception('error getting vitastor pool stats: '+str(e))
+
+        self._stats = stats
+        
+    def get_volume_stats(self, refresh=False):
+        """Get volume stats.
+        If 'refresh' is True, run update the stats first.
+        """
+        if not self._stats or refresh:
+            self._update_volume_stats()
+
+        return self._stats
+
+    def _next_id(self, resp):
+        if len(resp['kvs']) == 0:
+            return (1, 0)
+        else:
+            return (1 + resp['kvs'][0]['value'], resp['kvs'][0]['mod_revision'])
+
+    def create_volume(self, volume):
+        """Creates a logical volume."""
+
+        size = int(volume.size) * units.Gi
+        # FIXME: Check if convert_str is really required
+        vol_name = utils.convert_str(volume.name)
+        if vol_name.find('@') >= 0 or vol_name.find('/') >= 0:
+            raise exception.VolumeBackendAPIException(data = '@ and / are forbidden in volume and snapshot names')
+
+        LOG.debug("creating volume '%s'", vol_name)
+
+        self._create_image(vol_name, { 'size': size })
+
+        if volume.encryption_key_id:
+            self._create_encrypted_volume(volume, volume.obj_context)
+
+        volume_update = {}
+        return volume_update
+
+    def _create_encrypted_volume(self, volume, context):
+        """Create a new LUKS encrypted image directly in Vitastor."""
+        vol_name = utils.convert_str(volume.name)
+        f, opts = self._encrypt_opts(volume, context)
+        # FIXME: Check if it works at all :-)
+        self._execute(
+            'qemu-img', 'convert', '-f', 'luks', *opts,
+            'vitastor:image='+vol_name.replace(':', '\\:')+self._qemu_args(),
+            '%sM' % (volume.size * 1024)
+        )
+        f.close()
+
+    def _encrypt_opts(self, volume, context):
+        encryption = volume_utils.check_encryption_provider(self.db, volume, context)
+        # Fetch the key associated with the volume and decode the passphrase
+        keymgr = key_manager.API(CONF)
+        key = keymgr.get(context, encryption['encryption_key_id'])
+        passphrase = binascii.hexlify(key.get_encoded()).decode('utf-8')
+        # Decode the dm-crypt style cipher spec into something qemu-img can use
+        cipher_spec = image_utils.decode_cipher(encryption['cipher'], encryption['key_size'])
+        tmp_dir = volume_utils.image_conversion_dir()
+        f = tempfile.NamedTemporaryFile(prefix = 'luks_', dir = tmp_dir)
+        f.write(passphrase)
+        f.flush()
+        return (f, [
+            '--object', 'secret,id=luks_sec,format=raw,file=%(passfile)s' % {'passfile': f.name},
+            '-o', 'key-secret=luks_sec,cipher-alg=%(cipher_alg)s,cipher-mode=%(cipher_mode)s,ivgen-alg=%(ivgen_alg)s' % cipher_spec,
+        ])
+
+    def create_snapshot(self, snapshot):
+        """Creates a volume snapshot."""
+
+        vol_name = utils.convert_str(snapshot.volume_name)
+        snap_name = utils.convert_str(snapshot.name)
+        if snap_name.find('@') >= 0 or snap_name.find('/') >= 0:
+            raise exception.VolumeBackendAPIException(data = '@ and / are forbidden in volume and snapshot names')
+        self._create_snapshot(vol_name, vol_name+'@'+snap_name)
+
+    def snapshot_revert_use_temp_snapshot(self):
+        """Disable the use of a temporary snapshot on revert."""
+        return False
+
+    def revert_to_snapshot(self, context, volume, snapshot):
+        """Revert a volume to a given snapshot."""
+
+        vol_name = utils.convert_str(snapshot.volume_name)
+        snap_name = utils.convert_str(snapshot.name)
+
+        # Delete the image and recreate it from the snapshot
+        args = [ 'vitastor-cli', 'rm', vol_name, *(self._vitastor_args()) ]
+        try:
+            self._execute(*args)
+        except processutils.ProcessExecutionError as exc:
+            LOG.error("Failed to delete image "+vol_name+": "+exc)
+            raise exception.VolumeBackendAPIException(data = exc.stderr)
+        args = [
+            'vitastor-cli', 'create', '--parent', vol_name+'@'+snap_name,
+            vol_name, *(self._vitastor_args())
+        ]
+        try:
+            self._execute(*args)
+        except processutils.ProcessExecutionError as exc:
+            LOG.error("Failed to recreate image "+vol_name+" from "+vol_name+"@"+snap_name+": "+exc)
+            raise exception.VolumeBackendAPIException(data = exc.stderr)
+
+    def delete_snapshot(self, snapshot):
+        """Deletes a snapshot."""
+
+        vol_name = utils.convert_str(snapshot.volume_name)
+        snap_name = utils.convert_str(snapshot.name)
+
+        args = [
+            'vitastor-cli', 'rm', vol_name+'@'+snap_name,
+            *(self._vitastor_args())
+        ]
+        try:
+            self._execute(*args)
+        except processutils.ProcessExecutionError as exc:
+            LOG.error("Failed to remove snapshot "+vol_name+'@'+snap_name+": "+exc)
+            raise exception.VolumeBackendAPIException(data = exc.stderr)
+
+    def _child_count(self, parents):
+        children = 0
+        def add_child(kv):
+            nonlocal children
+            children += self._check_parent(kv, parents)
+        self._etcd_foreach('config/inode', lambda kv: add_child(kv))
+        return children
+
+    def _check_parent(self, kv, parents):
+        if 'parent_id' not in kv['value']:
+            return 0
+        parent_id = kv['value']['parent_id']
+        _, _, pool_id, inode_id = kv['key'].split('/')
+        parent_pool_id = pool_id
+        if 'parent_pool_id' in kv['value'] and kv['value']['parent_pool_id']:
+            parent_pool_id = kv['value']['parent_pool_id']
+        inode = (int(pool_id) << 48) | (int(inode_id) & 0xffffffffffff)
+        parent = (int(parent_pool_id) << 48) | (int(parent_id) & 0xffffffffffff)
+        if parent in parents and inode not in parents:
+            return 1
+        return 0
+
+    def create_cloned_volume(self, volume, src_vref):
+        """Create a cloned volume from another volume."""
+
+        size = int(volume.size) * units.Gi
+        src_name = utils.convert_str(src_vref.name)
+        dest_name = utils.convert_str(volume.name)
+        if dest_name.find('@') >= 0 or dest_name.find('/') >= 0:
+            raise exception.VolumeBackendAPIException(data = '@ and / are forbidden in volume and snapshot names')
+
+        # FIXME Do full copy if requested (cfg.disable_clone)
+
+        if src_vref.admin_metadata.get('readonly') == 'True':
+            # source volume is a volume-image cache entry or other readonly volume
+            # clone without intermediate snapshot
+            src = self._get_image(src_name)
+            LOG.debug("creating image '%s' from '%s'", dest_name, src_name)
+            new_cfg = self._create_image(dest_name, {
+                'size': size,
+                'parent_id': src['idx']['id'],
+                'parent_pool_id': src['idx']['pool_id'],
+            })
+            return {}
+
+        clone_snap = "%s@%s.clone_snap" % (src_name, dest_name)
+        make_img = True
+        if (volume.display_name and
+            volume.display_name.startswith('image-') and
+            src_vref.project_id != volume.project_id):
+            # idiotic openstack creates image-volume cache entries
+            # as clones of normal VM volumes... :-X prevent it :-D
+            clone_snap = dest_name
+            make_img = False
+
+        LOG.debug("creating layer '%s' under '%s'", clone_snap, src_name)
+        new_cfg = self._create_snapshot(src_name, clone_snap, True)
+        if make_img:
+            # Then create a clone from it
+            new_cfg = self._create_image(dest_name, {
+                'size': size,
+                'parent_id': new_cfg['parent_id'],
+                'parent_pool_id': new_cfg['parent_pool_id'],
+            })
+
+        return {}
+
+    def create_volume_from_snapshot(self, volume, snapshot):
+        """Creates a cloned volume from an existing snapshot."""
+
+        vol_name = utils.convert_str(volume.name)
+        snap_name = utils.convert_str(snapshot.name)
+
+        snap = self._get_image('volume-'+snapshot.volume_id+'@'+snap_name)
+        if not snap:
+            raise exception.SnapshotNotFound(snapshot_id = snap_name)
+        snap_inode_id = int(resp['responses'][0]['kvs'][0]['value']['id'])
+        snap_pool_id = int(resp['responses'][0]['kvs'][0]['value']['pool_id'])
+
+        size = snap['cfg']['size']
+        if int(volume.size):
+            size = int(volume.size) * units.Gi
+        new_cfg = self._create_image(vol_name, {
+            'size': size,
+            'parent_id': snap['idx']['id'],
+            'parent_pool_id': snap['idx']['pool_id'],
+        })
+
+        return {}
+
+    def _vitastor_args(self):
+        args = []
+        for k in [ 'config_path', 'etcd_address', 'etcd_prefix' ]:
+            v = self.configuration.safe_get('vitastor_'+k)
+            if v:
+                args.extend(['--'+k, v])
+        return args
+
+    def _qemu_args(self):
+        args = ''
+        for k in [ 'config_path', 'etcd_address', 'etcd_prefix' ]:
+            v = self.configuration.safe_get('vitastor_'+k)
+            kk = k
+            if kk == 'etcd_address':
+                # FIXME use etcd_address in qemu driver
+                kk = 'etcd_host'
+            if v:
+                args += ':'+kk.replace('_', '-')+'='+v.replace(':', '\\:')
+        return args
+
+    def delete_volume(self, volume):
+        """Deletes a logical volume."""
+
+        vol_name = utils.convert_str(volume.name)
+
+        # Find the volume and all its snapshots
+        range_end = b'index/image/' + vol_name.encode('utf-8')
+        range_end = range_end[0 : len(range_end)-1] + six.int2byte(range_end[len(range_end)-1] + 1)
+        resp = self._etcd_txn({ 'success': [
+            { 'request_range': { 'key': 'index/image/'+vol_name, 'range_end': range_end } },
+        ] })
+        if len(resp['responses'][0]['kvs']) == 0:
+            # already deleted
+            LOG.info("volume %s no longer exists in backend", vol_name)
+            return
+        layers = resp['responses'][0]['kvs']
+        layer_ids = {}
+        for kv in layers:
+            inode_id = int(kv['value']['id'])
+            pool_id = int(kv['value']['pool_id'])
+            inode_pool_id = (pool_id << 48) | (inode_id & 0xffffffffffff)
+            layer_ids[inode_pool_id] = True
+
+        # Check if the volume has clones and raise 'busy' if so
+        children = self._child_count(layer_ids)
+        if children > 0:
+            raise exception.VolumeIsBusy(volume_name = vol_name)
+
+        # Clear data
+        for kv in layers:
+            args = [
+                'vitastor-cli', 'rm-data', '--pool', str(kv['value']['pool_id']),
+                '--inode', str(kv['value']['id']), '--progress', '0',
+                *(self._vitastor_args())
+            ]
+            try:
+                self._execute(*args)
+            except processutils.ProcessExecutionError as exc:
+                LOG.error("Failed to remove layer "+kv['key']+": "+exc)
+                raise exception.VolumeBackendAPIException(data = exc.stderr)
+
+        # Delete all layers from etcd
+        requests = []
+        for kv in layers:
+            requests.append({ 'request_delete_range': { 'key': kv['key'] } })
+            requests.append({ 'request_delete_range': { 'key': 'config/inode/'+str(kv['value']['pool_id'])+'/'+str(kv['value']['id']) } })
+        self._etcd_txn({ 'success': requests })
+
+    def retype(self, context, volume, new_type, diff, host):
+        """Change extra type specifications for a volume."""
+
+        # FIXME Maybe (in the future) support multiple pools as different types
+        return True, {}
+
+    def ensure_export(self, context, volume):
+        """Synchronously recreates an export for a logical volume."""
+        pass
+
+    def create_export(self, context, volume, connector):
+        """Exports the volume."""
+        pass
+
+    def remove_export(self, context, volume):
+        """Removes an export for a logical volume."""
+        pass
+
+    def _create_image(self, vol_name, cfg):
+        pool_s = str(self.cfg['pool_id'])
+        image_id = 0
+        while image_id == 0:
+            # check if the image already exists and find a free ID
+            resp = self._etcd_txn({ 'success': [
+                { 'request_range': { 'key': 'index/image/'+vol_name } },
+                { 'request_range': { 'key': 'index/maxid/'+pool_s } },
+            ] })
+            if len(resp['responses'][0]['kvs']) > 0:
+                # already exists
+                raise exception.VolumeBackendAPIException(data = 'Volume '+vol_name+' already exists')
+            image_id, id_mod = self._next_id(resp['responses'][1])
+            # try to create the image
+            resp = self._etcd_txn({ 'compare': [
+                { 'target': 'MOD', 'mod_revision': id_mod, 'key': 'index/maxid/'+pool_s },
+                { 'target': 'VERSION', 'version': 0, 'key': 'index/image/'+vol_name },
+                { 'target': 'VERSION', 'version': 0, 'key': 'config/inode/'+pool_s+'/'+str(image_id) },
+            ], 'success': [
+                { 'request_put': { 'key': 'index/maxid/'+pool_s, 'value': image_id } },
+                { 'request_put': { 'key': 'index/image/'+vol_name, 'value': json.dumps({
+                    'id': image_id, 'pool_id': self.cfg['pool_id']
+                }) } },
+                { 'request_put': { 'key': 'config/inode/'+pool_s+'/'+str(image_id), 'value': json.dumps({
+                    **cfg, 'name': vol_name,
+                }) } },
+            ] })
+            if not resp.get('succeeded'):
+                # repeat
+                image_id = 0
+
+    def _create_snapshot(self, vol_name, snap_vol_name, allow_existing = False):
+        while True:
+            # check if the image already exists and snapshot doesn't
+            resp = self._etcd_txn({ 'success': [
+                { 'request_range': { 'key': 'index/image/'+vol_name } },
+                { 'request_range': { 'key': 'index/image/'+snap_vol_name } },
+            ] })
+            if len(resp['responses'][0]['kvs']) == 0:
+                raise exception.VolumeBackendAPIException(data = 'Volume '+vol_name+' does not exist')
+            if len(resp['responses'][1]['kvs']) > 0:
+                if allow_existing:
+                    snap_idx = resp['responses'][1]['kvs'][0]['value']
+                    resp = self._etcd_txn({ 'success': [
+                        { 'request_range': { 'key': 'config/inode/'+str(snap_idx['pool_id'])+'/'+str(snap_idx['id']) } },
+                    ] })
+                    if len(resp['responses'][0]['kvs']) == 0:
+                        raise exception.VolumeBackendAPIException(data =
+                            'Volume '+snap_vol_name+' is already indexed, but does not exist'
+                        )
+                    return resp['responses'][0]['kvs'][0]['value']
+                raise exception.VolumeBackendAPIException(
+                    data = 'Volume '+snap_vol_name+' already exists'
+                )
+            vol_idx = resp['responses'][0]['kvs'][0]['value']
+            vol_idx_mod = resp['responses'][0]['kvs'][0]['mod_revision']
+            # get image inode config and find a new ID
+            resp = self._etcd_txn({ 'success': [
+                { 'request_range': { 'key': 'config/inode/'+str(vol_idx['pool_id'])+'/'+str(vol_idx['id']) } },
+                { 'request_range': { 'key': 'index/maxid/'+str(self.cfg['pool_id']) } },
+            ] })
+            if len(resp['responses'][0]['kvs']) == 0:
+                raise exception.VolumeBackendAPIException(data = 'Volume '+vol_name+' does not exist')
+            vol_cfg = resp['responses'][0]['kvs'][0]['value']
+            vol_mod = resp['responses'][0]['kvs'][0]['mod_revision']
+            new_id, id_mod = self._next_id(resp['responses'][1])
+            # try to redirect image to the new inode
+            new_cfg = {
+                **vol_cfg, 'name': vol_name, 'parent_id': vol_idx['id'], 'parent_pool_id': vol_idx['pool_id']
+            }
+            resp = self._etcd_txn({ 'compare': [
+                { 'target': 'MOD', 'mod_revision': vol_idx_mod, 'key': 'index/image/'+vol_name },
+                { 'target': 'MOD', 'mod_revision': vol_mod, 'key': 'config/inode/'+str(vol_idx['pool_id'])+'/'+str(vol_idx['id']) },
+                { 'target': 'MOD', 'mod_revision': id_mod, 'key': 'index/maxid/'+str(self.cfg['pool_id']) },
+                { 'target': 'VERSION', 'version': 0, 'key': 'index/image/'+snap_vol_name },
+                { 'target': 'VERSION', 'version': 0, 'key': 'config/inode/'+str(self.cfg['pool_id'])+'/'+str(new_id) },
+            ], 'success': [
+                { 'request_put': { 'key': 'index/maxid/'+str(self.cfg['pool_id']), 'value': new_id } },
+                { 'request_put': { 'key': 'index/image/'+vol_name, 'value': json.dumps({
+                    'id': new_id, 'pool_id': self.cfg['pool_id']
+                }) } },
+                { 'request_put': { 'key': 'config/inode/'+str(self.cfg['pool_id'])+'/'+str(new_id), 'value': json.dumps(new_cfg) } },
+                { 'request_put': { 'key': 'index/image/'+snap_vol_name, 'value': json.dumps({
+                    'id': vol_idx['id'], 'pool_id': vol_idx['pool_id']
+                }) } },
+                { 'request_put': { 'key': 'config/inode/'+str(vol_idx['pool_id'])+'/'+str(vol_idx['id']), 'value': json.dumps({
+                    **vol_cfg, 'name': snap_vol_name, 'readonly': True
+                }) } }
+            ] })
+            if resp.get('succeeded'):
+                return new_cfg
+
+    def initialize_connection(self, volume, connector):
+        data = {
+            'driver_volume_type': 'vitastor',
+            'data': {
+                'config_path': self.configuration.vitastor_config_path,
+                'etcd_address': self.configuration.vitastor_etcd_address,
+                'etcd_prefix': self.configuration.vitastor_etcd_prefix,
+                'name': volume.name,
+                'logical_block_size': '512',
+                'physical_block_size': '4096',
+            }
+        }
+        LOG.debug('connection data: %s', data)
+        return data
+
+    def terminate_connection(self, volume, connector, **kwargs):
+        pass
+
+    def clone_image(self, context, volume, image_location, image_meta, image_service):
+        if image_location:
+            # Note: image_location[0] is glance image direct_url.
+            # image_location[1] contains the list of all locations (including
+            # direct_url) or None if show_multiple_locations is False in
+            # glance configuration.
+            if image_location[1]:
+                url_locations = [location['url'] for location in image_location[1]]
+            else:
+                url_locations = [image_location[0]]
+            # iterate all locations to look for a cloneable one.
+            for url_location in url_locations:
+                if url_location and url_location.startswith('cinder://'):
+                    # The idea is to use cinder://<volume-id> Glance volumes as base images
+                    base_vol = self.db.volume_get(context, url_location[len('cinder://') : ])
+                    if not base_vol or base_vol.volume_type_id != volume.volume_type_id:
+                        continue
+                    size = int(volume.size) * units.Gi
+                    dest_name = utils.convert_str(volume.name)
+                    # Find or create the base snapshot
+                    snap_cfg = self._create_snapshot(base_vol.name, base_vol.name+'@.clone_snap', True)
+                    # Then create a clone from it
+                    new_cfg = self._create_image(dest_name, {
+                        'size': size,
+                        'parent_id': snap_cfg['parent_id'],
+                        'parent_pool_id': snap_cfg['parent_pool_id'],
+                    })
+                    return ({}, True)
+        return ({}, False)
+
+    def copy_image_to_encrypted_volume(self, context, volume, image_service, image_id):
+        self.copy_image_to_volume(context, volume, image_service, image_id, encrypted = True)
+
+    def copy_image_to_volume(self, context, volume, image_service, image_id, encrypted = False, disable_sparse = False):
+        tmp_dir = volume_utils.image_conversion_dir()
+        with tempfile.NamedTemporaryFile(dir = tmp_dir) as tmp:
+            image_utils.fetch_to_raw(
+                context, image_service, image_id, tmp.name,
+                self.configuration.volume_dd_blocksize, size = volume.size
+            )
+            out_format = [ '-O', 'raw' ]
+            if encrypted:
+                key_file, opts = self._encrypt_opts(volume, context)
+                out_format = [ '-O', 'luks', *opts ]
+            dest_name = utils.convert_str(volume.name)
+            self._try_execute(
+                'qemu-img', 'convert', '-f', 'raw', tmp.name, *out_format,
+                'vitastor:image='+dest_name.replace(':', '\\:')+self._qemu_args()
+            )
+            if encrypted:
+                key_file.close()
+
+    def copy_volume_to_image(self, context, volume, image_service, image_meta):
+        tmp_dir = volume_utils.image_conversion_dir()
+        tmp_file = os.path.join(tmp_dir, volume.name + '-' + image_meta['id'])
+        with fileutils.remove_path_on_error(tmp_file):
+            vol_name = utils.convert_str(volume.name)
+            self._try_execute(
+                'qemu-img', 'convert', '-f', 'raw',
+                'vitastor:image='+vol_name.replace(':', '\\:')+self._qemu_args(),
+                '-O', 'raw', tmp_file
+            )
+            # FIXME: Copy directly if the destination image is also in Vitastor
+            volume_utils.upload_volume(context, image_service, image_meta, tmp_file, volume)
+        os.unlink(tmp_file)
+
+    def _get_image(self, vol_name):
+        # find the image
+        resp = self._etcd_txn({ 'success': [
+            { 'request_range': { 'key': 'index/image/'+vol_name } },
+        ] })
+        if len(resp['responses'][0]['kvs']) == 0:
+            return None
+        vol_idx = resp['responses'][0]['kvs'][0]['value']
+        vol_idx_mod = resp['responses'][0]['kvs'][0]['mod_revision']
+        # get image inode config
+        resp = self._etcd_txn({ 'success': [
+            { 'request_range': { 'key': 'config/inode/'+str(vol_idx['pool_id'])+'/'+str(vol_idx['id']) } },
+        ] })
+        if len(resp['responses'][0]['kvs']) == 0:
+            return None
+        vol_cfg = resp['responses'][0]['kvs'][0]['value']
+        vol_cfg_mod = resp['responses'][0]['kvs'][0]['mod_revision']
+        return {
+            'cfg': vol_cfg,
+            'cfg_mod': vol_cfg_mod,
+            'idx': vol_idx,
+            'idx_mod': vol_idx_mod,
+        }
+
+    def extend_volume(self, volume, new_size):
+        """Extend an existing volume."""
+        vol_name = utils.convert_str(volume.name)
+        while True:
+            vol = self._get_image(vol_name)
+            if not vol:
+                raise exception.VolumeBackendAPIException(data = 'Volume '+vol_name+' does not exist')
+            # change size
+            size = int(new_size) * units.Gi
+            if size == vol['cfg']['size']:
+                break
+            resp = self._etcd_txn({ 'compare': [ {
+                'target': 'MOD',
+                'mod_revision': vol['cfg_mod'],
+                'key': 'config/inode/'+str(vol['idx']['pool_id'])+'/'+str(vol['idx']['id']),
+            } ], 'success': [
+                { 'request_put': {
+                    'key': 'config/inode/'+str(vol['idx']['pool_id'])+'/'+str(vol['idx']['id']),
+                    'value': json.dumps({ **vol['cfg'], 'size': size }),
+                } },
+            ] })
+            if resp.get('succeeded'):
+                break
+        LOG.debug(
+            "Extend volume from %(old_size)s GB to %(new_size)s GB.",
+            {'old_size': volume.size, 'new_size': new_size}
+        )
+
+    def _add_manageable_volume(self, kv, manageable_volumes, cinder_ids):
+        cfg = kv['value']
+        if kv['key'].find('@') >= 0:
+            # snapshot
+            return
+        image_id = volume_utils.extract_id_from_volume_name(cfg['name'])
+        image_info = {
+            'reference': {'source-name': image_name},
+            'size': int(math.ceil(float(cfg['size']) / units.Gi)),
+            'cinder_id': None,
+            'extra_info': None,
+        }
+        if image_id in cinder_ids:
+            image_info['cinder_id'] = image_id
+            image_info['safe_to_manage'] = False
+            image_info['reason_not_safe'] = 'already managed'
+        else:
+            image_info['safe_to_manage'] = True
+            image_info['reason_not_safe'] = None
+        manageable_volumes.append(image_info)
+
+    def get_manageable_volumes(self, cinder_volumes, marker, limit, offset, sort_keys, sort_dirs):
+        manageable_volumes = []
+        cinder_ids = [resource['id'] for resource in cinder_volumes]
+
+        # List all volumes
+        # FIXME: It's possible to use pagination in our case, but.. do we want it?
+        self._etcd_foreach('config/inode/'+str(self.cfg['pool_id']),
+            lambda kv: self._add_manageable_volume(kv, manageable_volumes, cinder_ids))
+
+        return volume_utils.paginate_entries_list(
+            manageable_volumes, marker, limit, offset, sort_keys, sort_dirs)
+
+    def _get_existing_name(existing_ref):
+        if not isinstance(existing_ref, dict):
+            existing_ref = {"source-name": existing_ref}
+        if 'source-name' not in existing_ref:
+            reason = _('Reference must contain source-name element.')
+            raise exception.ManageExistingInvalidReference(existing_ref=existing_ref, reason=reason)
+        src_name = utils.convert_str(existing_ref['source-name'])
+        if not src_name:
+            reason = _('Reference must contain source-name element.')
+            raise exception.ManageExistingInvalidReference(existing_ref=existing_ref, reason=reason)
+        return src_name
+
+    def manage_existing_get_size(self, volume, existing_ref):
+        """Return size of an existing image for manage_existing.
+
+        :param volume: volume ref info to be set
+        :param existing_ref: {'source-name': <image name>}
+        """
+        src_name = self._get_existing_name(existing_ref)
+        vol = self._get_image(src_name)
+        if not vol:
+            raise exception.VolumeBackendAPIException(data = 'Volume '+src_name+' does not exist')
+        return int(math.ceil(float(vol['cfg']['size']) / units.Gi))
+
+    def manage_existing(self, volume, existing_ref):
+        """Manages an existing image.
+
+        Renames the image name to match the expected name for the volume.
+
+        :param volume: volume ref info to be set
+        :param existing_ref: {'source-name': <image name>}
+        """
+        from_name = self._get_existing_name(existing_ref)
+        to_name = utils.convert_str(volume.name)
+        self._rename(from_name, to_name)
+
+    def _rename(self, from_name, to_name):
+        while True:
+            vol = self._get_image(from_name)
+            if not vol:
+                raise exception.VolumeBackendAPIException(data = 'Volume '+from_name+' does not exist')
+            to = self._get_image(to_name)
+            if to:
+                raise exception.VolumeBackendAPIException(data = 'Volume '+to_name+' already exists')
+            resp = self._etcd_txn({ 'compare': [
+                { 'target': 'MOD', 'mod_revision': vol['idx_mod'], 'key': 'index/image/'+vol['cfg']['name'] },
+                { 'target': 'MOD', 'mod_revision': vol['cfg_mod'], 'key': 'config/inode/'+str(vol['idx']['pool_id'])+'/'+str(vol['idx']['id']) },
+                { 'target': 'VERSION', 'version': 0, 'key': 'index/image/'+to_name },
+            ], 'success': [
+                { 'request_delete_range': { 'key': 'index/image/'+vol['cfg']['name'] } },
+                { 'request_put': { 'key': 'index/image/'+to_name, 'value': json.dumps(vol['idx']) } },
+                { 'request_put': { 'key': 'config/inode/'+str(vol['idx']['pool_id'])+'/'+str(vol['idx']['id']),
+                    'value': json.dumps({ **vol['cfg'], 'name': to_name }) } },
+            ] })
+            if resp.get('succeeded'):
+                break
+
+    def unmanage(self, volume):
+        pass
+
+    def _add_manageable_snapshot(self, kv, manageable_snapshots, cinder_ids):
+        cfg = kv['value']
+        dog = kv['key'].find('@')
+        if dog < 0:
+            # snapshot
+            return
+        image_name = kv['key'][0 : dog]
+        snap_name = kv['key'][dog+1 : ]
+        snapshot_id = volume_utils.extract_id_from_snapshot_name(snap_name)
+        snapshot_info = {
+            'reference': {'source-name': snap_name},
+            'size': int(math.ceil(float(cfg['size']) / units.Gi)),
+            'cinder_id': None,
+            'extra_info': None,
+            'safe_to_manage': False,
+            'reason_not_safe': None,
+            'source_reference': {'source-name': image_name}
+        }
+        if snapshot_id in cinder_ids:
+            # Exclude snapshots already managed.
+            snapshot_info['reason_not_safe'] = ('already managed')
+            snapshot_info['cinder_id'] = snapshot_id
+        elif snap_name.endswith('.clone_snap'):
+            # Exclude clone snapshot.
+            snapshot_info['reason_not_safe'] = ('used for clone snap')
+        else:
+            snapshot_info['safe_to_manage'] = True
+        manageable_snapshots.append(snapshot_info)
+
+    def get_manageable_snapshots(self, cinder_snapshots, marker, limit, offset, sort_keys, sort_dirs):
+        """List manageable snapshots in Vitastor."""
+        manageable_snapshots = []
+        cinder_snapshot_ids = [resource['id'] for resource in cinder_snapshots]
+        # List all volumes
+        # FIXME: It's possible to use pagination in our case, but.. do we want it?
+        self._etcd_foreach('config/inode/'+str(self.cfg['pool_id']),
+            lambda kv: self._add_manageable_volume(kv, manageable_snapshots, cinder_snapshot_ids))
+        return volume_utils.paginate_entries_list(
+            manageable_snapshots, marker, limit, offset, sort_keys, sort_dirs)
+
+    def manage_existing_snapshot_get_size(self, snapshot, existing_ref):
+        """Return size of an existing image for manage_existing.
+
+        :param snapshot: snapshot ref info to be set
+        :param existing_ref: {'source-name': <name of snapshot>}
+        """
+        vol_name = utils.convert_str(snapshot.volume_name)
+        snap_name = self._get_existing_name(existing_ref)
+        vol = self._get_image(vol_name+'@'+snap_name)
+        if not vol:
+            raise exception.ManageExistingInvalidReference(
+                existing_ref=snapshot_name, reason='Specified snapshot does not exist.'
+            )
+        return int(math.ceil(float(vol['cfg']['size']) / units.Gi))
+
+    def manage_existing_snapshot(self, snapshot, existing_ref):
+        """Manages an existing snapshot.
+
+        Renames the snapshot name to match the expected name for the snapshot.
+        Error checking done by manage_existing_get_size is not repeated.
+
+        :param snapshot: snapshot ref info to be set
+        :param existing_ref: {'source-name': <name of snapshot>}
+        """
+        vol_name = utils.convert_str(snapshot.volume_name)
+        snap_name = self._get_existing_name(existing_ref)
+        from_name = vol_name+'@'+snap_name
+        to_name = vol_name+'@'+utils.convert_str(snapshot.name)
+        self._rename(from_name, to_name)
+
+    def unmanage_snapshot(self, snapshot):
+        """Removes the specified snapshot from Cinder management."""
+        pass
+
+    def _dumps(self, obj):
+        return json.dumps(obj, separators=(',', ':'), sort_keys=True)

patches/libvirt-10.4-vitastor.diff

@@ -0,0 +1,643 @@
+commit 1f7e90e36b2afca0312392979b96d31951a8d66b
+Author: Vitaliy Filippov <vitalif@yourcmc.ru>
+Date:   Thu Jun 27 01:34:54 2024 +0300
+
+    Add Vitastor support
+
+diff --git a/include/libvirt/libvirt-storage.h b/include/libvirt/libvirt-storage.h
+index aaad4a3da1..5f5daa8341 100644
+--- a/include/libvirt/libvirt-storage.h
++++ b/include/libvirt/libvirt-storage.h
+@@ -326,6 +326,7 @@ typedef enum {
+     VIR_CONNECT_LIST_STORAGE_POOLS_ZFS           = 1 << 17, /* (Since: 1.2.8) */
+     VIR_CONNECT_LIST_STORAGE_POOLS_VSTORAGE      = 1 << 18, /* (Since: 3.1.0) */
+     VIR_CONNECT_LIST_STORAGE_POOLS_ISCSI_DIRECT  = 1 << 19, /* (Since: 5.6.0) */
++    VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR      = 1 << 20, /* (Since: 5.0.0) */
+ } virConnectListAllStoragePoolsFlags;
+ 
+ int                     virConnectListAllStoragePools(virConnectPtr conn,
+diff --git a/src/conf/domain_conf.c b/src/conf/domain_conf.c
+index fde594f811..66537db3e3 100644
+--- a/src/conf/domain_conf.c
++++ b/src/conf/domain_conf.c
+@@ -7220,7 +7220,8 @@ virDomainDiskSourceNetworkParse(xmlNodePtr node,
+     src->configFile = virXPathString("string(./config/@file)", ctxt);
+ 
+     if (src->protocol == VIR_STORAGE_NET_PROTOCOL_HTTP ||
+-        src->protocol == VIR_STORAGE_NET_PROTOCOL_HTTPS)
++        src->protocol == VIR_STORAGE_NET_PROTOCOL_HTTPS ||
++        src->protocol == VIR_STORAGE_NET_PROTOCOL_VITASTOR)
+         src->query = virXMLPropString(node, "query");
+ 
+     if (virDomainStorageNetworkParseHosts(node, ctxt, &src->hosts, &src->nhosts) < 0)
+@@ -30734,6 +30735,7 @@ virDomainStorageSourceTranslateSourcePool(virStorageSource *src,
+ 
+     case VIR_STORAGE_POOL_MPATH:
+     case VIR_STORAGE_POOL_RBD:
++    case VIR_STORAGE_POOL_VITASTOR:
+     case VIR_STORAGE_POOL_SHEEPDOG:
+     case VIR_STORAGE_POOL_GLUSTER:
+     case VIR_STORAGE_POOL_LAST:
+diff --git a/src/conf/domain_validate.c b/src/conf/domain_validate.c
+index 395e036e8f..8a0190f85b 100644
+--- a/src/conf/domain_validate.c
++++ b/src/conf/domain_validate.c
+@@ -495,6 +495,7 @@ virDomainDiskDefValidateSourceChainOne(const virStorageSource *src)
+         case VIR_STORAGE_NET_PROTOCOL_RBD:
+             break;
+ 
++        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+         case VIR_STORAGE_NET_PROTOCOL_NBD:
+         case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+         case VIR_STORAGE_NET_PROTOCOL_GLUSTER:
+@@ -541,7 +542,7 @@ virDomainDiskDefValidateSourceChainOne(const virStorageSource *src)
+         }
+     }
+ 
+-    /* internal snapshots and config files are currently supported only with rbd: */
++    /* internal snapshots are currently supported only with rbd: */
+     if (virStorageSourceGetActualType(src) != VIR_STORAGE_TYPE_NETWORK &&
+         src->protocol != VIR_STORAGE_NET_PROTOCOL_RBD) {
+         if (src->snapshot) {
+@@ -549,10 +550,15 @@ virDomainDiskDefValidateSourceChainOne(const virStorageSource *src)
+                            _("<snapshot> element is currently supported only with 'rbd' disks"));
+             return -1;
+         }
++    }
+ 
++    /* config files are currently supported only with rbd and vitastor: */
++    if (virStorageSourceGetActualType(src) != VIR_STORAGE_TYPE_NETWORK &&
++        src->protocol != VIR_STORAGE_NET_PROTOCOL_RBD &&
++        src->protocol != VIR_STORAGE_NET_PROTOCOL_VITASTOR) {
+         if (src->configFile) {
+             virReportError(VIR_ERR_XML_ERROR, "%s",
+-                           _("<config> element is currently supported only with 'rbd' disks"));
++                           _("<config> element is currently supported only with 'rbd' and 'vitastor' disks"));
+             return -1;
+         }
+     }
+diff --git a/src/conf/schemas/domaincommon.rng b/src/conf/schemas/domaincommon.rng
+index a46a824f88..4c5b720643 100644
+--- a/src/conf/schemas/domaincommon.rng
++++ b/src/conf/schemas/domaincommon.rng
+@@ -1997,6 +1997,35 @@
+     </element>
+   </define>
+ 
++  <define name="diskSourceNetworkProtocolVitastor">
++    <element name="source">
++      <interleave>
++        <attribute name="protocol">
++          <value>vitastor</value>
++        </attribute>
++        <ref name="diskSourceCommon"/>
++        <optional>
++          <attribute name="name"/>
++        </optional>
++        <optional>
++          <attribute name="query"/>
++        </optional>
++        <zeroOrMore>
++          <ref name="diskSourceNetworkHost"/>
++        </zeroOrMore>
++        <optional>
++          <element name="config">
++            <attribute name="file">
++              <ref name="absFilePath"/>
++            </attribute>
++            <empty/>
++          </element>
++        </optional>
++        <empty/>
++      </interleave>
++    </element>
++  </define>
++
+   <define name="diskSourceNetworkProtocolISCSI">
+     <element name="source">
+       <attribute name="protocol">
+@@ -2347,6 +2376,7 @@
+       <ref name="diskSourceNetworkProtocolSimple"/>
+       <ref name="diskSourceNetworkProtocolVxHS"/>
+       <ref name="diskSourceNetworkProtocolNFS"/>
++      <ref name="diskSourceNetworkProtocolVitastor"/>
+     </choice>
+   </define>
+ 
+diff --git a/src/conf/storage_conf.c b/src/conf/storage_conf.c
+index 68842004b7..1d69a788b6 100644
+--- a/src/conf/storage_conf.c
++++ b/src/conf/storage_conf.c
+@@ -56,7 +56,7 @@ VIR_ENUM_IMPL(virStoragePool,
+               "logical", "disk", "iscsi",
+               "iscsi-direct", "scsi", "mpath",
+               "rbd", "sheepdog", "gluster",
+-              "zfs", "vstorage",
++              "zfs", "vstorage", "vitastor",
+ );
+ 
+ VIR_ENUM_IMPL(virStoragePoolFormatFileSystem,
+@@ -242,6 +242,18 @@ static virStoragePoolTypeInfo poolTypeInfo[] = {
+           .formatToString = virStorageFileFormatTypeToString,
+       }
+     },
++    {.poolType = VIR_STORAGE_POOL_VITASTOR,
++     .poolOptions = {
++         .flags = (VIR_STORAGE_POOL_SOURCE_HOST |
++                   VIR_STORAGE_POOL_SOURCE_NETWORK |
++                   VIR_STORAGE_POOL_SOURCE_NAME),
++      },
++      .volOptions = {
++          .defaultFormat = VIR_STORAGE_FILE_RAW,
++          .formatFromString = virStorageVolumeFormatFromString,
++          .formatToString = virStorageFileFormatTypeToString,
++      }
++    },
+     {.poolType = VIR_STORAGE_POOL_SHEEPDOG,
+      .poolOptions = {
+          .flags = (VIR_STORAGE_POOL_SOURCE_HOST |
+@@ -538,6 +550,11 @@ virStoragePoolDefParseSource(xmlXPathContextPtr ctxt,
+                        _("element 'name' is mandatory for RBD pool"));
+         return -1;
+     }
++    if (pool_type == VIR_STORAGE_POOL_VITASTOR && source->name == NULL) {
++        virReportError(VIR_ERR_XML_ERROR, "%s",
++                       _("element 'name' is mandatory for Vitastor pool"));
++        return -1;
++    }
+ 
+     if (options->formatFromString) {
+         g_autofree char *format = NULL;
+@@ -1127,6 +1144,7 @@ virStoragePoolDefFormatBuf(virBuffer *buf,
+     /* RBD, Sheepdog, Gluster and Iscsi-direct devices are not local block devs nor
+      * files, so they don't have a target */
+     if (def->type != VIR_STORAGE_POOL_RBD &&
++        def->type != VIR_STORAGE_POOL_VITASTOR &&
+         def->type != VIR_STORAGE_POOL_SHEEPDOG &&
+         def->type != VIR_STORAGE_POOL_GLUSTER &&
+         def->type != VIR_STORAGE_POOL_ISCSI_DIRECT) {
+diff --git a/src/conf/storage_conf.h b/src/conf/storage_conf.h
+index fc67957cfe..720c07ef74 100644
+--- a/src/conf/storage_conf.h
++++ b/src/conf/storage_conf.h
+@@ -103,6 +103,7 @@ typedef enum {
+     VIR_STORAGE_POOL_GLUSTER,  /* Gluster device */
+     VIR_STORAGE_POOL_ZFS,      /* ZFS */
+     VIR_STORAGE_POOL_VSTORAGE, /* Virtuozzo Storage */
++    VIR_STORAGE_POOL_VITASTOR, /* Vitastor */
+ 
+     VIR_STORAGE_POOL_LAST,
+ } virStoragePoolType;
+@@ -454,6 +455,7 @@ VIR_ENUM_DECL(virStoragePartedFs);
+                  VIR_CONNECT_LIST_STORAGE_POOLS_SCSI     | \
+                  VIR_CONNECT_LIST_STORAGE_POOLS_MPATH    | \
+                  VIR_CONNECT_LIST_STORAGE_POOLS_RBD      | \
++                 VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR | \
+                  VIR_CONNECT_LIST_STORAGE_POOLS_SHEEPDOG | \
+                  VIR_CONNECT_LIST_STORAGE_POOLS_GLUSTER  | \
+                  VIR_CONNECT_LIST_STORAGE_POOLS_ZFS      | \
+diff --git a/src/conf/storage_source_conf.c b/src/conf/storage_source_conf.c
+index 959ec5ed40..e751dd4d6a 100644
+--- a/src/conf/storage_source_conf.c
++++ b/src/conf/storage_source_conf.c
+@@ -88,6 +88,7 @@ VIR_ENUM_IMPL(virStorageNetProtocol,
+               "ssh",
+               "vxhs",
+               "nfs",
++              "vitastor",
+ );
+ 
+ 
+@@ -1301,6 +1302,7 @@ virStorageSourceNetworkDefaultPort(virStorageNetProtocol protocol)
+         case VIR_STORAGE_NET_PROTOCOL_GLUSTER:
+             return 24007;
+ 
++        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+         case VIR_STORAGE_NET_PROTOCOL_RBD:
+             /* we don't provide a default for RBD */
+             return 0;
+diff --git a/src/conf/storage_source_conf.h b/src/conf/storage_source_conf.h
+index 05b4bda16c..b5ed143c39 100644
+--- a/src/conf/storage_source_conf.h
++++ b/src/conf/storage_source_conf.h
+@@ -129,6 +129,7 @@ typedef enum {
+     VIR_STORAGE_NET_PROTOCOL_SSH,
+     VIR_STORAGE_NET_PROTOCOL_VXHS,
+     VIR_STORAGE_NET_PROTOCOL_NFS,
++    VIR_STORAGE_NET_PROTOCOL_VITASTOR,
+ 
+     VIR_STORAGE_NET_PROTOCOL_LAST
+ } virStorageNetProtocol;
+diff --git a/src/conf/virstorageobj.c b/src/conf/virstorageobj.c
+index 59fa5da372..4739167f5f 100644
+--- a/src/conf/virstorageobj.c
++++ b/src/conf/virstorageobj.c
+@@ -1438,6 +1438,7 @@ virStoragePoolObjSourceFindDuplicateCb(const void *payload,
+             return 1;
+         break;
+ 
++    case VIR_STORAGE_POOL_VITASTOR:
+     case VIR_STORAGE_POOL_ISCSI_DIRECT:
+     case VIR_STORAGE_POOL_RBD:
+     case VIR_STORAGE_POOL_LAST:
+@@ -1921,6 +1922,8 @@ virStoragePoolObjMatch(virStoragePoolObj *obj,
+                (obj->def->type == VIR_STORAGE_POOL_MPATH))   ||
+               (MATCH(VIR_CONNECT_LIST_STORAGE_POOLS_RBD) &&
+                (obj->def->type == VIR_STORAGE_POOL_RBD))     ||
++              (MATCH(VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR) &&
++               (obj->def->type == VIR_STORAGE_POOL_VITASTOR)) ||
+               (MATCH(VIR_CONNECT_LIST_STORAGE_POOLS_SHEEPDOG) &&
+                (obj->def->type == VIR_STORAGE_POOL_SHEEPDOG)) ||
+               (MATCH(VIR_CONNECT_LIST_STORAGE_POOLS_GLUSTER) &&
+diff --git a/src/libvirt-storage.c b/src/libvirt-storage.c
+index db7660aac4..561df34709 100644
+--- a/src/libvirt-storage.c
++++ b/src/libvirt-storage.c
+@@ -94,6 +94,7 @@ virStoragePoolGetConnect(virStoragePoolPtr pool)
+  * VIR_CONNECT_LIST_STORAGE_POOLS_SCSI
+  * VIR_CONNECT_LIST_STORAGE_POOLS_MPATH
+  * VIR_CONNECT_LIST_STORAGE_POOLS_RBD
++ * VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR
+  * VIR_CONNECT_LIST_STORAGE_POOLS_SHEEPDOG
+  * VIR_CONNECT_LIST_STORAGE_POOLS_GLUSTER
+  * VIR_CONNECT_LIST_STORAGE_POOLS_ZFS
+diff --git a/src/libxl/libxl_conf.c b/src/libxl/libxl_conf.c
+index 62e1be6672..71a1d42896 100644
+--- a/src/libxl/libxl_conf.c
++++ b/src/libxl/libxl_conf.c
+@@ -979,6 +979,7 @@ libxlMakeNetworkDiskSrcStr(virStorageSource *src,
+     case VIR_STORAGE_NET_PROTOCOL_SSH:
+     case VIR_STORAGE_NET_PROTOCOL_VXHS:
+     case VIR_STORAGE_NET_PROTOCOL_NFS:
++    case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+     case VIR_STORAGE_NET_PROTOCOL_LAST:
+     case VIR_STORAGE_NET_PROTOCOL_NONE:
+         virReportError(VIR_ERR_NO_SUPPORT,
+diff --git a/src/libxl/xen_xl.c b/src/libxl/xen_xl.c
+index 53f6871efc..c34b8cee1a 100644
+--- a/src/libxl/xen_xl.c
++++ b/src/libxl/xen_xl.c
+@@ -1456,6 +1456,7 @@ xenFormatXLDiskSrcNet(virStorageSource *src)
+     case VIR_STORAGE_NET_PROTOCOL_SSH:
+     case VIR_STORAGE_NET_PROTOCOL_VXHS:
+     case VIR_STORAGE_NET_PROTOCOL_NFS:
++    case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+     case VIR_STORAGE_NET_PROTOCOL_LAST:
+     case VIR_STORAGE_NET_PROTOCOL_NONE:
+         virReportError(VIR_ERR_NO_SUPPORT,
+diff --git a/src/qemu/qemu_block.c b/src/qemu/qemu_block.c
+index 738b72d7ea..5dd082fc89 100644
+--- a/src/qemu/qemu_block.c
++++ b/src/qemu/qemu_block.c
+@@ -758,6 +758,38 @@ qemuBlockStorageSourceGetRBDProps(virStorageSource *src,
+ }
+ 
+ 
++static virJSONValue *
++qemuBlockStorageSourceGetVitastorProps(virStorageSource *src)
++{
++    virJSONValue *ret = NULL;
++    virStorageNetHostDef *host;
++    size_t i;
++    g_auto(virBuffer) buf = VIR_BUFFER_INITIALIZER;
++    g_autofree char *etcd = NULL;
++
++    for (i = 0; i < src->nhosts; i++) {
++        host = src->hosts + i;
++        if ((virStorageNetHostTransport)host->transport != VIR_STORAGE_NET_HOST_TRANS_TCP) {
++            return NULL;
++        }
++        virBufferAsprintf(&buf, i > 0 ? ",%s:%u" : "%s:%u", host->name, host->port);
++    }
++    if (src->nhosts > 0) {
++        etcd = virBufferContentAndReset(&buf);
++    }
++
++    if (virJSONValueObjectAdd(&ret,
++                              "S:etcd-host", etcd,
++                              "S:etcd-prefix", src->query,
++                              "S:config-path", src->configFile,
++                              "s:image", src->path,
++                              NULL) < 0)
++        return NULL;
++
++    return ret;
++}
++
++
+ static virJSONValue *
+ qemuBlockStorageSourceGetSheepdogProps(virStorageSource *src)
+ {
+@@ -1140,6 +1172,12 @@ qemuBlockStorageSourceGetBackendProps(virStorageSource *src,
+                 return NULL;
+             break;
+ 
++        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
++            driver = "vitastor";
++            if (!(fileprops = qemuBlockStorageSourceGetVitastorProps(src)))
++                return NULL;
++            break;
++
+         case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+             driver = "sheepdog";
+             if (!(fileprops = qemuBlockStorageSourceGetSheepdogProps(src)))
+@@ -2020,6 +2058,7 @@ qemuBlockGetBackingStoreString(virStorageSource *src,
+ 
+             case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+             case VIR_STORAGE_NET_PROTOCOL_RBD:
++            case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+             case VIR_STORAGE_NET_PROTOCOL_VXHS:
+             case VIR_STORAGE_NET_PROTOCOL_NFS:
+             case VIR_STORAGE_NET_PROTOCOL_SSH:
+@@ -2400,6 +2439,12 @@ qemuBlockStorageSourceCreateGetStorageProps(virStorageSource *src,
+                 return -1;
+             break;
+ 
++        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
++            driver = "vitastor";
++            if (!(location = qemuBlockStorageSourceGetVitastorProps(src)))
++                return -1;
++            break;
++
+         case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+             driver = "sheepdog";
+             if (!(location = qemuBlockStorageSourceGetSheepdogProps(src)))
+diff --git a/src/qemu/qemu_domain.c b/src/qemu/qemu_domain.c
+index bda62f2e5c..84b4e5f2b8 100644
+--- a/src/qemu/qemu_domain.c
++++ b/src/qemu/qemu_domain.c
+@@ -5260,7 +5260,8 @@ qemuDomainValidateStorageSource(virStorageSource *src,
+     if (src->query &&
+         (actualType != VIR_STORAGE_TYPE_NETWORK ||
+          (src->protocol != VIR_STORAGE_NET_PROTOCOL_HTTPS &&
+-          src->protocol != VIR_STORAGE_NET_PROTOCOL_HTTP))) {
++          src->protocol != VIR_STORAGE_NET_PROTOCOL_HTTP &&
++          src->protocol != VIR_STORAGE_NET_PROTOCOL_VITASTOR))) {
+         virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
+                        _("query is supported only with HTTP(S) protocols"));
+         return -1;
+@@ -10514,6 +10515,7 @@ qemuDomainPrepareStorageSourceTLS(virStorageSource *src,
+         break;
+ 
+     case VIR_STORAGE_NET_PROTOCOL_RBD:
++    case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+     case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+     case VIR_STORAGE_NET_PROTOCOL_GLUSTER:
+     case VIR_STORAGE_NET_PROTOCOL_ISCSI:
+diff --git a/src/qemu/qemu_snapshot.c b/src/qemu/qemu_snapshot.c
+index f5260c4a22..2f9d8406fe 100644
+--- a/src/qemu/qemu_snapshot.c
++++ b/src/qemu/qemu_snapshot.c
+@@ -423,6 +423,7 @@ qemuSnapshotPrepareDiskExternalInactive(virDomainSnapshotDiskDef *snapdisk,
+         case VIR_STORAGE_NET_PROTOCOL_NONE:
+         case VIR_STORAGE_NET_PROTOCOL_NBD:
+         case VIR_STORAGE_NET_PROTOCOL_RBD:
++        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+         case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+         case VIR_STORAGE_NET_PROTOCOL_GLUSTER:
+         case VIR_STORAGE_NET_PROTOCOL_ISCSI:
+@@ -648,6 +649,7 @@ qemuSnapshotPrepareDiskInternal(virDomainDiskDef *disk,
+         case VIR_STORAGE_NET_PROTOCOL_NONE:
+         case VIR_STORAGE_NET_PROTOCOL_NBD:
+         case VIR_STORAGE_NET_PROTOCOL_RBD:
++        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+         case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+         case VIR_STORAGE_NET_PROTOCOL_GLUSTER:
+         case VIR_STORAGE_NET_PROTOCOL_ISCSI:
+diff --git a/src/storage/storage_driver.c b/src/storage/storage_driver.c
+index 86c03762d2..630c6eff1a 100644
+--- a/src/storage/storage_driver.c
++++ b/src/storage/storage_driver.c
+@@ -1626,6 +1626,7 @@ storageVolLookupByPathCallback(virStoragePoolObj *obj,
+ 
+         case VIR_STORAGE_POOL_GLUSTER:
+         case VIR_STORAGE_POOL_RBD:
++        case VIR_STORAGE_POOL_VITASTOR:
+         case VIR_STORAGE_POOL_SHEEPDOG:
+         case VIR_STORAGE_POOL_ZFS:
+         case VIR_STORAGE_POOL_LAST:
+diff --git a/src/storage_file/storage_source_backingstore.c b/src/storage_file/storage_source_backingstore.c
+index 80681924ea..8a3ade9ec0 100644
+--- a/src/storage_file/storage_source_backingstore.c
++++ b/src/storage_file/storage_source_backingstore.c
+@@ -287,6 +287,75 @@ virStorageSourceParseRBDColonString(const char *rbdstr,
+ }
+ 
+ 
++static int
++virStorageSourceParseVitastorColonString(const char *colonstr,
++                                         virStorageSource *src)
++{
++    char *p, *e, *next;
++    g_autofree char *options = NULL;
++
++    /* optionally skip the "vitastor:" prefix if provided */
++    if (STRPREFIX(colonstr, "vitastor:"))
++        colonstr += strlen("vitastor:");
++
++    options = g_strdup(colonstr);
++
++    p = options;
++    while (*p) {
++        /* find : delimiter or end of string */
++        for (e = p; *e && *e != ':'; ++e) {
++            if (*e == '\\') {
++                e++;
++                if (*e == '\0')
++                    break;
++            }
++        }
++        if (*e == '\0') {
++            next = e;    /* last kv pair */
++        } else {
++            next = e + 1;
++            *e = '\0';
++        }
++
++        if (STRPREFIX(p, "image=")) {
++            src->path = g_strdup(p + strlen("image="));
++        } else if (STRPREFIX(p, "etcd-prefix=")) {
++            src->query = g_strdup(p + strlen("etcd-prefix="));
++        } else if (STRPREFIX(p, "config-path=")) {
++            src->configFile = g_strdup(p + strlen("config-path="));
++        } else if (STRPREFIX(p, "etcd-host=")) {
++            char *h, *sep;
++
++            h = p + strlen("etcd-host=");
++            while (h < e) {
++                for (sep = h; sep < e; ++sep) {
++                    if (*sep == '\\' && (sep[1] == ',' ||
++                                         sep[1] == ';' ||
++                                         sep[1] == ' ')) {
++                        *sep = '\0';
++                        sep += 2;
++                        break;
++                    }
++                }
++
++                if (virStorageSourceRBDAddHost(src, h) < 0)
++                    return -1;
++
++                h = sep;
++            }
++        }
++
++        p = next;
++    }
++
++    if (!src->path) {
++        return -1;
++    }
++
++    return 0;
++}
++
++
+ static int
+ virStorageSourceParseNBDColonString(const char *nbdstr,
+                                     virStorageSource *src)
+@@ -399,6 +468,11 @@ virStorageSourceParseBackingColon(virStorageSource *src,
+             return -1;
+         break;
+ 
++    case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
++        if (virStorageSourceParseVitastorColonString(path, src) < 0)
++            return -1;
++        break;
++
+     case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+     case VIR_STORAGE_NET_PROTOCOL_LAST:
+     case VIR_STORAGE_NET_PROTOCOL_NONE:
+@@ -975,6 +1049,54 @@ virStorageSourceParseBackingJSONRBD(virStorageSource *src,
+     return 0;
+ }
+ 
++static int
++virStorageSourceParseBackingJSONVitastor(virStorageSource *src,
++                                         virJSONValue *json,
++                                         const char *jsonstr G_GNUC_UNUSED,
++                                         int opaque G_GNUC_UNUSED)
++{
++    const char *filename;
++    const char *image = virJSONValueObjectGetString(json, "image");
++    const char *conf = virJSONValueObjectGetString(json, "config-path");
++    const char *etcd_prefix = virJSONValueObjectGetString(json, "etcd-prefix");
++    virJSONValue *servers = virJSONValueObjectGetArray(json, "server");
++    size_t nservers;
++    size_t i;
++
++    src->type = VIR_STORAGE_TYPE_NETWORK;
++    src->protocol = VIR_STORAGE_NET_PROTOCOL_VITASTOR;
++
++    /* legacy syntax passed via 'filename' option */
++    if ((filename = virJSONValueObjectGetString(json, "filename")))
++        return virStorageSourceParseVitastorColonString(filename, src);
++
++    if (!image) {
++        virReportError(VIR_ERR_INVALID_ARG, "%s",
++                       _("missing image name in Vitastor backing volume "
++                         "JSON specification"));
++        return -1;
++    }
++
++    src->path = g_strdup(image);
++    src->configFile = g_strdup(conf);
++    src->query = g_strdup(etcd_prefix);
++
++    if (servers) {
++        nservers = virJSONValueArraySize(servers);
++
++        src->hosts = g_new0(virStorageNetHostDef, nservers);
++        src->nhosts = nservers;
++
++        for (i = 0; i < nservers; i++) {
++            if (virStorageSourceParseBackingJSONInetSocketAddress(src->hosts + i,
++                                                                  virJSONValueArrayGet(servers, i)) < 0)
++                return -1;
++        }
++    }
++
++    return 0;
++}
++
+ static int
+ virStorageSourceParseBackingJSONRaw(virStorageSource *src,
+                                     virJSONValue *json,
+@@ -1152,6 +1274,7 @@ static const struct virStorageSourceJSONDriverParser jsonParsers[] = {
+     {"sheepdog", false, virStorageSourceParseBackingJSONSheepdog, 0},
+     {"ssh", false, virStorageSourceParseBackingJSONSSH, 0},
+     {"rbd", false, virStorageSourceParseBackingJSONRBD, 0},
++    {"vitastor", false, virStorageSourceParseBackingJSONVitastor, 0},
+     {"raw", true, virStorageSourceParseBackingJSONRaw, 0},
+     {"nfs", false, virStorageSourceParseBackingJSONNFS, 0},
+     {"vxhs", false, virStorageSourceParseBackingJSONVxHS, 0},
+diff --git a/src/test/test_driver.c b/src/test/test_driver.c
+index d2d1bc43e3..31a92e4a01 100644
+--- a/src/test/test_driver.c
++++ b/src/test/test_driver.c
+@@ -7339,6 +7339,7 @@ testStorageVolumeTypeForPool(int pooltype)
+     case VIR_STORAGE_POOL_ISCSI_DIRECT:
+     case VIR_STORAGE_POOL_GLUSTER:
+     case VIR_STORAGE_POOL_RBD:
++    case VIR_STORAGE_POOL_VITASTOR:
+         return VIR_STORAGE_VOL_NETWORK;
+     case VIR_STORAGE_POOL_LOGICAL:
+     case VIR_STORAGE_POOL_DISK:
+diff --git a/tests/storagepoolcapsschemadata/poolcaps-fs.xml b/tests/storagepoolcapsschemadata/poolcaps-fs.xml
+index eee75af746..8bd0a57bdd 100644
+--- a/tests/storagepoolcapsschemadata/poolcaps-fs.xml
++++ b/tests/storagepoolcapsschemadata/poolcaps-fs.xml
+@@ -204,4 +204,11 @@
+       </enum>
+     </volOptions>
+   </pool>
++  <pool type='vitastor' supported='no'>
++    <volOptions>
++      <defaultFormat type='raw'/>
++      <enum name='targetFormatType'>
++      </enum>
++    </volOptions>
++  </pool>
+ </storagepoolCapabilities>
+diff --git a/tests/storagepoolcapsschemadata/poolcaps-full.xml b/tests/storagepoolcapsschemadata/poolcaps-full.xml
+index 805950a937..852df0de16 100644
+--- a/tests/storagepoolcapsschemadata/poolcaps-full.xml
++++ b/tests/storagepoolcapsschemadata/poolcaps-full.xml
+@@ -204,4 +204,11 @@
+       </enum>
+     </volOptions>
+   </pool>
++  <pool type='vitastor' supported='yes'>
++    <volOptions>
++      <defaultFormat type='raw'/>
++      <enum name='targetFormatType'>
++      </enum>
++    </volOptions>
++  </pool>
+ </storagepoolCapabilities>
+diff --git a/tests/storagepoolxml2argvtest.c b/tests/storagepoolxml2argvtest.c
+index e8e40d695e..db55fe5f3a 100644
+--- a/tests/storagepoolxml2argvtest.c
++++ b/tests/storagepoolxml2argvtest.c
+@@ -65,6 +65,7 @@ testCompareXMLToArgvFiles(bool shouldFail,
+     case VIR_STORAGE_POOL_GLUSTER:
+     case VIR_STORAGE_POOL_ZFS:
+     case VIR_STORAGE_POOL_VSTORAGE:
++    case VIR_STORAGE_POOL_VITASTOR:
+     case VIR_STORAGE_POOL_LAST:
+     default:
+         VIR_TEST_DEBUG("pool type '%s' has no xml2argv test", defTypeStr);
+diff --git a/tools/virsh-pool.c b/tools/virsh-pool.c
+index f9aad8ded0..64704b4288 100644
+--- a/tools/virsh-pool.c
++++ b/tools/virsh-pool.c
+@@ -1187,6 +1187,9 @@ cmdPoolList(vshControl *ctl, const vshCmd *cmd G_GNUC_UNUSED)
+             case VIR_STORAGE_POOL_VSTORAGE:
+                 flags |= VIR_CONNECT_LIST_STORAGE_POOLS_VSTORAGE;
+                 break;
++            case VIR_STORAGE_POOL_VITASTOR:
++                flags |= VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR;
++                break;
+             case VIR_STORAGE_POOL_LAST:
+                 break;
+             }

patches/qemu-make-patches.sh

@@ -9,7 +9,7 @@ for i in "$DIR"/qemu-*-vitastor.patch "$DIR"/pve-qemu-*-vitastor.patch; do
         echo '===================================================================' >> $i
         echo '--- /dev/null' >> $i
         echo '+++ a/block/vitastor.c' >> $i
-        echo '@@ -0,0 +1,'$(wc -l "$DIR"/../src/qemu_driver.c | cut -d ' ' -f 1)' @@' >> $i
-        cat "$DIR"/../src/qemu_driver.c | sed 's/^/+/' >> $i
+        echo '@@ -0,0 +1,'$(wc -l "$DIR"/../src/client/qemu_driver.c | cut -d ' ' -f 1)' @@' >> $i
+        cat "$DIR"/../src/client/qemu_driver.c | sed 's/^/+/' >> $i
     fi
 done

rpm/qemu-el8.Dockerfile

@@ -1,5 +1,5 @@
 # This is an attempt to automatically build patched RPM specs
-# More or less broken, better use *.spec.patch for now (and copy src/qemu_driver.c to SOURCES/qemu-vitastor.c)
+# More or less broken, better use *.spec.patch for now (and copy src/client/qemu_driver.c to SOURCES/qemu-vitastor.c)
 
 # Build packages for CentOS 8 inside a container
 # cd ..; podman build -t qemu-el8 -v `pwd`/packages:/root/packages -f rpm/qemu-el8.Dockerfile .

rpm/vitastor-el7.spec

@@ -108,10 +108,11 @@ npm install --production
 cd ..
 mkdir -p %buildroot/usr/lib/vitastor
 cp -r mon %buildroot/usr/lib/vitastor
+mv %buildroot/usr/lib/vitastor/mon/scripts/make-etcd %buildroot/usr/lib/vitastor/mon/
 mkdir -p %buildroot/lib/systemd/system
-cp mon/vitastor.target mon/vitastor-mon.service mon/vitastor-osd@.service %buildroot/lib/systemd/system
+cp mon/scripts/vitastor.target mon/scripts/vitastor-mon.service mon/scripts/vitastor-osd@.service %buildroot/lib/systemd/system
 mkdir -p %buildroot/lib/udev/rules.d
-cp mon/90-vitastor.rules %buildroot/lib/udev/rules.d
+cp mon/scripts/90-vitastor.rules %buildroot/lib/udev/rules.d
 
 
 %files

rpm/vitastor-el8.spec

@@ -105,10 +105,11 @@ npm install --production
 cd ..
 mkdir -p %buildroot/usr/lib/vitastor
 cp -r mon %buildroot/usr/lib/vitastor
+mv %buildroot/usr/lib/vitastor/mon/scripts/make-etcd %buildroot/usr/lib/vitastor/mon/
 mkdir -p %buildroot/lib/systemd/system
-cp mon/vitastor.target mon/vitastor-mon.service mon/vitastor-osd@.service %buildroot/lib/systemd/system
+cp mon/scripts/vitastor.target mon/scripts/vitastor-mon.service mon/scripts/vitastor-osd@.service %buildroot/lib/systemd/system
 mkdir -p %buildroot/lib/udev/rules.d
-cp mon/90-vitastor.rules %buildroot/lib/udev/rules.d
+cp mon/scripts/90-vitastor.rules %buildroot/lib/udev/rules.d
 
 
 %files

rpm/vitastor-el9.spec

@@ -98,10 +98,11 @@ npm install --production
 cd ..
 mkdir -p %buildroot/usr/lib/vitastor
 cp -r mon %buildroot/usr/lib/vitastor
+mv %buildroot/usr/lib/vitastor/mon/scripts/make-etcd %buildroot/usr/lib/vitastor/mon/
 mkdir -p %buildroot/lib/systemd/system
-cp mon/vitastor.target mon/vitastor-mon.service mon/vitastor-osd@.service %buildroot/lib/systemd/system
+cp mon/scripts/vitastor.target mon/scripts/vitastor-mon.service mon/scripts/vitastor-osd@.service %buildroot/lib/systemd/system
 mkdir -p %buildroot/lib/udev/rules.d
-cp mon/90-vitastor.rules %buildroot/lib/udev/rules.d
+cp mon/scripts/90-vitastor.rules %buildroot/lib/udev/rules.d
 
 
 %files

src/CMakeLists.txt

@@ -72,300 +72,28 @@ add_dependencies(test build_tests)
 
 include_directories(
 	../
+	${CMAKE_SOURCE_DIR}/src/blockstore
+	${CMAKE_SOURCE_DIR}/src/cmd
+	${CMAKE_SOURCE_DIR}/src/client
+	${CMAKE_SOURCE_DIR}/src/disk_tool
+	${CMAKE_SOURCE_DIR}/src/kv
+	${CMAKE_SOURCE_DIR}/src/nfs
+	${CMAKE_SOURCE_DIR}/src/osd
+	${CMAKE_SOURCE_DIR}/src/test
+	${CMAKE_SOURCE_DIR}/src/util
 	/usr/include/jerasure
 	${LIBURING_INCLUDE_DIRS}
 	${IBVERBS_INCLUDE_DIRS}
 )
 
-# libvitastor_blk.so
-add_library(vitastor_blk SHARED
-	allocator.cpp blockstore.cpp blockstore_impl.cpp blockstore_disk.cpp blockstore_init.cpp blockstore_open.cpp blockstore_journal.cpp blockstore_read.cpp
-	blockstore_write.cpp blockstore_sync.cpp blockstore_stable.cpp blockstore_rollback.cpp blockstore_flush.cpp crc32c.c ringloop.cpp
-)
-target_link_libraries(vitastor_blk
-	${LIBURING_LIBRARIES}
-	tcmalloc_minimal
-	# for timerfd_manager
-	vitastor_common
-)
-set_target_properties(vitastor_blk PROPERTIES VERSION ${VERSION} SOVERSION 0)
-
-if (${WITH_FIO})
-	# libfio_vitastor_blk.so
-	add_library(fio_vitastor_blk SHARED
-		fio_engine.cpp
-		../json11/json11.cpp
-	)
-	target_link_libraries(fio_vitastor_blk
-		vitastor_blk
-	)
-endif (${WITH_FIO})
-
-# libvitastor_common.a
-set(MSGR_RDMA "")
-if (IBVERBS_LIBRARIES)
-	set(MSGR_RDMA "msgr_rdma.cpp")
-endif (IBVERBS_LIBRARIES)
-add_library(vitastor_common STATIC
-	epoll_manager.cpp etcd_state_client.cpp messenger.cpp addr_util.cpp
-	msgr_stop.cpp msgr_op.cpp msgr_send.cpp msgr_receive.cpp ringloop.cpp ../json11/json11.cpp
-	http_client.cpp osd_ops.cpp pg_states.cpp timerfd_manager.cpp str_util.cpp ${MSGR_RDMA}
-)
-target_compile_options(vitastor_common PUBLIC -fPIC)
-
-# vitastor-osd
-add_executable(vitastor-osd
-	osd_main.cpp osd.cpp osd_secondary.cpp osd_peering.cpp osd_flush.cpp osd_peering_pg.cpp
-	osd_primary.cpp osd_primary_chain.cpp osd_primary_sync.cpp osd_primary_write.cpp osd_primary_subops.cpp
-	osd_cluster.cpp osd_rmw.cpp osd_scrub.cpp osd_primary_describe.cpp
-)
-target_link_libraries(vitastor-osd
-	vitastor_common
-	vitastor_blk
-	Jerasure
-	${ISAL_LIBRARIES}
-	${IBVERBS_LIBRARIES}
-)
-
-if (${WITH_FIO})
-	# libfio_vitastor_sec.so
-	add_library(fio_vitastor_sec SHARED
-		fio_sec_osd.cpp
-		rw_blocking.cpp
-		addr_util.cpp
-	)
-	target_link_libraries(fio_vitastor_sec
-		tcmalloc_minimal
-	)
-endif (${WITH_FIO})
-
-# libvitastor_client.so
-add_library(vitastor_client SHARED
-	cluster_client.cpp
-	cluster_client_list.cpp
-	cluster_client_wb.cpp
-	vitastor_c.cpp
-	cli_common.cpp
-	cli_alloc_osd.cpp
-	cli_status.cpp
-	cli_describe.cpp
-	cli_fix.cpp
-	cli_ls.cpp
-	cli_create.cpp
-	cli_modify.cpp
-	cli_flatten.cpp
-	cli_merge.cpp
-	cli_rm_data.cpp
-	cli_rm.cpp
-	cli_rm_osd.cpp
-	cli_pool_cfg.cpp
-	cli_pool_create.cpp
-	cli_pool_ls.cpp
-	cli_pool_modify.cpp
-	cli_pool_rm.cpp
-)
-set_target_properties(vitastor_client PROPERTIES PUBLIC_HEADER "vitastor_c.h")
-target_link_libraries(vitastor_client
-	vitastor_common
-	${LIBURING_LIBRARIES}
-	${IBVERBS_LIBRARIES}
-)
-set_target_properties(vitastor_client PROPERTIES VERSION ${VERSION} SOVERSION 0)
-
-if (${WITH_FIO})
-	# libfio_vitastor.so
-	add_library(fio_vitastor SHARED
-		fio_cluster.cpp
-	)
-	target_link_libraries(fio_vitastor
-		vitastor_client
-	)
-endif (${WITH_FIO})
-
-# vitastor-nbd
-pkg_check_modules(NL3 libnl-3.0 libnl-genl-3.0)
-add_executable(vitastor-nbd
-	nbd_proxy.cpp
-)
-target_include_directories(vitastor-nbd PUBLIC ${NL3_INCLUDE_DIRS})
-target_link_libraries(vitastor-nbd vitastor_client ${NL3_LIBRARIES})
-if (HAVE_NBD_NETLINK_H AND NL3_LIBRARIES)
-	target_compile_definitions(vitastor-nbd PUBLIC HAVE_NBD_NETLINK_H)
-endif (HAVE_NBD_NETLINK_H AND NL3_LIBRARIES)
-
-# libvitastor_kv.so
-add_library(vitastor_kv SHARED
-	kv_db.cpp
-	kv_db.h
-)
-target_link_libraries(vitastor_kv
-	vitastor_client
-)
-set_target_properties(vitastor_kv PROPERTIES VERSION ${VERSION} SOVERSION 0)
-
-# vitastor-kv
-add_executable(vitastor-kv
-	kv_cli.cpp
-)
-target_link_libraries(vitastor-kv
-	vitastor_kv
-)
-
-add_executable(vitastor-kv-stress
-	kv_stress.cpp
-)
-target_link_libraries(vitastor-kv-stress
-	vitastor_kv
-)
-
-# vitastor-nfs
-add_executable(vitastor-nfs
-	nfs_proxy.cpp
-	nfs_block.cpp
-	nfs_kv.cpp
-	nfs_kv_create.cpp
-	nfs_kv_getattr.cpp
-	nfs_kv_link.cpp
-	nfs_kv_lookup.cpp
-	nfs_kv_read.cpp
-	nfs_kv_readdir.cpp
-	nfs_kv_remove.cpp
-	nfs_kv_rename.cpp
-	nfs_kv_setattr.cpp
-	nfs_kv_write.cpp
-	nfs_fsstat.cpp
-	nfs_mount.cpp
-	nfs_portmap.cpp
-	sha256.c
-	nfs/xdr_impl.cpp
-	nfs/rpc_xdr.cpp
-	nfs/portmap_xdr.cpp
-	nfs/nfs_xdr.cpp
-)
-target_link_libraries(vitastor-nfs
-	vitastor_client
-	vitastor_kv
-)
-
-# vitastor-cli
-add_executable(vitastor-cli
-	cli.cpp
-)
-target_link_libraries(vitastor-cli
-	vitastor_client
-)
-configure_file(vitastor.pc.in vitastor.pc @ONLY)
-
-# vitastor-disk
-add_executable(vitastor-disk
-	disk_tool.cpp disk_simple_offsets.cpp
-	disk_tool_journal.cpp disk_tool_meta.cpp disk_tool_prepare.cpp disk_tool_resize.cpp disk_tool_udev.cpp disk_tool_utils.cpp disk_tool_upgrade.cpp
-	crc32c.c str_util.cpp ../json11/json11.cpp rw_blocking.cpp allocator.cpp ringloop.cpp blockstore_disk.cpp
-)
-target_link_libraries(vitastor-disk
-	tcmalloc_minimal
-	${LIBURING_LIBRARIES}
-)
-
-if (${WITH_QEMU})
-	# qemu_driver.so
-	add_library(qemu_vitastor SHARED
-		qemu_driver.c
-	)
-	target_compile_options(qemu_vitastor PUBLIC -DVITASTOR_SOURCE_TREE)
-	target_include_directories(qemu_vitastor PUBLIC
-		../qemu/b/qemu
-		../qemu/include
-		${GLIB_INCLUDE_DIRS}
-	)
-	target_link_libraries(qemu_vitastor
-		vitastor_client
-	)
-	set_target_properties(qemu_vitastor PROPERTIES
-		PREFIX ""
-		OUTPUT_NAME "block-vitastor"
-	)
-endif (${WITH_QEMU})
-
-### Test stubs
-
-# stub_osd, stub_bench, osd_test
-add_executable(stub_osd stub_osd.cpp rw_blocking.cpp addr_util.cpp)
-target_link_libraries(stub_osd tcmalloc_minimal)
-add_executable(stub_bench stub_bench.cpp rw_blocking.cpp addr_util.cpp)
-target_link_libraries(stub_bench tcmalloc_minimal)
-add_executable(osd_test osd_test.cpp rw_blocking.cpp addr_util.cpp)
-target_link_libraries(osd_test tcmalloc_minimal)
-
-# osd_rmw_test
-add_executable(osd_rmw_test EXCLUDE_FROM_ALL osd_rmw_test.cpp allocator.cpp)
-target_link_libraries(osd_rmw_test Jerasure ${ISAL_LIBRARIES} tcmalloc_minimal)
-add_dependencies(build_tests osd_rmw_test)
-add_test(NAME osd_rmw_test COMMAND osd_rmw_test)
-
-if (ISAL_LIBRARIES)
-	add_executable(osd_rmw_test_je EXCLUDE_FROM_ALL osd_rmw_test.cpp allocator.cpp)
-	target_compile_definitions(osd_rmw_test_je PUBLIC -DNO_ISAL)
-	target_link_libraries(osd_rmw_test_je Jerasure tcmalloc_minimal)
-	add_dependencies(build_tests osd_rmw_test_je)
-	add_test(NAME osd_rmw_test_jerasure COMMAND osd_rmw_test_je)
-endif (ISAL_LIBRARIES)
-
-# stub_uring_osd
-add_executable(stub_uring_osd
-	stub_uring_osd.cpp
-)
-target_link_libraries(stub_uring_osd
-	vitastor_common
-	${LIBURING_LIBRARIES}
-	${IBVERBS_LIBRARIES}
-	tcmalloc_minimal
-)
-
-# osd_peering_pg_test
-add_executable(osd_peering_pg_test EXCLUDE_FROM_ALL osd_peering_pg_test.cpp osd_peering_pg.cpp)
-target_link_libraries(osd_peering_pg_test tcmalloc_minimal)
-add_dependencies(build_tests osd_peering_pg_test)
-add_test(NAME osd_peering_pg_test COMMAND osd_peering_pg_test)
-
-# test_allocator
-add_executable(test_allocator EXCLUDE_FROM_ALL test_allocator.cpp allocator.cpp)
-add_dependencies(build_tests test_allocator)
-add_test(NAME test_allocator COMMAND test_allocator)
-
-# test_cas
-add_executable(test_cas
-	test_cas.cpp
-)
-target_link_libraries(test_cas
-	vitastor_client
-)
-
-# test_crc32
-add_executable(test_crc32
-	test_crc32.cpp
-)
-target_link_libraries(test_crc32
-	vitastor_blk
-)
-
-# test_cluster_client
-add_executable(test_cluster_client
-	EXCLUDE_FROM_ALL
-	test_cluster_client.cpp
-	pg_states.cpp osd_ops.cpp cluster_client.cpp cluster_client_list.cpp cluster_client_wb.cpp msgr_op.cpp mock/messenger.cpp msgr_stop.cpp
-	etcd_state_client.cpp timerfd_manager.cpp str_util.cpp ../json11/json11.cpp
-)
-target_compile_definitions(test_cluster_client PUBLIC -D__MOCK__)
-target_include_directories(test_cluster_client PUBLIC ${CMAKE_SOURCE_DIR}/src/mock)
-add_dependencies(build_tests test_cluster_client)
-add_test(NAME test_cluster_client COMMAND test_cluster_client)
-
-## test_blockstore, test_shit
-#add_executable(test_blockstore test_blockstore.cpp)
-#target_link_libraries(test_blockstore blockstore)
-#add_executable(test_shit test_shit.cpp osd_peering_pg.cpp)
-#target_link_libraries(test_shit ${LIBURING_LIBRARIES} m)
+add_subdirectory(blockstore)
+add_subdirectory(cmd)
+add_subdirectory(client)
+add_subdirectory(disk_tool)
+add_subdirectory(kv)
+add_subdirectory(nfs)
+add_subdirectory(osd)
+add_subdirectory(test)
 
 ### Install
 
@@ -378,7 +106,7 @@ install(
 	LIBRARY DESTINATION ${CMAKE_INSTALL_LIBDIR}
 	PUBLIC_HEADER DESTINATION ${CMAKE_INSTALL_INCLUDEDIR}
 )
-install(FILES ${CMAKE_CURRENT_BINARY_DIR}/vitastor.pc DESTINATION ${CMAKE_INSTALL_LIBDIR}/pkgconfig)
+install(FILES ${CMAKE_CURRENT_BINARY_DIR}/client/vitastor.pc DESTINATION ${CMAKE_INSTALL_LIBDIR}/pkgconfig)
 if (${WITH_FIO})
 	install(TARGETS fio_vitastor fio_vitastor_blk fio_vitastor_sec LIBRARY DESTINATION ${CMAKE_INSTALL_LIBDIR})
 endif (${WITH_FIO})

src/blockstore/CMakeLists.txt

@@ -0,0 +1,27 @@
+cmake_minimum_required(VERSION 2.8.12)
+
+project(vitastor)
+
+# libvitastor_blk.so
+add_library(vitastor_blk SHARED
+	../util/allocator.cpp blockstore.cpp blockstore_impl.cpp blockstore_disk.cpp blockstore_init.cpp blockstore_open.cpp blockstore_journal.cpp blockstore_read.cpp
+	blockstore_write.cpp blockstore_sync.cpp blockstore_stable.cpp blockstore_rollback.cpp blockstore_flush.cpp ../util/crc32c.c ../util/ringloop.cpp
+)
+target_link_libraries(vitastor_blk
+	${LIBURING_LIBRARIES}
+	tcmalloc_minimal
+	# for timerfd_manager
+	vitastor_common
+)
+set_target_properties(vitastor_blk PROPERTIES VERSION ${VERSION} SOVERSION 0)
+
+if (${WITH_FIO})
+	# libfio_vitastor_blk.so
+	add_library(fio_vitastor_blk SHARED
+		fio_engine.cpp
+		../../json11/json11.cpp
+	)
+	target_link_libraries(fio_vitastor_blk
+		vitastor_blk
+	)
+endif (${WITH_FIO})

src/blockstore/blockstore_flush.cpp

@@ -366,6 +366,7 @@ resume_0:
         !flusher->flush_queue.size() || !flusher->dequeuing)
     {
 stop_flusher:
+        flusher->dequeuing = false;
         if (flusher->trim_wanted > 0 && try_trim)
         {
             // Attempt forced trim
@@ -373,7 +374,6 @@ stop_flusher:
             flusher->active_flushers++;
             goto trim_journal;
         }
-        flusher->dequeuing = false;
         wait_state = 0;
         return true;
     }

src/client/CMakeLists.txt

@@ -0,0 +1,95 @@
+cmake_minimum_required(VERSION 2.8.12)
+
+project(vitastor)
+
+# libvitastor_common.a
+set(MSGR_RDMA "")
+if (IBVERBS_LIBRARIES)
+	set(MSGR_RDMA "msgr_rdma.cpp")
+endif (IBVERBS_LIBRARIES)
+add_library(vitastor_common STATIC
+	../util/epoll_manager.cpp etcd_state_client.cpp messenger.cpp ../util/addr_util.cpp
+	msgr_stop.cpp msgr_op.cpp msgr_send.cpp msgr_receive.cpp ../util/ringloop.cpp ../../json11/json11.cpp
+	http_client.cpp osd_ops.cpp pg_states.cpp ../util/timerfd_manager.cpp ../util/str_util.cpp ${MSGR_RDMA}
+)
+target_compile_options(vitastor_common PUBLIC -fPIC)
+
+# libvitastor_client.so
+add_library(vitastor_client SHARED
+	cluster_client.cpp
+	cluster_client_list.cpp
+	cluster_client_wb.cpp
+	vitastor_c.cpp
+)
+set_target_properties(vitastor_client PROPERTIES PUBLIC_HEADER "client/vitastor_c.h")
+target_link_libraries(vitastor_client
+	vitastor_common
+	vitastor_cli
+	${LIBURING_LIBRARIES}
+	${IBVERBS_LIBRARIES}
+)
+set_target_properties(vitastor_client PROPERTIES VERSION ${VERSION} SOVERSION 0)
+configure_file(vitastor.pc.in vitastor.pc @ONLY)
+
+if (${WITH_FIO})
+	# libfio_vitastor.so
+	add_library(fio_vitastor SHARED
+		fio_cluster.cpp
+	)
+	target_link_libraries(fio_vitastor
+		vitastor_client
+	)
+
+	# libfio_vitastor_sec.so
+	add_library(fio_vitastor_sec SHARED
+		fio_sec_osd.cpp
+		../util/rw_blocking.cpp
+		../util/addr_util.cpp
+	)
+	target_link_libraries(fio_vitastor_sec
+		tcmalloc_minimal
+	)
+endif (${WITH_FIO})
+
+# vitastor-nbd
+pkg_check_modules(NL3 libnl-3.0 libnl-genl-3.0)
+add_executable(vitastor-nbd
+	nbd_proxy.cpp
+)
+target_include_directories(vitastor-nbd PUBLIC ${NL3_INCLUDE_DIRS})
+target_link_libraries(vitastor-nbd vitastor_client ${NL3_LIBRARIES})
+if (HAVE_NBD_NETLINK_H AND NL3_LIBRARIES)
+	target_compile_definitions(vitastor-nbd PUBLIC HAVE_NBD_NETLINK_H)
+endif (HAVE_NBD_NETLINK_H AND NL3_LIBRARIES)
+
+if (${WITH_QEMU})
+	# qemu_driver.so
+	add_library(qemu_vitastor SHARED
+		qemu_driver.c
+	)
+	target_compile_options(qemu_vitastor PUBLIC -DVITASTOR_SOURCE_TREE)
+	target_include_directories(qemu_vitastor PUBLIC
+		../../qemu/b/qemu
+		../../qemu/include
+		${GLIB_INCLUDE_DIRS}
+	)
+	target_link_libraries(qemu_vitastor
+		vitastor_client
+	)
+	set_target_properties(qemu_vitastor PROPERTIES
+		PREFIX ""
+		OUTPUT_NAME "block-vitastor"
+	)
+endif (${WITH_QEMU})
+
+# test_cluster_client
+add_executable(test_cluster_client
+	EXCLUDE_FROM_ALL
+	../test/test_cluster_client.cpp
+	pg_states.cpp osd_ops.cpp cluster_client.cpp cluster_client_list.cpp cluster_client_wb.cpp msgr_op.cpp ../test/mock/messenger.cpp msgr_stop.cpp
+	etcd_state_client.cpp ../util/timerfd_manager.cpp ../util/str_util.cpp ../../json11/json11.cpp
+)
+target_compile_definitions(test_cluster_client PUBLIC -D__MOCK__)
+target_include_directories(test_cluster_client BEFORE PUBLIC ${CMAKE_SOURCE_DIR}/src/test/mock)
+add_dependencies(build_tests test_cluster_client)
+add_test(NAME test_cluster_client COMMAND test_cluster_client)

src/client/cluster_client.cpp

@@ -34,7 +34,7 @@ cluster_client_t::cluster_client_t(ring_loop_t *ringloop, timerfd_manager_t *tfd
         {
             // peer_osd just dropped connection
             // determine WHICH dirty_buffers are now obsolete and repeat them
-            if (wb->repeat_ops_for(this, peer_osd) > 0)
+            if (wb->repeat_ops_for(this, peer_osd, 0, 0) > 0)
             {
                 continue_ops();
             }
@@ -52,7 +52,8 @@ cluster_client_t::cluster_client_t(ring_loop_t *ringloop, timerfd_manager_t *tfd
     st_cli.tfd = tfd;
     st_cli.on_load_config_hook = [this](json11::Json::object & cfg) { on_load_config_hook(cfg); };
     st_cli.on_change_osd_state_hook = [this](uint64_t peer_osd) { on_change_osd_state_hook(peer_osd); };
-    st_cli.on_change_hook = [this](std::map<std::string, etcd_kv_t> & changes) { on_change_hook(changes); };
+    st_cli.on_change_pool_config_hook = [this]() { on_change_pool_config_hook(); };
+    st_cli.on_change_pg_state_hook = [this](pool_id_t pool_id, pg_num_t pg_num, osd_num_t prev_primary) { on_change_pg_state_hook(pool_id, pg_num, prev_primary); };
     st_cli.on_load_pgs_hook = [this](bool success) { on_load_pgs_hook(success); };
     st_cli.on_reload_hook = [this]() { st_cli.load_global_config(); };
 
@@ -77,11 +78,6 @@ cluster_client_t::~cluster_client_t()
 
 cluster_op_t::~cluster_op_t()
 {
-    if (buf)
-    {
-        free(buf);
-        buf = NULL;
-    }
     if (bitmap_buf)
     {
         free(bitmap_buf);
@@ -427,7 +423,7 @@ void cluster_client_t::on_load_pgs_hook(bool success)
     continue_ops();
 }
 
-void cluster_client_t::on_change_hook(std::map<std::string, etcd_kv_t> & changes)
+void cluster_client_t::on_change_pool_config_hook()
 {
     for (auto pool_item: st_cli.pool_config)
     {
@@ -450,6 +446,19 @@ void cluster_client_t::on_change_hook(std::map<std::string, etcd_kv_t> & changes
     continue_ops();
 }
 
+void cluster_client_t::on_change_pg_state_hook(pool_id_t pool_id, pg_num_t pg_num, osd_num_t prev_primary)
+{
+    auto & pg_cfg = st_cli.pool_config[pool_id].pg_config[pg_num];
+    if (pg_cfg.cur_primary != prev_primary)
+    {
+        // Repeat this PG operations because an OSD which stopped being primary may not fsync operations
+        if (wb->repeat_ops_for(this, 0, pool_id, pg_num) > 0)
+        {
+            continue_ops();
+        }
+    }
+}
+
 bool cluster_client_t::get_immediate_commit(uint64_t inode)
 {
     if (enable_writeback)
@@ -570,6 +579,14 @@ void cluster_client_t::execute_internal(cluster_op_t *op)
 {
     op->cur_inode = op->inode;
     op->retval = 0;
+    op->state = 0;
+    op->retry_after = 0;
+    op->inflight_count = 0;
+    op->done_count = 0;
+    op->part_bitmaps = NULL;
+    op->bitmap_buf_size = 0;
+    op->prev_wait = 0;
+    assert(!op->prev && !op->next);
     // check alignment, readonly flag and so on
     if (!check_rw(op))
     {
@@ -600,7 +617,9 @@ void cluster_client_t::execute_internal(cluster_op_t *op)
     {
         if (!(op->flags & OP_FLUSH_BUFFER) && !op->version /* no CAS write-repeat */)
         {
-            wb->copy_write(op, CACHE_WRITTEN);
+            uint64_t flush_id = ++wb->last_flush_id;
+            wb->copy_write(op, CACHE_REPEATING, flush_id);
+            op->flush_id = flush_id;
         }
         if (dirty_bytes >= client_max_dirty_bytes || dirty_ops >= client_max_dirty_ops)
         {
@@ -816,6 +835,10 @@ resume_2:
             auto & pool_cfg = st_cli.pool_config.at(INODE_POOL(op->inode));
             op->retval = op->len / pool_cfg.bitmap_granularity;
         }
+        if (op->flush_id)
+        {
+            wb->mark_flush_written(op->inode, op->offset, op->len, op->flush_id);
+        }
         erase_op(op);
         return 1;
     }
@@ -988,6 +1011,29 @@ void cluster_client_t::slice_rw(cluster_op_t *op)
     }
 }
 
+bool cluster_client_t::affects_pg(uint64_t inode, uint64_t offset, uint64_t len, pool_id_t pool_id, pg_num_t pg_num)
+{
+    if (INODE_POOL(inode) != pool_id)
+    {
+        return false;
+    }
+    auto & pool_cfg = st_cli.pool_config.at(INODE_POOL(inode));
+    uint32_t pg_data_size = (pool_cfg.scheme == POOL_SCHEME_REPLICATED ? 1 : pool_cfg.pg_size-pool_cfg.parity_chunks);
+    uint64_t pg_block_size = pool_cfg.data_block_size * pg_data_size;
+    uint64_t first_stripe = (offset / pg_block_size) * pg_block_size;
+    uint64_t last_stripe = len > 0 ? ((offset + len - 1) / pg_block_size) * pg_block_size : first_stripe;
+    if ((last_stripe/pool_cfg.pg_stripe_size) - (first_stripe/pool_cfg.pg_stripe_size) + 1 >= pool_cfg.real_pg_count)
+    {
+        // All PGs are affected
+        return true;
+    }
+    pg_num_t first_pg_num = (first_stripe/pool_cfg.pg_stripe_size) % pool_cfg.real_pg_count + 1; // like map_to_pg()
+    pg_num_t last_pg_num = (last_stripe/pool_cfg.pg_stripe_size) % pool_cfg.real_pg_count + 1; // like map_to_pg()
+    return (first_pg_num <= last_pg_num
+        ? (pg_num >= first_pg_num && pg_num <= last_pg_num)
+        : (pg_num >= first_pg_num || pg_num <= last_pg_num));
+}
+
 bool cluster_client_t::affects_osd(uint64_t inode, uint64_t offset, uint64_t len, osd_num_t osd)
 {
     auto & pool_cfg = st_cli.pool_config.at(INODE_POOL(inode));
@@ -1210,7 +1256,9 @@ void cluster_client_t::handle_op_part(cluster_op_part_t *part)
         // So do all these things after modifying operation state, otherwise we may hit reenterability bugs
         // FIXME postpone such things to set_immediate here to avoid bugs
         // Set op->retry_after to retry operation after a short pause (not immediately)
-        if (!op->retry_after)
+        if (!op->retry_after && (op->retval == -EPIPE ||
+            op->retval == -EIO && client_eio_retry_interval ||
+            op->retval == -ENOSPC && client_retry_enospc))
         {
             op->retry_after = op->retval != -EPIPE ? client_eio_retry_interval : client_retry_interval;
         }

src/client/cluster_client.h

@@ -56,8 +56,6 @@ struct cluster_op_t
 protected:
     int state = 0;
     uint64_t cur_inode; // for snapshot reads
-    void *buf = NULL;
-    cluster_op_t *orig_op = NULL;
     bool needs_reslice = false;
     int retry_after = 0;
     int inflight_count = 0, done_count = 0;
@@ -66,6 +64,7 @@ protected:
     unsigned bitmap_buf_size = 0;
     cluster_op_t *prev = NULL, *next = NULL;
     int prev_wait = 0;
+    uint64_t flush_id = 0;
     friend class cluster_client_t;
     friend class writeback_cache_t;
 };
@@ -81,6 +80,7 @@ class cluster_client_t
     ring_loop_t *ringloop;
 
     std::map<pool_id_t, uint64_t> pg_counts;
+    std::map<pool_pg_num_t, osd_num_t> pg_primary;
     // client_max_dirty_* is actually "max unsynced", for the case when immediate_commit is off
     uint64_t client_max_dirty_bytes = 0;
     uint64_t client_max_dirty_ops = 0;
@@ -146,9 +146,11 @@ public:
 
 protected:
     bool affects_osd(uint64_t inode, uint64_t offset, uint64_t len, osd_num_t osd);
+    bool affects_pg(uint64_t inode, uint64_t offset, uint64_t len, pool_id_t pool_id, pg_num_t pg_num);
     void on_load_config_hook(json11::Json::object & config);
     void on_load_pgs_hook(bool success);
-    void on_change_hook(std::map<std::string, etcd_kv_t> & changes);
+    void on_change_pool_config_hook();
+    void on_change_pg_state_hook(pool_id_t pool_id, pg_num_t pg_num, osd_num_t prev_primary);
     void on_change_osd_state_hook(uint64_t peer_osd);
     void execute_internal(cluster_op_t *op);
     void unshift_op(cluster_op_t *op);

src/client/cluster_client_impl.h

@@ -46,11 +46,12 @@ public:
     bool is_left_merged(dirty_buf_it_t dirty_it);
     bool is_right_merged(dirty_buf_it_t dirty_it);
     bool is_merged(const dirty_buf_it_t & dirty_it);
-    void copy_write(cluster_op_t *op, int state);
-    int repeat_ops_for(cluster_client_t *cli, osd_num_t peer_osd);
+    void copy_write(cluster_op_t *op, int state, uint64_t new_flush_id = 0);
+    int repeat_ops_for(cluster_client_t *cli, osd_num_t peer_osd, pool_id_t pool_id, pg_num_t pg_num);
     void start_writebacks(cluster_client_t *cli, int count);
     bool read_from_cache(cluster_op_t *op, uint32_t bitmap_granularity);
     void flush_buffers(cluster_client_t *cli, dirty_buf_it_t from_it, dirty_buf_it_t to_it);
+    void mark_flush_written(uint64_t inode, uint64_t offset, uint64_t len, uint64_t flush_id);
     void fsync_start();
     void fsync_error();
     void fsync_ok();

src/client/cluster_client_wb.cpp

@@ -71,7 +71,7 @@ bool writeback_cache_t::is_merged(const dirty_buf_it_t & dirty_it)
     return is_left_merged(dirty_it) || is_right_merged(dirty_it);
 }
 
-void writeback_cache_t::copy_write(cluster_op_t *op, int state)
+void writeback_cache_t::copy_write(cluster_op_t *op, int state, uint64_t new_flush_id)
 {
     // Save operation for replay when one of PGs goes out of sync
     // (primary OSD drops our connection in this case)
@@ -180,6 +180,7 @@ void writeback_cache_t::copy_write(cluster_op_t *op, int state)
         .buf = buf,
         .len = op->len,
         .state = state,
+        .flush_id = new_flush_id,
         .refcnt = refcnt,
     });
     if (state == CACHE_DIRTY)
@@ -208,7 +209,7 @@ void writeback_cache_t::copy_write(cluster_op_t *op, int state)
     }
 }
 
-int writeback_cache_t::repeat_ops_for(cluster_client_t *cli, osd_num_t peer_osd)
+int writeback_cache_t::repeat_ops_for(cluster_client_t *cli, osd_num_t peer_osd, pool_id_t pool_id, pg_num_t pg_num)
 {
     int repeated = 0;
     if (dirty_buffers.size())
@@ -218,8 +219,11 @@ int writeback_cache_t::repeat_ops_for(cluster_client_t *cli, osd_num_t peer_osd)
         for (auto wr_it = dirty_buffers.begin(), flush_it = wr_it, last_it = wr_it; ; )
         {
             bool end = wr_it == dirty_buffers.end();
-            bool flush_this = !end && wr_it->second.state != CACHE_REPEATING &&
-                cli->affects_osd(wr_it->first.inode, wr_it->first.stripe, wr_it->second.len, peer_osd);
+            bool flush_this = !end && wr_it->second.state != CACHE_REPEATING;
+            if (peer_osd)
+                flush_this = flush_this && cli->affects_osd(wr_it->first.inode, wr_it->first.stripe, wr_it->second.len, peer_osd);
+            if (pool_id && pg_num)
+                flush_this = flush_this && cli->affects_pg(wr_it->first.inode, wr_it->first.stripe, wr_it->second.len, pool_id, pg_num);
             if (flush_it != wr_it && (end || !flush_this ||
                 wr_it->first.inode != flush_it->first.inode ||
                 wr_it->first.stripe != last_it->first.stripe+last_it->second.len))
@@ -265,7 +269,7 @@ void writeback_cache_t::flush_buffers(cluster_client_t *cli, dirty_buf_it_t from
     writebacks_active++;
     op->callback = [this, flush_id](cluster_op_t* op)
     {
-        // Buffer flushes should be always retried, regardless of the error,
+        // Buffer flushes are always retried, regardless of the error,
         // so they should never result in an error here
         assert(op->retval == op->len);
         for (auto fl_it = flushed_buffers.find(flush_id);
@@ -277,16 +281,7 @@ void writeback_cache_t::flush_buffers(cluster_client_t *cli, dirty_buf_it_t from
             }
             flushed_buffers.erase(fl_it++);
         }
-        for (auto dirty_it = find_dirty(op->inode, op->offset);
-            dirty_it != dirty_buffers.end() && dirty_it->first.inode == op->inode &&
-            dirty_it->first.stripe < op->offset+op->len; dirty_it++)
-        {
-            if (dirty_it->second.flush_id == flush_id && dirty_it->second.state == CACHE_REPEATING)
-            {
-                dirty_it->second.flush_id = 0;
-                dirty_it->second.state = CACHE_WRITTEN;
-            }
-        }
+        mark_flush_written(op->inode, op->offset, op->len, flush_id);
         delete op;
         writebacks_active--;
         // We can't call execute_internal because it affects an invalid copy of the list here
@@ -304,6 +299,20 @@ void writeback_cache_t::flush_buffers(cluster_client_t *cli, dirty_buf_it_t from
     }
 }
 
+void writeback_cache_t::mark_flush_written(uint64_t inode, uint64_t offset, uint64_t len, uint64_t flush_id)
+{
+    for (auto dirty_it = find_dirty(inode, offset);
+        dirty_it != dirty_buffers.end() && dirty_it->first.inode == inode &&
+        dirty_it->first.stripe < offset+len; dirty_it++)
+    {
+        if (dirty_it->second.flush_id == flush_id && dirty_it->second.state == CACHE_REPEATING)
+        {
+            dirty_it->second.flush_id = 0;
+            dirty_it->second.state = CACHE_WRITTEN;
+        }
+    }
+}
+
 void writeback_cache_t::start_writebacks(cluster_client_t *cli, int count)
 {
     if (!writeback_queue.size())

src/client/etcd_state_client.cpp

@@ -253,7 +253,7 @@ void etcd_state_client_t::parse_config(const json11::Json & config)
     this->etcd_ws_keepalive_interval = config["etcd_ws_keepalive_interval"].uint64_value();
     if (this->etcd_ws_keepalive_interval <= 0)
     {
-        this->etcd_ws_keepalive_interval = 30;
+        this->etcd_ws_keepalive_interval = 5;
     }
     this->max_etcd_attempts = config["max_etcd_attempts"].uint64_value();
     if (this->max_etcd_attempts <= 0)
@@ -890,6 +890,10 @@ void etcd_state_client_t::parse_state(const etcd_kv_t & kv)
                 }
             }
         }
+        if (on_change_pool_config_hook)
+        {
+            on_change_pool_config_hook();
+        }
     }
     else if (key == etcd_prefix+"/config/pgs")
     {
@@ -1028,13 +1032,19 @@ void etcd_state_client_t::parse_state(const etcd_kv_t & kv)
         else if (value.is_null())
         {
             auto & pg_cfg = this->pool_config[pool_id].pg_config[pg_num];
+            auto prev_primary = pg_cfg.cur_primary;
             pg_cfg.state_exists = false;
             pg_cfg.cur_primary = 0;
             pg_cfg.cur_state = 0;
+            if (on_change_pg_state_hook)
+            {
+                on_change_pg_state_hook(pool_id, pg_num, prev_primary);
+            }
         }
         else
         {
             auto & pg_cfg = this->pool_config[pool_id].pg_config[pg_num];
+            auto prev_primary = pg_cfg.cur_primary;
             pg_cfg.state_exists = true;
             osd_num_t cur_primary = value["primary"].uint64_value();
             int state = 0;
@@ -1065,6 +1075,10 @@ void etcd_state_client_t::parse_state(const etcd_kv_t & kv)
             }
             pg_cfg.cur_primary = cur_primary;
             pg_cfg.cur_state = state;
+            if (on_change_pg_state_hook)
+            {
+                on_change_pg_state_hook(pool_id, pg_num, prev_primary);
+            }
         }
     }
     else if (key.substr(0, etcd_prefix.length()+11) == etcd_prefix+"/osd/state/")

src/client/etcd_state_client.h

@@ -20,9 +20,11 @@
 #define MAX_DATA_BLOCK_SIZE 128*1024*1024
 #define DEFAULT_BITMAP_GRANULARITY 4096
 
+#ifndef IMMEDIATE_NONE
 #define IMMEDIATE_NONE 0
 #define IMMEDIATE_SMALL 1
 #define IMMEDIATE_ALL 2
+#endif
 
 struct etcd_kv_t
 {
@@ -101,7 +103,7 @@ protected:
     void pick_next_etcd();
 public:
     int etcd_keepalive_timeout = 30;
-    int etcd_ws_keepalive_interval = 30;
+    int etcd_ws_keepalive_interval = 5;
     int max_etcd_attempts = 5;
     int etcd_quick_timeout = 1000;
     int etcd_slow_timeout = 5000;
@@ -125,6 +127,8 @@ public:
     std::function<void(json11::Json::object &)> on_load_config_hook;
     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, 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;
     std::function<void()> on_reload_hook;