Release 0.6.12

etcd connection stability, clang & elbrus support

- Fix build under CLang and Elbrus LCC compilers, making Vitastor compatible
  with Elbrus CPUs :)
- Completely fix the bug where OSDs didn't connect to peers and incorrectly marked
  PGs as incomplete
- Limit I/O depth for deletes the same way as for small writes. Makes OSD crashes
  with "Assertion failed: sqe != NULL" during image deletion go away
- Fix a very old, but rare, journaling bug (credits to https://github.com/mirrorll)
- Fix flushing of unclean journaled objects leading to OSDs sometimes hanging
  after failover in EC setups (bug was introduced in 0.6.7)
- Fix several problems that could prevent smooth operation of a Vitastor cluster
  under the condition of partial etcd failure:
  - OSDs could randomly fail due to too strict error handling
  - New clients and OSDs could be unable to start because of the lack of retries
  - CLI could fail some commands because of the lack of retries
  - Monitor could stop receiving state updates because of the lack of websocket pings
- Fix monitor being unable to rebalance PGs after a downscale of pool pg_size (3->2)
- Exit with failure when trying to nbd map or benchmark a non-existing image
- Use HTTP keep-alive for etcd connections
- Allow to configure etcd request timeouts and retries
- Allow to configure NBD timeout, max devices and partitions, and set default to
  up to 64 devices with up to 3 partitions each

CMakeLists.txt

@@ -2,6 +2,6 @@ cmake_minimum_required(VERSION 2.8)
 
 project(vitastor)
 
-set(VERSION "0.6.10")
+set(VERSION "0.6.12")
 
 add_subdirectory(src)

csi/Makefile

@@ -1,4 +1,4 @@
-VERSION ?= v0.6.10
+VERSION ?= v0.6.12
 
 all: build push
 

csi/deploy/004-csi-nodeplugin.yaml

@@ -49,7 +49,7 @@ spec:
             capabilities:
               add: ["SYS_ADMIN"]
             allowPrivilegeEscalation: true
-          image: vitalif/vitastor-csi:v0.6.10
+          image: vitalif/vitastor-csi:v0.6.12
           args:
             - "--node=$(NODE_ID)"
             - "--endpoint=$(CSI_ENDPOINT)"

csi/deploy/007-csi-provisioner.yaml

@@ -116,7 +116,7 @@ spec:
             privileged: true
             capabilities:
               add: ["SYS_ADMIN"]
-          image: vitalif/vitastor-csi:v0.6.10
+          image: vitalif/vitastor-csi:v0.6.12
           args:
             - "--node=$(NODE_ID)"
             - "--endpoint=$(CSI_ENDPOINT)"

csi/src/config.go

@@ -5,7 +5,7 @@ package vitastor
 
 const (
     vitastorCSIDriverName    = "csi.vitastor.io"
-    vitastorCSIDriverVersion = "0.6.10"
+    vitastorCSIDriverVersion = "0.6.12"
 )
 
 // Config struct fills the parameters of request or user input

debian/changelog

@@ -1,4 +1,4 @@
-vitastor (0.6.10-1) unstable; urgency=medium
+vitastor (0.6.12-1) unstable; urgency=medium
 
   * RDMA support
   * Bugfixes

debian/vitastor.Dockerfile

@@ -33,8 +33,8 @@ RUN set -e -x; \
     mkdir -p /root/packages/vitastor-$REL; \
     rm -rf /root/packages/vitastor-$REL/*; \
     cd /root/packages/vitastor-$REL; \
-    cp -r /root/vitastor vitastor-0.6.10; \
-    cd vitastor-0.6.10; \
+    cp -r /root/vitastor vitastor-0.6.12; \
+    cd vitastor-0.6.12; \
     ln -s /root/fio-build/fio-*/ ./fio; \
     FIO=$(head -n1 fio/debian/changelog | perl -pe 's/^.*\((.*?)\).*$/$1/'); \
     ls /usr/include/linux/raw.h || cp ./debian/raw.h /usr/include/linux/raw.h; \
@@ -47,8 +47,8 @@ RUN set -e -x; \
     rm -rf a b; \
     echo "dep:fio=$FIO" > debian/fio_version; \
     cd /root/packages/vitastor-$REL; \
-    tar --sort=name --mtime='2020-01-01' --owner=0 --group=0 --exclude=debian -cJf vitastor_0.6.10.orig.tar.xz vitastor-0.6.10; \
-    cd vitastor-0.6.10; \
+    tar --sort=name --mtime='2020-01-01' --owner=0 --group=0 --exclude=debian -cJf vitastor_0.6.12.orig.tar.xz vitastor-0.6.12; \
+    cd vitastor-0.6.12; \
     V=$(head -n1 debian/changelog | perl -pe 's/^.*\((.*?)\).*$/$1/'); \
     DEBFULLNAME="Vitaliy Filippov <vitalif@yourcmc.ru>" dch -D $REL -v "$V""$REL" "Rebuild for $REL"; \
     DEB_BUILD_OPTIONS=nocheck dpkg-buildpackage --jobs=auto -sa; \

mon/lp-optimizer.js

@@ -50,7 +50,7 @@ async function lp_solve(text)
     return { score, vars };
 }
 
-async function optimize_initial({ osd_tree, pg_count, pg_size = 3, pg_minsize = 2, max_combinations = 10000, parity_space = 1, round_robin = false })
+async function optimize_initial({ osd_tree, pg_count, pg_size = 3, pg_minsize = 2, max_combinations = 10000, parity_space = 1, ordered = false })
 {
     if (!pg_count || !osd_tree)
     {
@@ -92,7 +92,7 @@ async function optimize_initial({ osd_tree, pg_count, pg_size = 3, pg_minsize =
         console.log(lp);
         throw new Error('Problem is infeasible or unbounded - is it a bug?');
     }
-    const int_pgs = make_int_pgs(lp_result.vars, pg_count, round_robin);
+    const int_pgs = make_int_pgs(lp_result.vars, pg_count, ordered);
     const eff = pg_list_space_efficiency(int_pgs, all_weights, pg_minsize, parity_space);
     const res = {
         score: lp_result.score,
@@ -140,20 +140,20 @@ function make_int_pgs(weights, pg_count, round_robin)
     return int_pgs;
 }
 
-function calc_intersect_weights(pg_size, pg_count, prev_weights, all_pgs)
+function calc_intersect_weights(old_pg_size, pg_size, pg_count, prev_weights, all_pgs, ordered)
 {
     const move_weights = {};
-    if ((1 << pg_size) < pg_count)
+    if ((1 << old_pg_size) < pg_count)
     {
         const intersect = {};
         for (const pg_name in prev_weights)
         {
             const pg = pg_name.substr(3).split(/_/);
-            for (let omit = 1; omit < (1 << pg_size); omit++)
+            for (let omit = 1; omit < (1 << old_pg_size); omit++)
             {
                 let pg_omit = [ ...pg ];
-                let intersect_count = pg_size;
-                for (let i = 0; i < pg_size; i++)
+                let intersect_count = old_pg_size;
+                for (let i = 0; i < old_pg_size; i++)
                 {
                     if (omit & (1 << i))
                     {
@@ -161,6 +161,8 @@ function calc_intersect_weights(pg_size, pg_count, prev_weights, all_pgs)
                         intersect_count--;
                     }
                 }
+                if (!ordered)
+                    pg_omit = pg_omit.filter(n => n).sort();
                 pg_omit = pg_omit.join(':');
                 intersect[pg_omit] = Math.max(intersect[pg_omit] || 0, intersect_count);
             }
@@ -174,10 +176,10 @@ function calc_intersect_weights(pg_size, pg_count, prev_weights, all_pgs)
                 for (let i = 0; i < pg_size; i++)
                 {
                     if (omit & (1 << i))
-                    {
                         pg_omit[i] = '';
-                    }
                 }
+                if (!ordered)
+                    pg_omit = pg_omit.filter(n => n).sort();
                 pg_omit = pg_omit.join(':');
                 max_int = Math.max(max_int, intersect[pg_omit] || 0);
             }
@@ -186,15 +188,18 @@ function calc_intersect_weights(pg_size, pg_count, prev_weights, all_pgs)
     }
     else
     {
-        const prev_pg_hashed = Object.keys(prev_weights).map(pg_name => pg_name.substr(3).split(/_/).reduce((a, c) => { a[c] = 1; return a; }, {}));
+        const prev_pg_hashed = Object.keys(prev_weights).map(pg_name => pg_name
+            .substr(3).split(/_/).reduce((a, c, i) => { a[c] = i+1; return a; }, {}));
         for (const pg of all_pgs)
         {
             if (!prev_weights['pg_'+pg.join('_')])
             {
                 let max_int = 0;
-                for (const prev_hash in prev_pg_hashed)
+                for (const prev_hash of prev_pg_hashed)
                 {
-                    const intersect_count = pg.reduce((a, osd) => a + (prev_hash[osd] ? 1 : 0), 0);
+                    const intersect_count = ordered
+                        ? pg.reduce((a, osd, i) => a + (prev_hash[osd] == 1+i ? 1 : 0), 0)
+                        : pg.reduce((a, osd, i) => a + (prev_hash[osd] ? 1 : 0), 0);
                     if (max_int < intersect_count)
                     {
                         max_int = intersect_count;
@@ -243,7 +248,7 @@ function add_valid_previous(osd_tree, prev_weights, all_pgs)
 }
 
 // Try to minimize data movement
-async function optimize_change({ prev_pgs: prev_int_pgs, osd_tree, pg_size = 3, pg_minsize = 2, max_combinations = 10000, parity_space = 1 })
+async function optimize_change({ prev_pgs: prev_int_pgs, osd_tree, pg_size = 3, pg_minsize = 2, max_combinations = 10000, parity_space = 1, ordered = false })
 {
     if (!osd_tree)
     {
@@ -266,9 +271,13 @@ async function optimize_change({ prev_pgs: prev_int_pgs, osd_tree, pg_size = 3,
             prev_pg_per_osd[osd].push([ pg_name, (i >= pg_minsize ? parity_space : 1) ]);
         }
     }
+    const old_pg_size = prev_int_pgs[0].length;
     // Get all combinations
     let all_pgs = random_combinations(osd_tree, pg_size, max_combinations, parity_space > 1);
-    add_valid_previous(osd_tree, prev_weights, all_pgs);
+    if (old_pg_size == pg_size)
+    {
+        add_valid_previous(osd_tree, prev_weights, all_pgs);
+    }
     all_pgs = Object.values(all_pgs);
     const pg_per_osd = {};
     for (const pg of all_pgs)
@@ -282,7 +291,7 @@ async function optimize_change({ prev_pgs: prev_int_pgs, osd_tree, pg_size = 3,
         }
     }
     // Penalize PGs based on their similarity to old PGs
-    const move_weights = calc_intersect_weights(pg_size, pg_count, prev_weights, all_pgs);
+    const move_weights = calc_intersect_weights(old_pg_size, pg_size, pg_count, prev_weights, all_pgs, ordered);
     // Calculate total weight - old PG weights
     const all_pg_names = all_pgs.map(pg => 'pg_'+pg.join('_'));
     const all_pgs_hash = all_pg_names.reduce((a, c) => { a[c] = true; return a; }, {});
@@ -373,11 +382,35 @@ async function optimize_change({ prev_pgs: prev_int_pgs, osd_tree, pg_size = 3,
         {
             differs++;
         }
-        for (let j = 0; j < pg_size; j++)
+    }
+    if (ordered)
+    {
+        for (let i = 0; i < pg_count; i++)
         {
-            if (new_pgs[i][j] != prev_int_pgs[i][j])
+            for (let j = 0; j < pg_size; j++)
             {
-                osd_differs++;
+                if (new_pgs[i][j] != prev_int_pgs[i][j])
+                {
+                    osd_differs++;
+                }
+            }
+        }
+    }
+    else
+    {
+        for (let i = 0; i < pg_count; i++)
+        {
+            const old_map = prev_int_pgs[i].reduce((a, c) => { a[c] = (a[c]|0) + 1; return a; }, {});
+            for (let j = 0; j < pg_size; j++)
+            {
+                if ((0|old_map[new_pgs[i][j]]) > 0)
+                {
+                    old_map[new_pgs[i][j]]--;
+                }
+                else
+                {
+                    osd_differs++;
+                }
             }
         }
     }

mon/mon.js

@@ -84,7 +84,12 @@ const etcd_tree = {
             osd_ping_timeout: 5, // seconds. min: 1
             up_wait_retry_interval: 500, // ms. min: 50
             // osd
-            etcd_report_interval: 5,
+            etcd_report_interval: 5, // seconds
+            max_etcd_attempts: 5,
+            etcd_quick_timeout: 1000, // ms
+            etcd_slow_timeout: 5000, // ms
+            etcd_keepalive_timeout: 30, // seconds, default is min(30, etcd_report_interval*2)
+            etcd_ws_keepalive_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",
@@ -341,6 +346,9 @@ class Mon
         this.etcd_start_timeout = (config.etcd_start_timeout || 5) * 1000;
         this.state = JSON.parse(JSON.stringify(this.constructor.etcd_tree));
         this.signals_set = false;
+        this.ws = null;
+        this.ws_alive = false;
+        this.ws_keepalive_timer = null;
         this.on_stop_cb = () => this.on_stop(0).catch(console.error);
     }
 
@@ -383,7 +391,7 @@ class Mon
         for (const pool_id in this.state.config.pools)
         {
             if (!this.state.pool.stats[pool_id] ||
-                !this.state.pool.stats[pool_id].pg_real_size)
+                !Number(this.state.pool.stats[pool_id].pg_real_size))
             {
                 // Generate missing data in etcd
                 this.state.config.pgs.hash = null;
@@ -461,8 +469,20 @@ class Mon
 
     restart_watcher(cur_addr)
     {
+        if (this.ws)
+        {
+            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;
+        }
         this.start_watcher(this.config.etcd_mon_retries).catch(this.die);
     }
 
@@ -482,6 +502,7 @@ class Mon
                 const timer_id = setTimeout(() =>
                 {
                     this.ws.close();
+                    this.ws = null;
                     ok(false);
                 }, this.config.etcd_mon_timeout);
                 this.ws = new WebSocket(base+'/watch');
@@ -510,6 +531,20 @@ class Mon
             this.die('Failed to open etcd watch websocket');
         }
         const cur_addr = this.selected_etcd_url;
+        this.ws_alive = true;
+        this.ws_keepalive_timer = setInterval(() =>
+        {
+            if (this.ws_alive)
+            {
+                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.config.etcd_keepalive_interval) || 30)*1000);
         this.ws.on('error', () => this.restart_watcher(cur_addr));
         this.ws.send(JSON.stringify({
             create_request: {
@@ -522,6 +557,7 @@ class Mon
         }));
         this.ws.on('message', (msg) =>
         {
+            this.ws_alive = true;
             let data;
             try
             {
@@ -558,7 +594,7 @@ class Mon
                     console.log('Revision '+data.result.header.revision+' events: ');
                 }
                 this.etcd_watch_revision = BigInt(data.result.header.revision)+BigInt(1);
-                for (const e of data.result.events)
+                for (const e of data.result.events||[])
                 {
                     this.parse_kv(e.kv);
                     const key = e.kv.key.substr(this.etcd_prefix.length);
@@ -709,10 +745,13 @@ class Mon
         for (const node_id in this.state.config.node_placement||{})
         {
             const node_cfg = this.state.config.node_placement[node_id];
-            if (!node_id || /^\d/.exec(node_id) ||
-                !node_cfg.level || !levels[node_cfg.level])
+            if (/^\d+$/.exec(node_id))
             {
-                // All nodes must have non-empty non-numeric IDs and valid levels
+                node_cfg.level = 'osd';
+            }
+            if (!node_id || !node_cfg.level || !levels[node_cfg.level])
+            {
+                // All nodes must have non-empty IDs and valid levels
                 continue;
             }
             tree[node_id] = { id: node_id, level: node_cfg.level, parent: node_cfg.parent, children: [] };
@@ -745,10 +784,10 @@ class Mon
                         .reduce((a, c) => { a[c] = true; return a; }, {});
                 }
                 delete tree[osd_num].children;
-                if (!tree[tree[osd_num].parent])
+                if (!tree[stat.host])
                 {
-                    tree[tree[osd_num].parent] = {
-                        id: tree[osd_num].parent,
+                    tree[stat.host] = {
+                        id: stat.host,
                         level: 'host',
                         parent: null,
                         children: [],
@@ -1094,7 +1133,7 @@ class Mon
                     pg_size: pool_cfg.pg_size,
                     pg_minsize: pool_cfg.pg_minsize,
                     max_combinations: pool_cfg.max_osd_combinations,
-                    round_robin: pool_cfg.scheme != 'replicated',
+                    ordered: pool_cfg.scheme != 'replicated',
                 };
                 let optimize_result;
                 if (old_pg_count > 0)
@@ -1117,10 +1156,6 @@ class Mon
                         {
                             pg.push(0);
                         }
-                        while (pg.length > pool_cfg.pg_size)
-                        {
-                            pg.pop();
-                        }
                     }
                     if (!this.state.config.pgs.hash)
                     {
@@ -1156,8 +1191,8 @@ class Mon
                 this.state.pool.stats[pool_id] = {
                     used_raw_tb: (this.state.pool.stats[pool_id]||{}).used_raw_tb || 0,
                     total_raw_tb: optimize_result.space,
-                    pg_real_size: pg_effsize,
-                    raw_to_usable: pg_effsize / (pool_cfg.scheme === 'replicated'
+                    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),
                 };

mon/test-optimize-simple.js

@@ -5,21 +5,45 @@ const LPOptimizer = require('./lp-optimizer.js');
 
 async function run()
 {
-    const osd_tree = { a: { 1: 1 }, b: { 2: 1 }, c: { 3: 1 } };
+    const osd_tree = {
+        100: { 1: 1 },
+        200: { 2: 1 },
+        300: { 3: 1 },
+    };
+
     let res;
 
     console.log('16 PGs, size=3');
-    res = await LPOptimizer.optimize_initial({ osd_tree, pg_size: 3, pg_count: 16 });
+    res = await LPOptimizer.optimize_initial({ osd_tree, pg_size: 3, pg_count: 16, ordered: false });
     LPOptimizer.print_change_stats(res, false);
-
-    console.log('\nReduce PG size to 2');
-    res = await LPOptimizer.optimize_change({ prev_pgs: res.int_pgs.map(pg => pg.slice(0, 2)), osd_tree, pg_size: 2 });
+    assert(res.space == 3, 'Initial distribution');
+    console.log('\nChange size to 2');
+    res = await LPOptimizer.optimize_change({ prev_pgs: res.int_pgs, osd_tree, pg_size: 2, ordered: false });
     LPOptimizer.print_change_stats(res, false);
-
+    assert(res.space >= 3*14/16 && res.osd_differs == 0, 'Redistribution');
     console.log('\nRemove OSD 3');
-    delete osd_tree['c'];
-    res = await LPOptimizer.optimize_change({ prev_pgs: res.int_pgs, osd_tree, pg_size: 2 });
+    const no3_tree = { ...osd_tree };
+    delete no3_tree['300'];
+    res = await LPOptimizer.optimize_change({ prev_pgs: res.int_pgs, osd_tree: no3_tree, pg_size: 2, ordered: false });
     LPOptimizer.print_change_stats(res, false);
+    assert(res.space == 2, 'Redistribution after OSD removal');
+
+    console.log('\n16 PGs, size=3, ordered');
+    res = await LPOptimizer.optimize_initial({ osd_tree, pg_size: 3, pg_count: 16, ordered: true });
+    LPOptimizer.print_change_stats(res, false);
+    assert(res.space == 3, 'Initial distribution');
+    console.log('\nChange size to 2, ordered');
+    res = await LPOptimizer.optimize_change({ prev_pgs: res.int_pgs, osd_tree, pg_size: 2, ordered: true });
+    LPOptimizer.print_change_stats(res, false);
+    assert(res.space >= 3*14/16 && res.osd_differs < 8, 'Redistribution');
+}
+
+function assert(cond, txt)
+{
+    if (!cond)
+    {
+        throw new Error((txt||'test')+' failed');
+    }
 }
 
 run().catch(console.error);

mon/test-optimize-undersized.js

@@ -45,30 +45,45 @@ async function run()
     console.log('Empty tree:');
     let res = await LPOptimizer.optimize_initial({ osd_tree: cur_tree, pg_size: 3, pg_count: 256 });
     LPOptimizer.print_change_stats(res, false);
+    assert(res.space == 0);
     console.log('\nAdding 1st failure domain:');
     cur_tree['dom1'] = osd_tree['dom1'];
     res = await LPOptimizer.optimize_change({ prev_pgs: res.int_pgs, osd_tree: cur_tree, pg_size: 3 });
     LPOptimizer.print_change_stats(res, false);
+    assert(res.space == 12 && res.total_space == 12);
     console.log('\nAdding 2nd failure domain:');
     cur_tree['dom2'] = osd_tree['dom2'];
     res = await LPOptimizer.optimize_change({ prev_pgs: res.int_pgs, osd_tree: cur_tree, pg_size: 3 });
     LPOptimizer.print_change_stats(res, false);
+    assert(res.space == 24 && res.total_space == 24);
     console.log('\nAdding 3rd failure domain:');
     cur_tree['dom3'] = osd_tree['dom3'];
     res = await LPOptimizer.optimize_change({ prev_pgs: res.int_pgs, osd_tree: cur_tree, pg_size: 3 });
     LPOptimizer.print_change_stats(res, false);
+    assert(res.space == 36 && res.total_space == 36);
     console.log('\nRemoving 3rd failure domain:');
     delete cur_tree['dom3'];
     res = await LPOptimizer.optimize_change({ prev_pgs: res.int_pgs, osd_tree: cur_tree, pg_size: 3 });
     LPOptimizer.print_change_stats(res, false);
+    assert(res.space == 24 && res.total_space == 24);
     console.log('\nRemoving 2nd failure domain:');
     delete cur_tree['dom2'];
     res = await LPOptimizer.optimize_change({ prev_pgs: res.int_pgs, osd_tree: cur_tree, pg_size: 3 });
     LPOptimizer.print_change_stats(res, false);
+    assert(res.space == 12 && res.total_space == 12);
     console.log('\nRemoving 1st failure domain:');
     delete cur_tree['dom1'];
     res = await LPOptimizer.optimize_change({ prev_pgs: res.int_pgs, osd_tree: cur_tree, pg_size: 3 });
     LPOptimizer.print_change_stats(res, false);
+    assert(res.space == 0);
+}
+
+function assert(cond, txt)
+{
+    if (!cond)
+    {
+        throw new Error((txt||'test')+' failed');
+    }
 }
 
 run().catch(console.error);

patches/cinder-vitastor.py

@@ -50,7 +50,7 @@ from cinder.volume import configuration
 from cinder.volume import driver
 from cinder.volume import volume_utils
 
-VERSION = '0.6.10'
+VERSION = '0.6.12'
 
 LOG = logging.getLogger(__name__)
 

rpm/build-tarball.sh

@@ -25,4 +25,4 @@ rm fio
 mv fio-copy fio
 FIO=`rpm -qi fio | perl -e 'while(<>) { /^Epoch[\s:]+(\S+)/ && print "$1:"; /^Version[\s:]+(\S+)/ && print $1; /^Release[\s:]+(\S+)/ && print "-$1"; }'`
 perl -i -pe 's/(Requires:\s*fio)([^\n]+)?/$1 = '$FIO'/' $VITASTOR/rpm/vitastor-el$EL.spec
-tar --transform 's#^#vitastor-0.6.10/#' --exclude 'rpm/*.rpm' -czf $VITASTOR/../vitastor-0.6.10$(rpm --eval '%dist').tar.gz *
+tar --transform 's#^#vitastor-0.6.12/#' --exclude 'rpm/*.rpm' -czf $VITASTOR/../vitastor-0.6.12$(rpm --eval '%dist').tar.gz *

rpm/vitastor-el7.Dockerfile

@@ -34,7 +34,7 @@ ADD . /root/vitastor
 RUN set -e; \
     cd /root/vitastor/rpm; \
     sh build-tarball.sh; \
-    cp /root/vitastor-0.6.10.el7.tar.gz ~/rpmbuild/SOURCES; \
+    cp /root/vitastor-0.6.12.el7.tar.gz ~/rpmbuild/SOURCES; \
     cp vitastor-el7.spec ~/rpmbuild/SPECS/vitastor.spec; \
     cd ~/rpmbuild/SPECS/; \
     rpmbuild -ba vitastor.spec; \

rpm/vitastor-el7.spec

@@ -1,11 +1,11 @@
 Name:           vitastor
-Version:        0.6.10
+Version:        0.6.12
 Release:        1%{?dist}
 Summary:        Vitastor, a fast software-defined clustered block storage
 
 License:        Vitastor Network Public License 1.1
 URL:            https://vitastor.io/
-Source0:        vitastor-0.6.10.el7.tar.gz
+Source0:        vitastor-0.6.12.el7.tar.gz
 
 BuildRequires:  liburing-devel >= 0.6
 BuildRequires:  gperftools-devel

rpm/vitastor-el8.Dockerfile

@@ -33,7 +33,7 @@ ADD . /root/vitastor
 RUN set -e; \
     cd /root/vitastor/rpm; \
     sh build-tarball.sh; \
-    cp /root/vitastor-0.6.10.el8.tar.gz ~/rpmbuild/SOURCES; \
+    cp /root/vitastor-0.6.12.el8.tar.gz ~/rpmbuild/SOURCES; \
     cp vitastor-el8.spec ~/rpmbuild/SPECS/vitastor.spec; \
     cd ~/rpmbuild/SPECS/; \
     rpmbuild -ba vitastor.spec; \

rpm/vitastor-el8.spec

@@ -1,11 +1,11 @@
 Name:           vitastor
-Version:        0.6.10
+Version:        0.6.12
 Release:        1%{?dist}
 Summary:        Vitastor, a fast software-defined clustered block storage
 
 License:        Vitastor Network Public License 1.1
 URL:            https://vitastor.io/
-Source0:        vitastor-0.6.10.el8.tar.gz
+Source0:        vitastor-0.6.12.el8.tar.gz
 
 BuildRequires:  liburing-devel >= 0.6
 BuildRequires:  gperftools-devel

src/CMakeLists.txt

@@ -15,7 +15,7 @@ if("${CMAKE_INSTALL_PREFIX}" MATCHES "^/usr/local/?$")
 	set(CMAKE_INSTALL_RPATH "${CMAKE_INSTALL_PREFIX}/${CMAKE_INSTALL_LIBDIR}")
 endif()
 
-add_definitions(-DVERSION="0.6.10")
+add_definitions(-DVERSION="0.6.12")
 add_definitions(-Wall -Wno-sign-compare -Wno-comment -Wno-parentheses -Wno-pointer-arith -fdiagnostics-color=always -I ${CMAKE_SOURCE_DIR}/src)
 if (${WITH_ASAN})
 	add_definitions(-fsanitize=address -fno-omit-frame-pointer)
@@ -88,8 +88,8 @@ 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 msgr_stop.cpp msgr_op.cpp msgr_send.cpp msgr_receive.cpp ringloop.cpp ../json11/json11.cpp
+	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 base64.cpp ${MSGR_RDMA}
 )
 target_compile_options(vitastor_common PUBLIC -fPIC)
@@ -112,6 +112,7 @@ if (${WITH_FIO})
 	add_library(fio_vitastor_sec SHARED
 		fio_sec_osd.cpp
 		rw_blocking.cpp
+		addr_util.cpp
 	)
 	target_link_libraries(fio_vitastor_sec
 		tcmalloc_minimal
@@ -189,11 +190,11 @@ endif (${WITH_QEMU})
 ### Test stubs
 
 # stub_osd, stub_bench, osd_test
-add_executable(stub_osd stub_osd.cpp rw_blocking.cpp)
+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)
+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)
+add_executable(osd_test osd_test.cpp rw_blocking.cpp addr_util.cpp)
 target_link_libraries(osd_test tcmalloc_minimal)
 
 # osd_rmw_test

src/addr_util.cpp

@@ -0,0 +1,188 @@
+#include <arpa/inet.h>
+#include <net/if.h>
+#include <sys/types.h>
+#include <ifaddrs.h>
+#include <string.h>
+#include <stdio.h>
+
+#include <stdexcept>
+
+#include "addr_util.h"
+
+bool string_to_addr(std::string str, bool parse_port, int default_port, struct sockaddr *addr)
+{
+    if (parse_port)
+    {
+        int p = str.rfind(':');
+        if (p != std::string::npos && !(str.length() > 0 && str[p-1] == ']')) // "[ipv6]" which contains ':'
+        {
+            char null_byte = 0;
+            int n = sscanf(str.c_str()+p+1, "%d%c", &default_port, &null_byte);
+            if (n != 1 || default_port >= 0x10000)
+                return false;
+            str = str.substr(0, p);
+        }
+    }
+    if (inet_pton(AF_INET, str.c_str(), &((struct sockaddr_in*)addr)->sin_addr) == 1)
+    {
+        addr->sa_family = AF_INET;
+        ((struct sockaddr_in*)addr)->sin_port = htons(default_port);
+        return true;
+    }
+    if (str.length() >= 2 && str[0] == '[' && str[str.length()-1] == ']')
+        str = str.substr(1, str.length()-2);
+    if (inet_pton(AF_INET6, str.c_str(), &((struct sockaddr_in6*)addr)->sin6_addr) == 1)
+    {
+        addr->sa_family = AF_INET6;
+        ((struct sockaddr_in6*)addr)->sin6_port = htons(default_port);
+        return true;
+    }
+    return false;
+}
+
+std::string addr_to_string(const sockaddr &addr)
+{
+    char peer_str[256];
+    bool ok = false;
+    int port;
+    if (addr.sa_family == AF_INET)
+    {
+        ok = !!inet_ntop(AF_INET, &((sockaddr_in*)&addr)->sin_addr, peer_str, 256);
+        port = ntohs(((sockaddr_in*)&addr)->sin_port);
+    }
+    else if (addr.sa_family == AF_INET6)
+    {
+        ok = !!inet_ntop(AF_INET6, &((sockaddr_in6*)&addr)->sin6_addr, peer_str, 256);
+        port = ntohs(((sockaddr_in6*)&addr)->sin6_port);
+    }
+    else
+        throw std::runtime_error("Unknown address family "+std::to_string(addr.sa_family));
+    if (!ok)
+        throw std::runtime_error(std::string("inet_ntop: ") + strerror(errno));
+    return std::string(peer_str)+":"+std::to_string(port);
+}
+
+static bool cidr_match(const in_addr &addr, const in_addr &net, uint8_t bits)
+{
+    if (bits == 0)
+    {
+        // C99 6.5.7 (3): u32 << 32 is undefined behaviour
+        return true;
+    }
+    return !((addr.s_addr ^ net.s_addr) & htonl(0xFFFFFFFFu << (32 - bits)));
+}
+
+static bool cidr6_match(const in6_addr &address, const in6_addr &network, uint8_t bits)
+{
+    const uint32_t *a = address.s6_addr32;
+    const uint32_t *n = network.s6_addr32;
+    int bits_whole, bits_incomplete;
+    bits_whole = bits >> 5;         // number of whole u32
+    bits_incomplete = bits & 0x1F;  // number of bits in incomplete u32
+    if (bits_whole && memcmp(a, n, bits_whole << 2))
+        return false;
+    if (bits_incomplete)
+    {
+        uint32_t mask = htonl((0xFFFFFFFFu) << (32 - bits_incomplete));
+        if ((a[bits_whole] ^ n[bits_whole]) & mask)
+            return false;
+    }
+    return true;
+}
+
+struct addr_mask_t
+{
+    sa_family_t family;
+    in_addr ipv4;
+    in6_addr ipv6;
+    uint8_t bits;
+};
+
+std::vector<std::string> getifaddr_list(std::vector<std::string> mask_cfg, bool include_v6)
+{
+    std::vector<addr_mask_t> masks;
+    for (auto mask: mask_cfg)
+    {
+        unsigned bits = 0;
+        int p = mask.find('/');
+        if (p != std::string::npos)
+        {
+            char null_byte = 0;
+            if (sscanf(mask.c_str()+p+1, "%u%c", &bits, &null_byte) != 1 || bits > 128)
+            {
+                throw std::runtime_error((include_v6 ? "Invalid IPv4 address mask: " : "Invalid IP address mask: ") + mask);
+            }
+            mask = mask.substr(0, p);
+        }
+        in_addr ipv4;
+        in6_addr ipv6;
+        if (inet_pton(AF_INET, mask.c_str(), &ipv4) == 1)
+        {
+            if (bits > 32)
+            {
+                throw std::runtime_error((include_v6 ? "Invalid IPv4 address mask: " : "Invalid IP address mask: ") + mask);
+            }
+            masks.push_back((addr_mask_t){ .family = AF_INET, .ipv4 = ipv4, .bits = (uint8_t)bits });
+        }
+        else if (include_v6 && inet_pton(AF_INET6, mask.c_str(), &ipv6) == 1)
+        {
+            masks.push_back((addr_mask_t){ .family = AF_INET6, .ipv6 = ipv6, .bits = (uint8_t)bits });
+        }
+        else
+        {
+            throw std::runtime_error((include_v6 ? "Invalid IPv4 address mask: " : "Invalid IP address mask: ") + mask);
+        }
+    }
+    std::vector<std::string> addresses;
+    ifaddrs *list, *ifa;
+    if (getifaddrs(&list) == -1)
+    {
+        throw std::runtime_error(std::string("getifaddrs: ") + strerror(errno));
+    }
+    for (ifa = list; ifa != NULL; ifa = ifa->ifa_next)
+    {
+        if (!ifa->ifa_addr)
+        {
+            continue;
+        }
+        int family = ifa->ifa_addr->sa_family;
+        if ((family == AF_INET || family == AF_INET6 && include_v6) &&
+            (ifa->ifa_flags & (IFF_UP | IFF_RUNNING | IFF_LOOPBACK)) == (IFF_UP | IFF_RUNNING))
+        {
+            void *addr_ptr;
+            if (family == AF_INET)
+            {
+                addr_ptr = &((sockaddr_in *)ifa->ifa_addr)->sin_addr;
+            }
+            else
+            {
+                addr_ptr = &((sockaddr_in6 *)ifa->ifa_addr)->sin6_addr;
+            }
+            if (masks.size() > 0)
+            {
+                int i;
+                for (i = 0; i < masks.size(); i++)
+                {
+                    if (masks[i].family == family && (family == AF_INET
+                        ? cidr_match(*(in_addr*)addr_ptr, masks[i].ipv4, masks[i].bits)
+                        : cidr6_match(*(in6_addr*)addr_ptr, masks[i].ipv6, masks[i].bits)))
+                    {
+                        break;
+                    }
+                }
+                if (i >= masks.size())
+                {
+                    continue;
+                }
+            }
+            char addr[INET6_ADDRSTRLEN];
+            if (!inet_ntop(family, addr_ptr, addr, INET6_ADDRSTRLEN))
+            {
+                throw std::runtime_error(std::string("inet_ntop: ") + strerror(errno));
+            }
+            addresses.push_back(std::string(addr));
+        }
+    }
+    freeifaddrs(list);
+    return addresses;
+}

src/addr_util.h

@@ -0,0 +1,9 @@
+#pragma once
+
+#include <sys/socket.h>
+#include <string>
+#include <vector>
+
+bool string_to_addr(std::string str, bool parse_port, int default_port, struct sockaddr *addr);
+std::string addr_to_string(const sockaddr &addr);
+std::vector<std::string> getifaddr_list(std::vector<std::string> mask_cfg = std::vector<std::string>(), bool include_v6 = false);

src/blockstore_flush.cpp

@@ -185,7 +185,7 @@ void journal_flusher_t::release_trim()
 void journal_flusher_t::dump_diagnostics()
 {
     const char *unflushable_type = "";
-    obj_ver_id unflushable = { 0 };
+    obj_ver_id unflushable = {};
     // Try to find out if there is a flushable object for information
     for (object_id cur_oid: flush_queue)
     {
@@ -486,8 +486,8 @@ resume_1:
         if (bs->clean_entry_bitmap_size)
         {
             new_clean_bitmap = (bs->inmemory_meta
-                ? meta_new.buf + meta_new.pos*bs->clean_entry_size + sizeof(clean_disk_entry)
-                : bs->clean_bitmap + (clean_loc >> bs->block_order)*(2*bs->clean_entry_bitmap_size));
+                ? (uint8_t*)meta_new.buf + meta_new.pos*bs->clean_entry_size + sizeof(clean_disk_entry)
+                : (uint8_t*)bs->clean_bitmap + (clean_loc >> bs->block_order)*(2*bs->clean_entry_bitmap_size));
             if (clean_init_bitmap)
             {
                 memset(new_clean_bitmap, 0, bs->clean_entry_bitmap_size);
@@ -533,7 +533,7 @@ resume_1:
                 return false;
             }
             // zero out old metadata entry
-            memset(meta_old.buf + meta_old.pos*bs->clean_entry_size, 0, bs->clean_entry_size);
+            memset((uint8_t*)meta_old.buf + meta_old.pos*bs->clean_entry_size, 0, bs->clean_entry_size);
             await_sqe(15);
             data->iov = (struct iovec){ meta_old.buf, bs->meta_block_size };
             data->callback = simple_callback_w;
@@ -544,23 +544,25 @@ resume_1:
         }
         if (has_delete)
         {
-            clean_disk_entry *new_entry = (clean_disk_entry*)(meta_new.buf + meta_new.pos*bs->clean_entry_size);
+            clean_disk_entry *new_entry = (clean_disk_entry*)((uint8_t*)meta_new.buf + meta_new.pos*bs->clean_entry_size);
             if (new_entry->oid.inode != 0 && new_entry->oid != cur.oid)
             {
-                printf("Fatal error (metadata corruption or bug): tried to delete metadata entry %lu (%lx:%lx) while deleting %lx:%lx\n",
-                    clean_loc >> bs->block_order, new_entry->oid.inode, new_entry->oid.stripe, cur.oid.inode, cur.oid.stripe);
+                printf("Fatal error (metadata corruption or bug): tried to delete metadata entry %lu (%lx:%lx v%lu) while deleting %lx:%lx\n",
+                    clean_loc >> bs->block_order, new_entry->oid.inode, new_entry->oid.stripe,
+                    new_entry->version, cur.oid.inode, cur.oid.stripe);
                 exit(1);
             }
             // zero out new metadata entry
-            memset(meta_new.buf + meta_new.pos*bs->clean_entry_size, 0, bs->clean_entry_size);
+            memset((uint8_t*)meta_new.buf + meta_new.pos*bs->clean_entry_size, 0, bs->clean_entry_size);
         }
         else
         {
-            clean_disk_entry *new_entry = (clean_disk_entry*)(meta_new.buf + meta_new.pos*bs->clean_entry_size);
+            clean_disk_entry *new_entry = (clean_disk_entry*)((uint8_t*)meta_new.buf + meta_new.pos*bs->clean_entry_size);
             if (new_entry->oid.inode != 0 && new_entry->oid != cur.oid)
             {
-                printf("Fatal error (metadata corruption or bug): tried to overwrite non-zero metadata entry %lu (%lx:%lx) with %lx:%lx\n",
-                    clean_loc >> bs->block_order, new_entry->oid.inode, new_entry->oid.stripe, cur.oid.inode, cur.oid.stripe);
+                printf("Fatal error (metadata corruption or bug): tried to overwrite non-zero metadata entry %lu (%lx:%lx v%lu) with %lx:%lx v%lu\n",
+                    clean_loc >> bs->block_order, new_entry->oid.inode, new_entry->oid.stripe, new_entry->version,
+                    cur.oid.inode, cur.oid.stripe, cur.version);
                 exit(1);
             }
             new_entry->oid = cur.oid;
@@ -573,7 +575,7 @@ resume_1:
             if (bs->clean_entry_bitmap_size)
             {
                 void *bmp_ptr = bs->clean_entry_bitmap_size > sizeof(void*) ? dirty_end->second.bitmap : &dirty_end->second.bitmap;
-                memcpy((void*)(new_entry+1) + bs->clean_entry_bitmap_size, bmp_ptr, bs->clean_entry_bitmap_size);
+                memcpy((uint8_t*)(new_entry+1) + bs->clean_entry_bitmap_size, bmp_ptr, bs->clean_entry_bitmap_size);
             }
         }
         await_sqe(6);
@@ -760,7 +762,7 @@ bool journal_flusher_co::scan_dirty(int wait_base)
                         if (bs->journal.inmemory)
                         {
                             // Take it from memory
-                            memcpy(it->buf, bs->journal.buffer + submit_offset, submit_len);
+                            memcpy(it->buf, (uint8_t*)bs->journal.buffer + submit_offset, submit_len);
                         }
                         else
                         {
@@ -824,7 +826,7 @@ bool journal_flusher_co::modify_meta_read(uint64_t meta_loc, flusher_meta_write_
     wr.pos = ((meta_loc >> bs->block_order) % (bs->meta_block_size / bs->clean_entry_size));
     if (bs->inmemory_meta)
     {
-        wr.buf = bs->metadata_buffer + wr.sector;
+        wr.buf = (uint8_t*)bs->metadata_buffer + wr.sector;
         return true;
     }
     wr.it = flusher->meta_sectors.find(wr.sector);

src/blockstore_impl.cpp

@@ -235,6 +235,12 @@ void blockstore_impl_t::loop()
         {
             throw std::runtime_error(std::string("io_uring_submit: ") + strerror(-ret));
         }
+        for (auto s: journal.submitting_sectors)
+        {
+            // Mark journal sector writes as submitted
+            journal.sector_info[s].submit_id = 0;
+        }
+        journal.submitting_sectors.clear();
         if ((initial_ring_space - ringloop->space_left()) > 0)
         {
             live = true;
@@ -366,7 +372,7 @@ void blockstore_impl_t::enqueue_op(blockstore_op_t *op)
                     };
                 }
                 unstable_writes.clear();
-                op->callback = [this, old_callback](blockstore_op_t *op)
+                op->callback = [old_callback](blockstore_op_t *op)
                 {
                     obj_ver_id *vers = (obj_ver_id*)op->buf;
                     delete[] vers;

src/blockstore_impl.h

@@ -54,6 +54,14 @@
 #define IS_BIG_WRITE(st) (((st) & 0x0F) == BS_ST_BIG_WRITE)
 #define IS_DELETE(st) (((st) & 0x0F) == BS_ST_DELETE)
 
+#define BS_SUBMIT_CHECK_SQES(n) \
+    if (ringloop->space_left() < (n))\
+    {\
+        /* Pause until there are more requests available */\
+        PRIV(op)->wait_for = WAIT_SQE;\
+        return 0;\
+    }
+
 #define BS_SUBMIT_GET_SQE(sqe, data) \
     BS_SUBMIT_GET_ONLY_SQE(sqe); \
     struct ring_data_t *data = ((ring_data_t*)sqe->user_data)
@@ -170,7 +178,7 @@ struct blockstore_op_private_t
     std::vector<fulfill_read_t> read_vec;
 
     // Sync, write
-    uint64_t min_flushed_journal_sector, max_flushed_journal_sector;
+    int min_flushed_journal_sector, max_flushed_journal_sector;
 
     // Write
     struct iovec iov_zerofill[3];
@@ -251,6 +259,7 @@ class blockstore_impl_t
     int data_fd;
     uint64_t meta_size, meta_area, meta_len;
     uint64_t data_size, data_len;
+    uint64_t data_device_sect, meta_device_sect, journal_device_sect;
 
     void *metadata_buffer = NULL;
 
@@ -271,7 +280,7 @@ class blockstore_impl_t
 
     friend class blockstore_init_meta;
     friend class blockstore_init_journal;
-    friend class blockstore_journal_check_t;
+    friend struct blockstore_journal_check_t;
     friend class journal_flusher_t;
     friend class journal_flusher_co;
 
@@ -282,6 +291,10 @@ class blockstore_impl_t
     void open_journal();
     uint8_t* get_clean_entry_bitmap(uint64_t block_loc, int offset);
 
+    // Journaling
+    void prepare_journal_sector_write(int sector, blockstore_op_t *op);
+    void handle_journal_write(ring_data_t *data, uint64_t flush_id);
+
     // Asynchronous init
     int initialized;
     int metadata_buf_size;
@@ -309,21 +322,18 @@ class blockstore_impl_t
 
     // Sync
     int continue_sync(blockstore_op_t *op, bool queue_has_in_progress_sync);
-    void handle_sync_event(ring_data_t *data, blockstore_op_t *op);
     void ack_sync(blockstore_op_t *op);
 
     // Stabilize
     int dequeue_stable(blockstore_op_t *op);
     int continue_stable(blockstore_op_t *op);
     void mark_stable(const obj_ver_id & ov, bool forget_dirty = false);
-    void handle_stable_event(ring_data_t *data, blockstore_op_t *op);
     void stabilize_object(object_id oid, uint64_t max_ver);
 
     // Rollback
     int dequeue_rollback(blockstore_op_t *op);
     int continue_rollback(blockstore_op_t *op);
     void mark_rolled_back(const obj_ver_id & ov);
-    void handle_rollback_event(ring_data_t *data, blockstore_op_t *op);
     void erase_dirty(blockstore_dirty_db_t::iterator dirty_start, blockstore_dirty_db_t::iterator dirty_end, uint64_t clean_loc);
 
     // List

src/blockstore_init.cpp

@@ -148,7 +148,7 @@ resume_1:
         {
             GET_SQE();
             data->iov = {
-                metadata_buffer + (bs->inmemory_meta
+                (uint8_t*)metadata_buffer + (bs->inmemory_meta
                     ? metadata_read
                     : (prev == 1 ? bs->metadata_buf_size : 0)),
                 bs->meta_len - metadata_read > bs->metadata_buf_size ? bs->metadata_buf_size : bs->meta_len - metadata_read,
@@ -169,13 +169,13 @@ resume_1:
         if (prev_done)
         {
             void *done_buf = bs->inmemory_meta
-                ? (metadata_buffer + done_pos)
-                : (metadata_buffer + (prev_done == 2 ? bs->metadata_buf_size : 0));
+                ? ((uint8_t*)metadata_buffer + done_pos)
+                : ((uint8_t*)metadata_buffer + (prev_done == 2 ? bs->metadata_buf_size : 0));
             unsigned count = bs->meta_block_size / bs->clean_entry_size;
             for (int sector = 0; sector < done_len; sector += bs->meta_block_size)
             {
                 // handle <count> entries
-                handle_entries(done_buf + sector, count, bs->block_order);
+                handle_entries((uint8_t*)done_buf + sector, count, bs->block_order);
                 done_cnt += count;
             }
             prev_done = 0;
@@ -215,7 +215,7 @@ void blockstore_init_meta::handle_entries(void* entries, unsigned count, int blo
 {
     for (unsigned i = 0; i < count; i++)
     {
-        clean_disk_entry *entry = (clean_disk_entry*)(entries + i*bs->clean_entry_size);
+        clean_disk_entry *entry = (clean_disk_entry*)((uint8_t*)entries + i*bs->clean_entry_size);
         if (!bs->inmemory_meta && bs->clean_entry_bitmap_size)
         {
             memcpy(bs->clean_bitmap + (done_cnt+i)*2*bs->clean_entry_bitmap_size, &entry->bitmap, 2*bs->clean_entry_bitmap_size);
@@ -440,7 +440,7 @@ resume_1:
                 if (!bs->journal.inmemory)
                     submitted_buf = memalign_or_die(MEM_ALIGNMENT, JOURNAL_BUFFER_SIZE);
                 else
-                    submitted_buf = bs->journal.buffer + journal_pos;
+                    submitted_buf = (uint8_t*)bs->journal.buffer + journal_pos;
                 data->iov = {
                     submitted_buf,
                     end - journal_pos < JOURNAL_BUFFER_SIZE ? end - journal_pos : JOURNAL_BUFFER_SIZE,
@@ -570,7 +570,7 @@ int blockstore_init_journal::handle_journal_part(void *buf, uint64_t done_pos, u
     resume:
         while (pos < bs->journal.block_size)
         {
-            journal_entry *je = (journal_entry*)(buf + proc_pos - done_pos + pos);
+            journal_entry *je = (journal_entry*)((uint8_t*)buf + proc_pos - done_pos + pos);
             if (je->magic != JOURNAL_MAGIC || je_crc32(je) != je->crc32 ||
                 je->type < JE_MIN || je->type > JE_MAX || started && je->crc32_prev != crc32_last)
             {
@@ -619,7 +619,7 @@ int blockstore_init_journal::handle_journal_part(void *buf, uint64_t done_pos, u
                 if (location >= done_pos && location+je->small_write.len <= done_pos+len)
                 {
                     // data is within this buffer
-                    data_crc32 = crc32c(0, buf + location - done_pos, je->small_write.len);
+                    data_crc32 = crc32c(0, (uint8_t*)buf + location - done_pos, je->small_write.len);
                 }
                 else
                 {
@@ -634,7 +634,7 @@ int blockstore_init_journal::handle_journal_part(void *buf, uint64_t done_pos, u
                                 ? location+je->small_write.len : done[i].pos+done[i].len);
                             uint64_t part_begin = (location < done[i].pos ? done[i].pos : location);
                             covered += part_end - part_begin;
-                            data_crc32 = crc32c(data_crc32, done[i].buf + part_begin - done[i].pos, part_end - part_begin);
+                            data_crc32 = crc32c(data_crc32, (uint8_t*)done[i].buf + part_begin - done[i].pos, part_end - part_begin);
                         }
                     }
                     if (covered < je->small_write.len)
@@ -650,9 +650,9 @@ int blockstore_init_journal::handle_journal_part(void *buf, uint64_t done_pos, u
                     // interesting thing is that we must clear the corrupt entry if we're not readonly,
                     // because we don't write next entries in the same journal block
                     printf("Journal entry data is corrupt (data crc32 %x != %x)\n", data_crc32, je->small_write.crc32_data);
-                    memset(buf + proc_pos - done_pos + pos, 0, bs->journal.block_size - pos);
+                    memset((uint8_t*)buf + proc_pos - done_pos + pos, 0, bs->journal.block_size - pos);
                     bs->journal.next_free = prev_free;
-                    init_write_buf = buf + proc_pos - done_pos;
+                    init_write_buf = (uint8_t*)buf + proc_pos - done_pos;
                     init_write_sector = proc_pos;
                     return 0;
                 }
@@ -665,7 +665,7 @@ int blockstore_init_journal::handle_journal_part(void *buf, uint64_t done_pos, u
                         .version = je->small_write.version,
                     };
                     void *bmp = NULL;
-                    void *bmp_from = (void*)je + sizeof(journal_entry_small_write);
+                    void *bmp_from = (uint8_t*)je + sizeof(journal_entry_small_write);
                     if (bs->clean_entry_bitmap_size <= sizeof(void*))
                     {
                         memcpy(&bmp, bmp_from, bs->clean_entry_bitmap_size);
@@ -745,7 +745,7 @@ int blockstore_init_journal::handle_journal_part(void *buf, uint64_t done_pos, u
                         .version = je->big_write.version,
                     };
                     void *bmp = NULL;
-                    void *bmp_from = (void*)je + sizeof(journal_entry_big_write);
+                    void *bmp_from = (uint8_t*)je + sizeof(journal_entry_big_write);
                     if (bs->clean_entry_bitmap_size <= sizeof(void*))
                     {
                         memcpy(&bmp, bmp_from, bs->clean_entry_bitmap_size);

src/blockstore_init.h

@@ -6,7 +6,7 @@
 class blockstore_init_meta
 {
     blockstore_impl_t *bs;
-    int wait_state = 0, wait_count = 0;
+    int wait_state = 0;
     bool zero_on_init = false;
     void *metadata_buffer = NULL;
     uint64_t metadata_read = 0;

src/blockstore_journal.cpp

@@ -96,7 +96,8 @@ int blockstore_journal_check_t::check_available(blockstore_op_t *op, int entries
         next_pos = next_pos + data_after;
         if (next_pos > bs->journal.len)
         {
-            next_pos = bs->journal.block_size + data_after;
+            if (right_dir)
+                next_pos = bs->journal.block_size + data_after;
             right_dir = false;
         }
     }
@@ -136,13 +137,13 @@ journal_entry* prefill_single_journal_entry(journal_t & journal, uint16_t type,
         journal.in_sector_pos = 0;
         journal.next_free = (journal.next_free+journal.block_size) < journal.len ? journal.next_free + journal.block_size : journal.block_size;
         memset(journal.inmemory
-            ? journal.buffer + journal.sector_info[journal.cur_sector].offset
-            : journal.sector_buf + journal.block_size*journal.cur_sector, 0, journal.block_size);
+            ? (uint8_t*)journal.buffer + journal.sector_info[journal.cur_sector].offset
+            : (uint8_t*)journal.sector_buf + journal.block_size*journal.cur_sector, 0, journal.block_size);
     }
     journal_entry *je = (struct journal_entry*)(
         (journal.inmemory
-            ? journal.buffer + journal.sector_info[journal.cur_sector].offset
-            : journal.sector_buf + journal.block_size*journal.cur_sector) + journal.in_sector_pos
+            ? (uint8_t*)journal.buffer + journal.sector_info[journal.cur_sector].offset
+            : (uint8_t*)journal.sector_buf + journal.block_size*journal.cur_sector) + journal.in_sector_pos
     );
     journal.in_sector_pos += size;
     je->magic = JOURNAL_MAGIC;
@@ -153,22 +154,73 @@ journal_entry* prefill_single_journal_entry(journal_t & journal, uint16_t type,
     return je;
 }
 
-void prepare_journal_sector_write(journal_t & journal, int cur_sector, io_uring_sqe *sqe, std::function<void(ring_data_t*)> cb)
+void blockstore_impl_t::prepare_journal_sector_write(int cur_sector, blockstore_op_t *op)
 {
+    // Don't submit the same sector twice in the same batch
+    if (!journal.sector_info[cur_sector].submit_id)
+    {
+        io_uring_sqe *sqe = get_sqe();
+        // Caller must ensure availability of an SQE
+        assert(sqe != NULL);
+        ring_data_t *data = ((ring_data_t*)sqe->user_data);
+        journal.sector_info[cur_sector].written = true;
+        journal.sector_info[cur_sector].submit_id = ++journal.submit_id;
+        journal.submitting_sectors.push_back(cur_sector);
+        journal.sector_info[cur_sector].flush_count++;
+        data->iov = (struct iovec){
+            (journal.inmemory
+                ? (uint8_t*)journal.buffer + journal.sector_info[cur_sector].offset
+                : (uint8_t*)journal.sector_buf + journal.block_size*cur_sector),
+            journal.block_size
+        };
+        data->callback = [this, flush_id = journal.submit_id](ring_data_t *data) { handle_journal_write(data, flush_id); };
+        my_uring_prep_writev(
+            sqe, journal.fd, &data->iov, 1, journal.offset + journal.sector_info[cur_sector].offset
+        );
+    }
     journal.sector_info[cur_sector].dirty = false;
-    journal.sector_info[cur_sector].written = true;
-    journal.sector_info[cur_sector].flush_count++;
-    ring_data_t *data = ((ring_data_t*)sqe->user_data);
-    data->iov = (struct iovec){
-        (journal.inmemory
-            ? journal.buffer + journal.sector_info[cur_sector].offset
-            : journal.sector_buf + journal.block_size*cur_sector),
-        journal.block_size
-    };
-    data->callback = cb;
-    my_uring_prep_writev(
-        sqe, journal.fd, &data->iov, 1, journal.offset + journal.sector_info[cur_sector].offset
-    );
+    // But always remember that this operation has to wait until this exact journal write is finished
+    journal.flushing_ops.insert((pending_journaling_t){
+        .flush_id = journal.sector_info[cur_sector].submit_id,
+        .sector = cur_sector,
+        .op = op,
+    });
+    auto priv = PRIV(op);
+    priv->pending_ops++;
+    if (!priv->min_flushed_journal_sector)
+        priv->min_flushed_journal_sector = 1+cur_sector;
+    priv->max_flushed_journal_sector = 1+cur_sector;
+}
+
+void blockstore_impl_t::handle_journal_write(ring_data_t *data, uint64_t flush_id)
+{
+    live = true;
+    if (data->res != data->iov.iov_len)
+    {
+        // FIXME: our state becomes corrupted after a write error. maybe do something better than just die
+        throw std::runtime_error(
+            "journal write failed ("+std::to_string(data->res)+" != "+std::to_string(data->iov.iov_len)+
+            "). in-memory state is corrupted. AAAAAAAaaaaaaaaa!!!111"
+        );
+    }
+    auto fl_it = journal.flushing_ops.upper_bound((pending_journaling_t){ .flush_id = flush_id });
+    if (fl_it != journal.flushing_ops.end() && fl_it->flush_id == flush_id)
+    {
+        journal.sector_info[fl_it->sector].flush_count--;
+    }
+    while (fl_it != journal.flushing_ops.end() && fl_it->flush_id == flush_id)
+    {
+        auto priv = PRIV(fl_it->op);
+        priv->pending_ops--;
+        assert(priv->pending_ops >= 0);
+        if (priv->pending_ops == 0)
+        {
+            release_journal_sectors(fl_it->op);
+            priv->op_state++;
+            ringloop->wakeup();
+        }
+        journal.flushing_ops.erase(fl_it++);
+    }
 }
 
 journal_t::~journal_t()

src/blockstore_journal.h

@@ -4,6 +4,7 @@
 #pragma once
 
 #include "crc32c.h"
+#include <set>
 
 #define MIN_JOURNAL_SIZE 4*1024*1024
 #define JOURNAL_MAGIC 0x4A33
@@ -145,8 +146,21 @@ struct journal_sector_info_t
     uint64_t flush_count;
     bool written;
     bool dirty;
+    uint64_t submit_id;
 };
 
+struct pending_journaling_t
+{
+    uint64_t flush_id;
+    int sector;
+    blockstore_op_t *op;
+};
+
+inline bool operator < (const pending_journaling_t & a, const pending_journaling_t & b)
+{
+    return a.flush_id < b.flush_id || a.flush_id == b.flush_id && a.op < b.op;
+}
+
 struct journal_t
 {
     int fd;
@@ -172,6 +186,9 @@ struct journal_t
     bool no_same_sector_overwrites = false;
     int cur_sector = 0;
     int in_sector_pos = 0;
+    std::vector<int> submitting_sectors;
+    std::set<pending_journaling_t> flushing_ops;
+    uint64_t submit_id = 0;
 
     // Used sector map
     // May use ~ 80 MB per 1 GB of used journal space in the worst case
@@ -200,5 +217,3 @@ struct blockstore_journal_check_t
 };
 
 journal_entry* prefill_single_journal_entry(journal_t & journal, uint16_t type, uint32_t size);
-
-void prepare_journal_sector_write(journal_t & journal, int sector, io_uring_sqe *sqe, std::function<void(ring_data_t*)> cb);

src/blockstore_open.cpp

@@ -295,9 +295,9 @@ void blockstore_impl_t::calc_lengths()
     }
 }
 
-void check_size(int fd, uint64_t *size, std::string name)
+static void check_size(int fd, uint64_t *size, uint64_t *sectsize, std::string name)
 {
-    int sectsize;
+    int sect;
     struct stat st;
     if (fstat(fd, &st) < 0)
     {
@@ -306,14 +306,21 @@ void check_size(int fd, uint64_t *size, std::string name)
     if (S_ISREG(st.st_mode))
     {
         *size = st.st_size;
+        if (sectsize)
+        {
+            *sectsize = st.st_blksize;
+        }
     }
     else if (S_ISBLK(st.st_mode))
     {
-        if (ioctl(fd, BLKSSZGET, &sectsize) < 0 ||
-            ioctl(fd, BLKGETSIZE64, size) < 0 ||
-            sectsize != 512)
+        if (ioctl(fd, BLKGETSIZE64, size) < 0 ||
+            ioctl(fd, BLKSSZGET, &sect) < 0)
         {
-            throw std::runtime_error(name+" sector is not equal to 512 bytes");
+            throw std::runtime_error("failed to get "+name+" size or block size: "+strerror(errno));
+        }
+        if (sectsize)
+        {
+            *sectsize = sect;
         }
     }
     else
@@ -329,7 +336,14 @@ void blockstore_impl_t::open_data()
     {
         throw std::runtime_error("Failed to open data device");
     }
-    check_size(data_fd, &data_size, "data device");
+    check_size(data_fd, &data_size, &data_device_sect, "data device");
+    if (disk_alignment % data_device_sect)
+    {
+        throw std::runtime_error(
+            "disk_alignment ("+std::to_string(disk_alignment)+
+            ") is not a multiple of data device sector size ("+std::to_string(data_device_sect)+")"
+        );
+    }
     if (data_offset >= data_size)
     {
         throw std::runtime_error("data_offset exceeds device size = "+std::to_string(data_size));
@@ -350,7 +364,7 @@ void blockstore_impl_t::open_meta()
         {
             throw std::runtime_error("Failed to open metadata device");
         }
-        check_size(meta_fd, &meta_size, "metadata device");
+        check_size(meta_fd, &meta_size, &meta_device_sect, "metadata device");
         if (meta_offset >= meta_size)
         {
             throw std::runtime_error("meta_offset exceeds device size = "+std::to_string(meta_size));
@@ -363,12 +377,20 @@ void blockstore_impl_t::open_meta()
     else
     {
         meta_fd = data_fd;
+        meta_device_sect = data_device_sect;
         meta_size = 0;
         if (meta_offset >= data_size)
         {
             throw std::runtime_error("meta_offset exceeds device size = "+std::to_string(data_size));
         }
     }
+    if (meta_block_size % meta_device_sect)
+    {
+        throw std::runtime_error(
+            "meta_block_size ("+std::to_string(meta_block_size)+
+            ") is not a multiple of data device sector size ("+std::to_string(meta_device_sect)+")"
+        );
+    }
 }
 
 void blockstore_impl_t::open_journal()
@@ -380,7 +402,7 @@ void blockstore_impl_t::open_journal()
         {
             throw std::runtime_error("Failed to open journal device");
         }
-        check_size(journal.fd, &journal.device_size, "journal device");
+        check_size(journal.fd, &journal.device_size, &journal_device_sect, "journal device");
         if (!disable_flock && flock(journal.fd, LOCK_EX|LOCK_NB) != 0)
         {
             throw std::runtime_error(std::string("Failed to lock journal device: ") + strerror(errno));
@@ -389,6 +411,7 @@ void blockstore_impl_t::open_journal()
     else
     {
         journal.fd = meta_fd;
+        journal_device_sect = meta_device_sect;
         journal.device_size = 0;
         if (journal.offset >= data_size)
         {
@@ -406,4 +429,11 @@ void blockstore_impl_t::open_journal()
         if (!journal.sector_buf)
             throw std::bad_alloc();
     }
+    if (journal_block_size % journal_device_sect)
+    {
+        throw std::runtime_error(
+            "journal_block_size ("+std::to_string(journal_block_size)+
+            ") is not a multiple of journal device sector size ("+std::to_string(journal_device_sect)+")"
+        );
+    }
 }

src/blockstore_read.cpp

@@ -24,7 +24,7 @@ int blockstore_impl_t::fulfill_read_push(blockstore_op_t *op, void *buf, uint64_
     }
     if (journal.inmemory && IS_JOURNAL(item_state))
     {
-        memcpy(buf, journal.buffer + offset, len);
+        memcpy(buf, (uint8_t*)journal.buffer + offset, len);
         return 1;
     }
     BS_SUBMIT_GET_SQE(sqe, data);
@@ -75,7 +75,7 @@ int blockstore_impl_t::fulfill_read(blockstore_op_t *read_op, uint64_t &fulfille
                 };
                 it = PRIV(read_op)->read_vec.insert(it, el);
                 if (!fulfill_read_push(read_op,
-                    read_op->buf + el.offset - read_op->offset,
+                    (uint8_t*)read_op->buf + el.offset - read_op->offset,
                     item_location + el.offset - item_start,
                     el.len, item_state, item_version))
                 {
@@ -102,7 +102,7 @@ uint8_t* blockstore_impl_t::get_clean_entry_bitmap(uint64_t block_loc, int offse
     {
         uint64_t sector = (meta_loc / (meta_block_size / clean_entry_size)) * meta_block_size;
         uint64_t pos = (meta_loc % (meta_block_size / clean_entry_size));
-        clean_entry_bitmap = (uint8_t*)(metadata_buffer + sector + pos*clean_entry_size + sizeof(clean_disk_entry) + offset);
+        clean_entry_bitmap = ((uint8_t*)metadata_buffer + sector + pos*clean_entry_size + sizeof(clean_disk_entry) + offset);
     }
     else
         clean_entry_bitmap = (uint8_t*)(clean_bitmap + meta_loc*2*clean_entry_bitmap_size + offset);

src/blockstore_rollback.cpp

@@ -74,24 +74,17 @@ skip_ov:
     {
         return 0;
     }
-    // There is sufficient space. Get SQEs
-    struct io_uring_sqe *sqe[space_check.sectors_to_write];
-    for (i = 0; i < space_check.sectors_to_write; i++)
-    {
-        BS_SUBMIT_GET_SQE_DECL(sqe[i]);
-    }
+    // There is sufficient space. Check SQEs
+    BS_SUBMIT_CHECK_SQES(space_check.sectors_to_write);
     // Prepare and submit journal entries
-    auto cb = [this, op](ring_data_t *data) { handle_rollback_event(data, op); };
-    int s = 0, cur_sector = -1;
+    int s = 0;
     for (i = 0, v = (obj_ver_id*)op->buf; i < op->len; i++, v++)
     {
         if (!journal.entry_fits(sizeof(journal_entry_rollback)) &&
             journal.sector_info[journal.cur_sector].dirty)
         {
-            if (cur_sector == -1)
-                PRIV(op)->min_flushed_journal_sector = 1 + journal.cur_sector;
-            prepare_journal_sector_write(journal, journal.cur_sector, sqe[s++], cb);
-            cur_sector = journal.cur_sector;
+            prepare_journal_sector_write(journal.cur_sector, op);
+            s++;
         }
         journal_entry_rollback *je = (journal_entry_rollback*)
             prefill_single_journal_entry(journal, JE_ROLLBACK, sizeof(journal_entry_rollback));
@@ -100,12 +93,9 @@ skip_ov:
         je->crc32 = je_crc32((journal_entry*)je);
         journal.crc32_last = je->crc32;
     }
-    prepare_journal_sector_write(journal, journal.cur_sector, sqe[s++], cb);
+    prepare_journal_sector_write(journal.cur_sector, op);
+    s++;
     assert(s == space_check.sectors_to_write);
-    if (cur_sector == -1)
-        PRIV(op)->min_flushed_journal_sector = 1 + journal.cur_sector;
-    PRIV(op)->max_flushed_journal_sector = 1 + journal.cur_sector;
-    PRIV(op)->pending_ops = s;
     PRIV(op)->op_state = 1;
     return 1;
 }
@@ -114,30 +104,23 @@ int blockstore_impl_t::continue_rollback(blockstore_op_t *op)
 {
     if (PRIV(op)->op_state == 2)
         goto resume_2;
-    else if (PRIV(op)->op_state == 3)
-        goto resume_3;
-    else if (PRIV(op)->op_state == 5)
-        goto resume_5;
+    else if (PRIV(op)->op_state == 4)
+        goto resume_4;
     else
         return 1;
 resume_2:
-    // Release used journal sectors
-    release_journal_sectors(op);
-resume_3:
     if (!disable_journal_fsync)
     {
-        io_uring_sqe *sqe;
-        BS_SUBMIT_GET_SQE_DECL(sqe);
-        ring_data_t *data = ((ring_data_t*)sqe->user_data);
+        BS_SUBMIT_GET_SQE(sqe, data);
         my_uring_prep_fsync(sqe, journal.fd, IORING_FSYNC_DATASYNC);
         data->iov = { 0 };
-        data->callback = [this, op](ring_data_t *data) { handle_rollback_event(data, op); };
+        data->callback = [this, op](ring_data_t *data) { handle_write_event(data, op); };
         PRIV(op)->min_flushed_journal_sector = PRIV(op)->max_flushed_journal_sector = 0;
         PRIV(op)->pending_ops = 1;
-        PRIV(op)->op_state = 4;
+        PRIV(op)->op_state = 3;
         return 1;
     }
-resume_5:
+resume_4:
     obj_ver_id* v;
     int i;
     for (i = 0, v = (obj_ver_id*)op->buf; i < op->len; i++, v++)
@@ -196,24 +179,6 @@ void blockstore_impl_t::mark_rolled_back(const obj_ver_id & ov)
     }
 }
 
-void blockstore_impl_t::handle_rollback_event(ring_data_t *data, blockstore_op_t *op)
-{
-    live = true;
-    if (data->res != data->iov.iov_len)
-    {
-        throw std::runtime_error(
-            "write operation failed ("+std::to_string(data->res)+" != "+std::to_string(data->iov.iov_len)+
-            "). in-memory state is corrupted. AAAAAAAaaaaaaaaa!!!111"
-        );
-    }
-    PRIV(op)->pending_ops--;
-    if (PRIV(op)->pending_ops == 0)
-    {
-        PRIV(op)->op_state++;
-        ringloop->wakeup();
-    }
-}
-
 void blockstore_impl_t::erase_dirty(blockstore_dirty_db_t::iterator dirty_start, blockstore_dirty_db_t::iterator dirty_end, uint64_t clean_loc)
 {
     if (dirty_end == dirty_start)

src/blockstore_stable.cpp

@@ -97,25 +97,18 @@ int blockstore_impl_t::dequeue_stable(blockstore_op_t *op)
     {
         return 0;
     }
-    // There is sufficient space. Get SQEs
-    struct io_uring_sqe *sqe[space_check.sectors_to_write];
-    for (i = 0; i < space_check.sectors_to_write; i++)
-    {
-        BS_SUBMIT_GET_SQE_DECL(sqe[i]);
-    }
+    // There is sufficient space. Check SQEs
+    BS_SUBMIT_CHECK_SQES(space_check.sectors_to_write);
     // Prepare and submit journal entries
-    auto cb = [this, op](ring_data_t *data) { handle_stable_event(data, op); };
-    int s = 0, cur_sector = -1;
+    int s = 0;
     for (i = 0, v = (obj_ver_id*)op->buf; i < op->len; i++, v++)
     {
         // FIXME: Only stabilize versions that aren't stable yet
         if (!journal.entry_fits(sizeof(journal_entry_stable)) &&
             journal.sector_info[journal.cur_sector].dirty)
         {
-            if (cur_sector == -1)
-                PRIV(op)->min_flushed_journal_sector = 1 + journal.cur_sector;
-            prepare_journal_sector_write(journal, journal.cur_sector, sqe[s++], cb);
-            cur_sector = journal.cur_sector;
+            prepare_journal_sector_write(journal.cur_sector, op);
+            s++;
         }
         journal_entry_stable *je = (journal_entry_stable*)
             prefill_single_journal_entry(journal, JE_STABLE, sizeof(journal_entry_stable));
@@ -124,12 +117,9 @@ int blockstore_impl_t::dequeue_stable(blockstore_op_t *op)
         je->crc32 = je_crc32((journal_entry*)je);
         journal.crc32_last = je->crc32;
     }
-    prepare_journal_sector_write(journal, journal.cur_sector, sqe[s++], cb);
+    prepare_journal_sector_write(journal.cur_sector, op);
+    s++;
     assert(s == space_check.sectors_to_write);
-    if (cur_sector == -1)
-        PRIV(op)->min_flushed_journal_sector = 1 + journal.cur_sector;
-    PRIV(op)->max_flushed_journal_sector = 1 + journal.cur_sector;
-    PRIV(op)->pending_ops = s;
     PRIV(op)->op_state = 1;
     return 1;
 }
@@ -138,30 +128,23 @@ int blockstore_impl_t::continue_stable(blockstore_op_t *op)
 {
     if (PRIV(op)->op_state == 2)
         goto resume_2;
-    else if (PRIV(op)->op_state == 3)
-        goto resume_3;
-    else if (PRIV(op)->op_state == 5)
-        goto resume_5;
+    else if (PRIV(op)->op_state == 4)
+        goto resume_4;
     else
         return 1;
 resume_2:
-    // Release used journal sectors
-    release_journal_sectors(op);
-resume_3:
     if (!disable_journal_fsync)
     {
-        io_uring_sqe *sqe;
-        BS_SUBMIT_GET_SQE_DECL(sqe);
-        ring_data_t *data = ((ring_data_t*)sqe->user_data);
+        BS_SUBMIT_GET_SQE(sqe, data);
         my_uring_prep_fsync(sqe, journal.fd, IORING_FSYNC_DATASYNC);
         data->iov = { 0 };
-        data->callback = [this, op](ring_data_t *data) { handle_stable_event(data, op); };
+        data->callback = [this, op](ring_data_t *data) { handle_write_event(data, op); };
         PRIV(op)->min_flushed_journal_sector = PRIV(op)->max_flushed_journal_sector = 0;
         PRIV(op)->pending_ops = 1;
-        PRIV(op)->op_state = 4;
+        PRIV(op)->op_state = 3;
         return 1;
     }
-resume_5:
+resume_4:
     // Mark dirty_db entries as stable, acknowledge op completion
     obj_ver_id* v;
     int i;
@@ -257,21 +240,3 @@ void blockstore_impl_t::mark_stable(const obj_ver_id & v, bool forget_dirty)
         unstable_writes.erase(unstab_it);
     }
 }
-
-void blockstore_impl_t::handle_stable_event(ring_data_t *data, blockstore_op_t *op)
-{
-    live = true;
-    if (data->res != data->iov.iov_len)
-    {
-        throw std::runtime_error(
-            "write operation failed ("+std::to_string(data->res)+" != "+std::to_string(data->iov.iov_len)+
-            "). in-memory state is corrupted. AAAAAAAaaaaaaaaa!!!111"
-        );
-    }
-    PRIV(op)->pending_ops--;
-    if (PRIV(op)->pending_ops == 0)
-    {
-        PRIV(op)->op_state++;
-        ringloop->wakeup();
-    }
-}

src/blockstore_sync.cpp

@@ -44,10 +44,8 @@ int blockstore_impl_t::continue_sync(blockstore_op_t *op, bool queue_has_in_prog
         if (journal.sector_info[journal.cur_sector].dirty)
         {
             // Write out the last journal sector if it happens to be dirty
-            BS_SUBMIT_GET_ONLY_SQE(sqe);
-            prepare_journal_sector_write(journal, journal.cur_sector, sqe, [this, op](ring_data_t *data) { handle_sync_event(data, op); });
-            PRIV(op)->min_flushed_journal_sector = PRIV(op)->max_flushed_journal_sector = 1 + journal.cur_sector;
-            PRIV(op)->pending_ops = 1;
+            BS_SUBMIT_CHECK_SQES(1);
+            prepare_journal_sector_write(journal.cur_sector, op);
             PRIV(op)->op_state = SYNC_JOURNAL_WRITE_SENT;
             return 1;
         }
@@ -64,7 +62,7 @@ int blockstore_impl_t::continue_sync(blockstore_op_t *op, bool queue_has_in_prog
             BS_SUBMIT_GET_SQE(sqe, data);
             my_uring_prep_fsync(sqe, data_fd, IORING_FSYNC_DATASYNC);
             data->iov = { 0 };
-            data->callback = [this, op](ring_data_t *data) { handle_sync_event(data, op); };
+            data->callback = [this, op](ring_data_t *data) { handle_write_event(data, op); };
             PRIV(op)->min_flushed_journal_sector = PRIV(op)->max_flushed_journal_sector = 0;
             PRIV(op)->pending_ops = 1;
             PRIV(op)->op_state = SYNC_DATA_SYNC_SENT;
@@ -85,24 +83,18 @@ int blockstore_impl_t::continue_sync(blockstore_op_t *op, bool queue_has_in_prog
         {
             return 0;
         }
-        // Get SQEs. Don't bother about merging, submit each journal sector as a separate request
-        struct io_uring_sqe *sqe[space_check.sectors_to_write];
-        for (int i = 0; i < space_check.sectors_to_write; i++)
-        {
-            BS_SUBMIT_GET_SQE_DECL(sqe[i]);
-        }
+        // Check SQEs. Don't bother about merging, submit each journal sector as a separate request
+        BS_SUBMIT_CHECK_SQES(space_check.sectors_to_write);
         // Prepare and submit journal entries
         auto it = PRIV(op)->sync_big_writes.begin();
-        int s = 0, cur_sector = -1;
+        int s = 0;
         while (it != PRIV(op)->sync_big_writes.end())
         {
             if (!journal.entry_fits(sizeof(journal_entry_big_write) + clean_entry_bitmap_size) &&
                 journal.sector_info[journal.cur_sector].dirty)
             {
-                if (cur_sector == -1)
-                    PRIV(op)->min_flushed_journal_sector = 1 + journal.cur_sector;
-                prepare_journal_sector_write(journal, journal.cur_sector, sqe[s++], [this, op](ring_data_t *data) { handle_sync_event(data, op); });
-                cur_sector = journal.cur_sector;
+                prepare_journal_sector_write(journal.cur_sector, op);
+                s++;
             }
             auto & dirty_entry = dirty_db.at(*it);
             journal_entry_big_write *je = (journal_entry_big_write*)prefill_single_journal_entry(
@@ -129,12 +121,9 @@ int blockstore_impl_t::continue_sync(blockstore_op_t *op, bool queue_has_in_prog
             journal.crc32_last = je->crc32;
             it++;
         }
-        prepare_journal_sector_write(journal, journal.cur_sector, sqe[s++], [this, op](ring_data_t *data) { handle_sync_event(data, op); });
+        prepare_journal_sector_write(journal.cur_sector, op);
+        s++;
         assert(s == space_check.sectors_to_write);
-        if (cur_sector == -1)
-            PRIV(op)->min_flushed_journal_sector = 1 + journal.cur_sector;
-        PRIV(op)->max_flushed_journal_sector = 1 + journal.cur_sector;
-        PRIV(op)->pending_ops = s;
         PRIV(op)->op_state = SYNC_JOURNAL_WRITE_SENT;
         return 1;
     }
@@ -145,7 +134,7 @@ int blockstore_impl_t::continue_sync(blockstore_op_t *op, bool queue_has_in_prog
             BS_SUBMIT_GET_SQE(sqe, data);
             my_uring_prep_fsync(sqe, journal.fd, IORING_FSYNC_DATASYNC);
             data->iov = { 0 };
-            data->callback = [this, op](ring_data_t *data) { handle_sync_event(data, op); };
+            data->callback = [this, op](ring_data_t *data) { handle_write_event(data, op); };
             PRIV(op)->min_flushed_journal_sector = PRIV(op)->max_flushed_journal_sector = 0;
             PRIV(op)->pending_ops = 1;
             PRIV(op)->op_state = SYNC_JOURNAL_SYNC_SENT;
@@ -164,42 +153,6 @@ int blockstore_impl_t::continue_sync(blockstore_op_t *op, bool queue_has_in_prog
     return 1;
 }
 
-void blockstore_impl_t::handle_sync_event(ring_data_t *data, blockstore_op_t *op)
-{
-    live = true;
-    if (data->res != data->iov.iov_len)
-    {
-        throw std::runtime_error(
-            "write operation failed ("+std::to_string(data->res)+" != "+std::to_string(data->iov.iov_len)+
-            "). in-memory state is corrupted. AAAAAAAaaaaaaaaa!!!111"
-        );
-    }
-    PRIV(op)->pending_ops--;
-    if (PRIV(op)->pending_ops == 0)
-    {
-        // Release used journal sectors
-        release_journal_sectors(op);
-        // Handle states
-        if (PRIV(op)->op_state == SYNC_DATA_SYNC_SENT)
-        {
-            PRIV(op)->op_state = SYNC_DATA_SYNC_DONE;
-        }
-        else if (PRIV(op)->op_state == SYNC_JOURNAL_WRITE_SENT)
-        {
-            PRIV(op)->op_state = SYNC_JOURNAL_WRITE_DONE;
-        }
-        else if (PRIV(op)->op_state == SYNC_JOURNAL_SYNC_SENT)
-        {
-            PRIV(op)->op_state = SYNC_DONE;
-        }
-        else
-        {
-            throw std::runtime_error("BUG: unexpected sync op state");
-        }
-        ringloop->wakeup();
-    }
-}
-
 void blockstore_impl_t::ack_sync(blockstore_op_t *op)
 {
     // Handle states

src/blockstore_write.cpp

@@ -102,7 +102,7 @@ bool blockstore_impl_t::enqueue_write(blockstore_op_t *op)
         // Issue an additional sync so that the previous big write can reach the journal
         blockstore_op_t *sync_op = new blockstore_op_t;
         sync_op->opcode = BS_OP_SYNC;
-        sync_op->callback = [this, op](blockstore_op_t *sync_op)
+        sync_op->callback = [](blockstore_op_t *sync_op)
         {
             delete sync_op;
         };
@@ -268,8 +268,8 @@ int blockstore_impl_t::dequeue_write(blockstore_op_t *op)
             cancel_all_writes(op, dirty_it, -ENOSPC);
             return 2;
         }
-        write_iodepth++;
         BS_SUBMIT_GET_SQE(sqe, data);
+        write_iodepth++;
         dirty_it->second.location = loc << block_order;
         dirty_it->second.state = (dirty_it->second.state & ~BS_ST_WORKFLOW_MASK) | BS_ST_SUBMITTED;
 #ifdef BLOCKSTORE_DEBUG
@@ -324,29 +324,21 @@ int blockstore_impl_t::dequeue_write(blockstore_op_t *op)
         {
             return 0;
         }
-        write_iodepth++;
-        // There is sufficient space. Get SQE(s)
-        struct io_uring_sqe *sqe1 = NULL;
-        if (immediate_commit != IMMEDIATE_NONE ||
-            !journal.entry_fits(sizeof(journal_entry_small_write) + clean_entry_bitmap_size))
-        {
+        // There is sufficient space. Check SQE(s)
+        BS_SUBMIT_CHECK_SQES(
             // Write current journal sector only if it's dirty and full, or in the immediate_commit mode
-            BS_SUBMIT_GET_SQE_DECL(sqe1);
-        }
-        struct io_uring_sqe *sqe2 = NULL;
-        if (op->len > 0)
-        {
-            BS_SUBMIT_GET_SQE_DECL(sqe2);
-        }
+            (immediate_commit != IMMEDIATE_NONE ||
+                !journal.entry_fits(sizeof(journal_entry_small_write) + clean_entry_bitmap_size) ? 1 : 0) +
+            (op->len > 0 ? 1 : 0)
+        );
+        write_iodepth++;
         // Got SQEs. Prepare previous journal sector write if required
         auto cb = [this, op](ring_data_t *data) { handle_write_event(data, op); };
         if (immediate_commit == IMMEDIATE_NONE)
         {
-            if (sqe1)
+            if (!journal.entry_fits(sizeof(journal_entry_small_write) + clean_entry_bitmap_size))
             {
-                prepare_journal_sector_write(journal, journal.cur_sector, sqe1, cb);
-                PRIV(op)->min_flushed_journal_sector = PRIV(op)->max_flushed_journal_sector = 1 + journal.cur_sector;
-                PRIV(op)->pending_ops++;
+                prepare_journal_sector_write(journal.cur_sector, op);
             }
             else
             {
@@ -380,9 +372,7 @@ int blockstore_impl_t::dequeue_write(blockstore_op_t *op)
         journal.crc32_last = je->crc32;
         if (immediate_commit != IMMEDIATE_NONE)
         {
-            prepare_journal_sector_write(journal, journal.cur_sector, sqe1, cb);
-            PRIV(op)->min_flushed_journal_sector = PRIV(op)->max_flushed_journal_sector = 1 + journal.cur_sector;
-            PRIV(op)->pending_ops++;
+            prepare_journal_sector_write(journal.cur_sector, op);
         }
         if (op->len > 0)
         {
@@ -390,9 +380,9 @@ int blockstore_impl_t::dequeue_write(blockstore_op_t *op)
             if (journal.inmemory)
             {
                 // Copy data
-                memcpy(journal.buffer + journal.next_free, op->buf, op->len);
+                memcpy((uint8_t*)journal.buffer + journal.next_free, op->buf, op->len);
             }
-            ring_data_t *data2 = ((ring_data_t*)sqe2->user_data);
+            BS_SUBMIT_GET_SQE(sqe2, data2);
             data2->iov = (struct iovec){ op->buf, op->len };
             data2->callback = cb;
             my_uring_prep_writev(
@@ -441,13 +431,12 @@ int blockstore_impl_t::continue_write(blockstore_op_t *op)
 resume_2:
     // Only for the immediate_commit mode: prepare and submit big_write journal entry
     {
+        BS_SUBMIT_CHECK_SQES(1);
         auto dirty_it = dirty_db.find((obj_ver_id){
             .oid = op->oid,
             .version = op->version,
         });
         assert(dirty_it != dirty_db.end());
-        io_uring_sqe *sqe = NULL;
-        BS_SUBMIT_GET_SQE_DECL(sqe);
         journal_entry_big_write *je = (journal_entry_big_write*)prefill_single_journal_entry(
             journal, op->opcode == BS_OP_WRITE_STABLE ? JE_BIG_WRITE_INSTANT : JE_BIG_WRITE,
             sizeof(journal_entry_big_write) + clean_entry_bitmap_size
@@ -469,10 +458,7 @@ resume_2:
         memcpy((void*)(je+1), (clean_entry_bitmap_size > sizeof(void*) ? dirty_it->second.bitmap : &dirty_it->second.bitmap), clean_entry_bitmap_size);
         je->crc32 = je_crc32((journal_entry*)je);
         journal.crc32_last = je->crc32;
-        prepare_journal_sector_write(journal, journal.cur_sector, sqe,
-            [this, op](ring_data_t *data) { handle_write_event(data, op); });
-        PRIV(op)->min_flushed_journal_sector = PRIV(op)->max_flushed_journal_sector = 1 + journal.cur_sector;
-        PRIV(op)->pending_ops = 1;
+        prepare_journal_sector_write(journal.cur_sector, op);
         PRIV(op)->op_state = 3;
         return 1;
     }
@@ -587,6 +573,7 @@ void blockstore_impl_t::handle_write_event(ring_data_t *data, blockstore_op_t *o
         );
     }
     PRIV(op)->pending_ops--;
+    assert(PRIV(op)->pending_ops >= 0);
     if (PRIV(op)->pending_ops == 0)
     {
         release_journal_sectors(op);
@@ -604,7 +591,6 @@ void blockstore_impl_t::release_journal_sectors(blockstore_op_t *op)
         uint64_t s = PRIV(op)->min_flushed_journal_sector;
         while (1)
         {
-            journal.sector_info[s-1].flush_count--;
             if (s != (1+journal.cur_sector) && journal.sector_info[s-1].flush_count == 0)
             {
                 // We know for sure that we won't write into this sector anymore
@@ -643,24 +629,24 @@ int blockstore_impl_t::dequeue_del(blockstore_op_t *op)
     {
         return 0;
     }
-    write_iodepth++;
-    io_uring_sqe *sqe = NULL;
-    if (immediate_commit != IMMEDIATE_NONE ||
-        (journal_block_size - journal.in_sector_pos) < sizeof(journal_entry_del) &&
-        journal.sector_info[journal.cur_sector].dirty)
+    // Write current journal sector only if it's dirty and full, or in the immediate_commit mode
+    BS_SUBMIT_CHECK_SQES(
+        (immediate_commit != IMMEDIATE_NONE ||
+            (journal_block_size - journal.in_sector_pos) < sizeof(journal_entry_del) &&
+            journal.sector_info[journal.cur_sector].dirty) ? 1 : 0
+    );
+    if (write_iodepth >= max_write_iodepth)
     {
-        // Write current journal sector only if it's dirty and full, or in the immediate_commit mode
-        BS_SUBMIT_GET_SQE_DECL(sqe);
+        return 0;
     }
-    auto cb = [this, op](ring_data_t *data) { handle_write_event(data, op); };
+    write_iodepth++;
     // Prepare journal sector write
     if (immediate_commit == IMMEDIATE_NONE)
     {
-        if (sqe)
+        if ((journal_block_size - journal.in_sector_pos) < sizeof(journal_entry_del) &&
+            journal.sector_info[journal.cur_sector].dirty)
         {
-            prepare_journal_sector_write(journal, journal.cur_sector, sqe, cb);
-            PRIV(op)->min_flushed_journal_sector = PRIV(op)->max_flushed_journal_sector = 1 + journal.cur_sector;
-            PRIV(op)->pending_ops++;
+            prepare_journal_sector_write(journal.cur_sector, op);
         }
         else
         {
@@ -687,9 +673,7 @@ int blockstore_impl_t::dequeue_del(blockstore_op_t *op)
     dirty_it->second.state = BS_ST_DELETE | BS_ST_SUBMITTED;
     if (immediate_commit != IMMEDIATE_NONE)
     {
-        prepare_journal_sector_write(journal, journal.cur_sector, sqe, cb);
-        PRIV(op)->min_flushed_journal_sector = PRIV(op)->max_flushed_journal_sector = 1 + journal.cur_sector;
-        PRIV(op)->pending_ops++;
+        prepare_journal_sector_write(journal.cur_sector, op);
     }
     if (!PRIV(op)->pending_ops)
     {

src/cli.cpp

@@ -177,7 +177,7 @@ void cli_tool_t::change_parent(inode_t cur, inode_t new_parent)
     new_cfg.parent_id = new_parent;
     json11::Json::object cur_cfg_json = cli->st_cli.serialize_inode_cfg(&new_cfg);
     waiting++;
-    cli->st_cli.etcd_txn(json11::Json::object {
+    cli->st_cli.etcd_txn_slow(json11::Json::object {
         { "compare", json11::Json::array {
             json11::Json::object {
                 { "target", "MOD" },
@@ -194,7 +194,7 @@ void cli_tool_t::change_parent(inode_t cur, inode_t new_parent)
                 } }
             },
         } },
-    }, ETCD_SLOW_TIMEOUT, [this, new_parent, cur, cur_name](std::string err, json11::Json res)
+    }, [this, new_parent, cur, cur_name](std::string err, json11::Json res)
     {
         if (err != "")
         {
@@ -229,6 +229,22 @@ void cli_tool_t::change_parent(inode_t cur, inode_t new_parent)
     });
 }
 
+void cli_tool_t::etcd_txn(json11::Json txn)
+{
+    waiting++;
+    cli->st_cli.etcd_txn_slow(txn, [this](std::string err, json11::Json res)
+    {
+        waiting--;
+        if (err != "")
+        {
+            fprintf(stderr, "Error reading from etcd: %s\n", err.c_str());
+            exit(1);
+        }
+        etcd_result = res;
+        ringloop->wakeup();
+    });
+}
+
 inode_config_t* cli_tool_t::get_inode_cfg(const std::string & name)
 {
     for (auto & ic: cli->st_cli.inode_config)
@@ -305,6 +321,10 @@ void cli_tool_t::run(json11::Json cfg)
         fprintf(stderr, "unknown command: %s\n", cmd[0].string_value().c_str());
         exit(1);
     }
+    if (action_cb == NULL)
+    {
+        return;
+    }
     color = !cfg["no-color"].bool_value();
     json_output = cfg["json"].bool_value();
     iodepth = cfg["iodepth"].uint64_value();

src/cli.h

@@ -34,6 +34,7 @@ public:
     cluster_client_t *cli = NULL;
 
     int waiting = 0;
+    json11::Json etcd_result;
     ring_consumer_t consumer;
     std::function<bool(void)> action_cb;
 
@@ -60,6 +61,8 @@ public:
     std::function<bool(void)> start_snap_rm(json11::Json);
     std::function<bool(void)> start_alloc_osd(json11::Json cfg, uint64_t *out = NULL);
     std::function<bool(void)> simple_offsets(json11::Json cfg);
+
+    void etcd_txn(json11::Json txn);
 };
 
 uint64_t parse_size(std::string size_str);

src/cli_alloc_osd.cpp

@@ -13,7 +13,6 @@ struct alloc_osd_t
 {
     cli_tool_t *parent;
 
-    json11::Json result;
     uint64_t new_id = 1;
 
     int state = 0;
@@ -29,7 +28,7 @@ struct alloc_osd_t
             goto resume_1;
         do
         {
-            etcd_txn(json11::Json::object {
+            parent->etcd_txn(json11::Json::object {
                 { "compare", json11::Json::array {
                     json11::Json::object {
                         { "target", "VERSION" },
@@ -63,10 +62,10 @@ struct alloc_osd_t
             state = 1;
             if (parent->waiting > 0)
                 return;
-            if (!result["succeeded"].bool_value())
+            if (!parent->etcd_result["succeeded"].bool_value())
             {
                 std::vector<osd_num_t> used;
-                for (auto kv: result["responses"][0]["response_range"]["kvs"].array_items())
+                for (auto kv: parent->etcd_result["responses"][0]["response_range"]["kvs"].array_items())
                 {
                     std::string key = base64_decode(kv["key"].string_value());
                     osd_num_t cur_osd;
@@ -98,25 +97,9 @@ struct alloc_osd_t
                     new_id = used[e-1]+1;
                 }
             }
-        } while (!result["succeeded"].bool_value());
+        } while (!parent->etcd_result["succeeded"].bool_value());
         state = 100;
     }
-
-    void etcd_txn(json11::Json txn)
-    {
-        parent->waiting++;
-        parent->cli->st_cli.etcd_txn(txn, ETCD_SLOW_TIMEOUT, [this](std::string err, json11::Json res)
-        {
-            parent->waiting--;
-            if (err != "")
-            {
-                fprintf(stderr, "Error reading from etcd: %s\n", err.c_str());
-                exit(1);
-            }
-            this->result = res;
-            parent->ringloop->wakeup();
-        });
-    }
 };
 
 std::function<bool(void)> cli_tool_t::start_alloc_osd(json11::Json cfg, uint64_t *out)

src/cli_create.cpp

@@ -31,7 +31,6 @@ struct image_creator_t
     inode_t new_parent_id = 0;
     inode_t new_id = 0, old_id = 0;
     uint64_t max_id_mod_rev = 0, cfg_mod_rev = 0, idx_mod_rev = 0;
-    json11::Json result;
 
     int state = 0;
 
@@ -125,26 +124,26 @@ struct image_creator_t
         }
         do
         {
-            etcd_txn(json11::Json::object {
+            parent->etcd_txn(json11::Json::object {
                 { "success", json11::Json::array { get_next_id() } }
             });
             state = 2;
 resume_2:
             if (parent->waiting > 0)
                 return;
-            extract_next_id(result["responses"][0]);
+            extract_next_id(parent->etcd_result["responses"][0]);
             attempt_create();
             state = 3;
 resume_3:
             if (parent->waiting > 0)
                 return;
-            if (!result["succeeded"].bool_value() &&
-                result["responses"][0]["response_range"]["kvs"].array_items().size() > 0)
+            if (!parent->etcd_result["succeeded"].bool_value() &&
+                parent->etcd_result["responses"][0]["response_range"]["kvs"].array_items().size() > 0)
             {
                 fprintf(stderr, "Image %s already exists\n", image_name.c_str());
                 exit(1);
             }
-        } while (!result["succeeded"].bool_value());
+        } while (!parent->etcd_result["succeeded"].bool_value());
         if (parent->progress)
         {
             printf("Image %s created\n", image_name.c_str());
@@ -196,13 +195,13 @@ resume_3:
 resume_4:
             if (parent->waiting > 0)
                 return;
-            if (!result["succeeded"].bool_value() &&
-                result["responses"][0]["response_range"]["kvs"].array_items().size() > 0)
+            if (!parent->etcd_result["succeeded"].bool_value() &&
+                parent->etcd_result["responses"][0]["response_range"]["kvs"].array_items().size() > 0)
             {
                 fprintf(stderr, "Snapshot %s@%s already exists\n", image_name.c_str(), new_snap.c_str());
                 exit(1);
             }
-        } while (!result["succeeded"].bool_value());
+        } while (!parent->etcd_result["succeeded"].bool_value());
         if (parent->progress)
         {
             printf("Snapshot %s@%s created\n", image_name.c_str(), new_snap.c_str());
@@ -246,7 +245,7 @@ resume_4:
             goto resume_2;
         else if (state == 3)
             goto resume_3;
-        etcd_txn(json11::Json::object { { "success", json11::Json::array {
+        parent->etcd_txn(json11::Json::object { { "success", json11::Json::array {
             get_next_id(),
             json11::Json::object {
                 { "request_range", json11::Json::object {
@@ -260,11 +259,11 @@ resume_4:
 resume_2:
         if (parent->waiting > 0)
             return;
-        extract_next_id(result["responses"][0]);
+        extract_next_id(parent->etcd_result["responses"][0]);
         old_id = 0;
         old_pool_id = 0;
         cfg_mod_rev = idx_mod_rev = 0;
-        if (result["responses"][1]["response_range"]["kvs"].array_items().size() == 0)
+        if (parent->etcd_result["responses"][1]["response_range"]["kvs"].array_items().size() == 0)
         {
             for (auto & ic: parent->cli->st_cli.inode_config)
             {
@@ -283,7 +282,7 @@ resume_2:
         {
             // FIXME: Parse kvs in etcd_state_client automatically
             {
-                auto kv = parent->cli->st_cli.parse_etcd_kv(result["responses"][1]["response_range"]["kvs"][0]);
+                auto kv = parent->cli->st_cli.parse_etcd_kv(parent->etcd_result["responses"][1]["response_range"]["kvs"][0]);
                 old_id = INODE_NO_POOL(kv.value["id"].uint64_value());
                 old_pool_id = (pool_id_t)kv.value["pool_id"].uint64_value();
                 idx_mod_rev = kv.mod_revision;
@@ -293,7 +292,7 @@ resume_2:
                     exit(1);
                 }
             }
-            etcd_txn(json11::Json::object {
+            parent->etcd_txn(json11::Json::object {
                 { "success", json11::Json::array {
                     json11::Json::object {
                         { "request_range", json11::Json::object {
@@ -310,7 +309,7 @@ resume_3:
             if (parent->waiting > 0)
                 return;
             {
-                auto kv = parent->cli->st_cli.parse_etcd_kv(result["responses"][0]["response_range"]["kvs"][0]);
+                auto kv = parent->cli->st_cli.parse_etcd_kv(parent->etcd_result["responses"][0]["response_range"]["kvs"][0]);
                 size = kv.value["size"].uint64_value();
                 new_parent_id = kv.value["parent_id"].uint64_value();
                 uint64_t parent_pool_id = kv.value["parent_pool_id"].uint64_value();
@@ -439,28 +438,12 @@ resume_3:
                 } },
             });
         };
-        etcd_txn(json11::Json::object {
+        parent->etcd_txn(json11::Json::object {
             { "compare", checks },
             { "success", success },
             { "failure", failure },
         });
     }
-
-    void etcd_txn(json11::Json txn)
-    {
-        parent->waiting++;
-        parent->cli->st_cli.etcd_txn(txn, ETCD_SLOW_TIMEOUT, [this](std::string err, json11::Json res)
-        {
-            parent->waiting--;
-            if (err != "")
-            {
-                fprintf(stderr, "Error reading from etcd: %s\n", err.c_str());
-                exit(1);
-            }
-            this->result = res;
-            parent->ringloop->wakeup();
-        });
-    }
 };
 
 uint64_t parse_size(std::string size_str)

src/cli_df.cpp

@@ -24,8 +24,7 @@ struct pool_lister_t
         if (state == 1)
             goto resume_1;
         // Space statistics - pool/stats/<pool>
-        parent->waiting++;
-        parent->cli->st_cli.etcd_txn(json11::Json::object {
+        parent->etcd_txn(json11::Json::object {
             { "success", json11::Json::array {
                 json11::Json::object {
                     { "request_range", json11::Json::object {
@@ -48,21 +47,12 @@ struct pool_lister_t
                     } },
                 },
             } },
-        }, ETCD_SLOW_TIMEOUT, [this](std::string err, json11::Json res)
-        {
-            parent->waiting--;
-            if (err != "")
-            {
-                fprintf(stderr, "Error reading from etcd: %s\n", err.c_str());
-                exit(1);
-            }
-            space_info = res;
-            parent->ringloop->wakeup();
         });
         state = 1;
 resume_1:
         if (parent->waiting > 0)
             return;
+        space_info = parent->etcd_result;
         std::map<pool_id_t, uint64_t> osd_free;
         for (auto & kv_item: space_info["responses"][0]["response_range"]["kvs"].array_items())
         {

src/cli_flatten.cpp

@@ -3,6 +3,7 @@
 
 #include "cli.h"
 #include "cluster_client.h"
+#include <sys/stat.h>
 
 // Flatten a layer: merge all parents into a layer and break the connection completely
 struct snap_flattener_t

src/cli_ls.cpp

@@ -84,8 +84,7 @@ struct image_lister_t
         // Space statistics
         // inode/stats/<pool>/<inode>::raw_used divided by pool/stats/<pool>::pg_real_size
         // multiplied by 1 or number of data drives
-        parent->waiting++;
-        parent->cli->st_cli.etcd_txn(json11::Json::object {
+        parent->etcd_txn(json11::Json::object {
             { "success", json11::Json::array {
                 json11::Json::object {
                     { "request_range", json11::Json::object {
@@ -112,21 +111,12 @@ struct image_lister_t
                     } },
                 },
             } },
-        }, ETCD_SLOW_TIMEOUT, [this](std::string err, json11::Json res)
-        {
-            parent->waiting--;
-            if (err != "")
-            {
-                fprintf(stderr, "Error reading from etcd: %s\n", err.c_str());
-                exit(1);
-            }
-            space_info = res;
-            parent->ringloop->wakeup();
         });
         state = 1;
 resume_1:
         if (parent->waiting > 0)
             return;
+        space_info = parent->etcd_result;
         std::map<pool_id_t, uint64_t> pool_pg_real_size;
         for (auto & kv_item: space_info["responses"][0]["response_range"]["kvs"].array_items())
         {

src/cli_merge.cpp

@@ -412,7 +412,7 @@ struct snap_merger_t
         uint64_t bitmap_size = target_block_size / gran;
         while (rwo->end < bitmap_size)
         {
-            auto bit = ((*(uint8_t*)(rwo->op.bitmap_buf + (rwo->end >> 3))) & (1 << (rwo->end & 0x7)));
+            auto bit = ((*((uint8_t*)rwo->op.bitmap_buf + (rwo->end >> 3))) & (1 << (rwo->end & 0x7)));
             if (!bit)
             {
                 if (rwo->end > rwo->start)
@@ -459,7 +459,7 @@ struct snap_merger_t
         subop->len = end-start;
         subop->version = version;
         subop->flags = OSD_OP_IGNORE_READONLY;
-        subop->iov.push_back(rwo->buf+start, end-start);
+        subop->iov.push_back((uint8_t*)rwo->buf+start, end-start);
         subop->callback = [this, rwo](cluster_op_t *subop)
         {
             rwo->todo--;
@@ -495,7 +495,7 @@ struct snap_merger_t
         subop->offset = offset;
         subop->len = 0;
         subop->flags = OSD_OP_IGNORE_READONLY;
-        subop->callback = [this](cluster_op_t *subop)
+        subop->callback = [](cluster_op_t *subop)
         {
             if (subop->retval != 0)
             {
@@ -519,10 +519,10 @@ struct snap_merger_t
                 deleted_unsynced++;
                 if (deleted_unsynced >= fsync_interval)
                 {
-                    uint64_t from = last_fsync_offset, to = last_written_offset;
+                    uint64_t to = last_written_offset;
                     cluster_op_t *subop = new cluster_op_t;
                     subop->opcode = OSD_OP_SYNC;
-                    subop->callback = [this, from, to](cluster_op_t *subop)
+                    subop->callback = [this, to](cluster_op_t *subop)
                     {
                         delete subop;
                         // We can now delete source data between <from> and <to>

src/cli_modify.cpp

@@ -170,29 +170,19 @@ resume_1:
                 } }
             });
         }
-        parent->waiting++;
-        parent->cli->st_cli.etcd_txn(json11::Json::object {
+        parent->etcd_txn(json11::Json::object {
             { "compare", checks },
             { "success", success },
-        }, ETCD_SLOW_TIMEOUT, [this](std::string err, json11::Json res)
-        {
-            if (err != "")
-            {
-                fprintf(stderr, "Error changing %s: %s\n", image_name.c_str(), err.c_str());
-                exit(1);
-            }
-            if (!res["succeeded"].bool_value())
-            {
-                fprintf(stderr, "Image %s was modified by someone else, please repeat your request\n", image_name.c_str());
-                exit(1);
-            }
-            parent->waiting--;
-            parent->ringloop->wakeup();
         });
         state = 2;
 resume_2:
         if (parent->waiting > 0)
             return;
+        if (!parent->etcd_result["succeeded"].bool_value())
+        {
+            fprintf(stderr, "Image %s was modified by someone else, please repeat your request\n", image_name.c_str());
+            exit(1);
+        }
         printf("Image %s modified\n", image_name.c_str());
         state = 100;
     }

src/cli_simple_offsets.cpp

@@ -8,6 +8,7 @@
 #include "cli.h"
 #include "cluster_client.h"
 #include "base64.h"
+#include <sys/stat.h>
 
 // Calculate offsets for a block device and print OSD command line parameters
 std::function<bool(void)> cli_tool_t::simple_offsets(json11::Json cfg)

src/cli_snap_rm.cpp

@@ -256,9 +256,9 @@ resume_9:
             });
         }
         parent->waiting++;
-        parent->cli->st_cli.etcd_txn(json11::Json::object {
+        parent->cli->st_cli.etcd_txn_slow(json11::Json::object {
             { "success", reads },
-        }, ETCD_SLOW_TIMEOUT, [this](std::string err, json11::Json data)
+        }, [this](std::string err, json11::Json data)
         {
             parent->waiting--;
             if (err != "")
@@ -414,10 +414,10 @@ resume_9:
             }
         }
         parent->waiting++;
-        parent->cli->st_cli.etcd_txn(json11::Json::object {
+        parent->cli->st_cli.etcd_txn_slow(json11::Json::object {
             { "compare", cmp },
             { "success", txn },
-        }, ETCD_SLOW_TIMEOUT, [this, target_name, child_name](std::string err, json11::Json res)
+        }, [this, target_name, child_name](std::string err, json11::Json res)
         {
             parent->waiting--;
             if (err != "")
@@ -454,7 +454,7 @@ resume_9:
             "/"+std::to_string(INODE_NO_POOL(cur))
         );
         parent->waiting++;
-        parent->cli->st_cli.etcd_txn(json11::Json::object {
+        parent->cli->st_cli.etcd_txn_slow(json11::Json::object {
             { "compare", json11::Json::array {
                 json11::Json::object {
                     { "target", "MOD" },
@@ -475,7 +475,7 @@ resume_9:
                     } },
                 },
             } },
-        }, ETCD_SLOW_TIMEOUT, [this, cur_name](std::string err, json11::Json res)
+        }, [this, cur_name](std::string err, json11::Json res)
         {
             parent->waiting--;
             if (err != "")

src/cluster_client.cpp

@@ -534,8 +534,8 @@ void cluster_client_t::copy_write(cluster_op_t *op, std::map<object_id, cluster_
             unsigned iov_len = (op->iov.buf[iov_idx].iov_len - iov_pos);
             if (iov_len <= cur_len)
             {
-                memcpy(dirty_it->second.buf + pos - dirty_it->first.stripe,
-                    op->iov.buf[iov_idx].iov_base + iov_pos, iov_len);
+                memcpy((uint8_t*)dirty_it->second.buf + pos - dirty_it->first.stripe,
+                    (uint8_t*)op->iov.buf[iov_idx].iov_base + iov_pos, iov_len);
                 pos += iov_len;
                 len -= iov_len;
                 cur_len -= iov_len;
@@ -544,8 +544,8 @@ void cluster_client_t::copy_write(cluster_op_t *op, std::map<object_id, cluster_
             }
             else
             {
-                memcpy(dirty_it->second.buf + pos - dirty_it->first.stripe,
-                    op->iov.buf[iov_idx].iov_base + iov_pos, cur_len);
+                memcpy((uint8_t*)dirty_it->second.buf + pos - dirty_it->first.stripe,
+                    (uint8_t*)op->iov.buf[iov_idx].iov_base + iov_pos, cur_len);
                 pos += cur_len;
                 len -= cur_len;
                 iov_pos += cur_len;
@@ -762,7 +762,7 @@ static void add_iov(int size, bool skip, cluster_op_t *op, int &iov_idx, size_t
         {
             if (!skip)
             {
-                iov.push_back(op->iov.buf[iov_idx].iov_base + iov_pos, cur_left);
+                iov.push_back((uint8_t*)op->iov.buf[iov_idx].iov_base + iov_pos, cur_left);
             }
             left -= cur_left;
             iov_pos = 0;
@@ -772,7 +772,7 @@ static void add_iov(int size, bool skip, cluster_op_t *op, int &iov_idx, size_t
         {
             if (!skip)
             {
-                iov.push_back(op->iov.buf[iov_idx].iov_base + iov_pos, left);
+                iov.push_back((uint8_t*)op->iov.buf[iov_idx].iov_base + iov_pos, left);
             }
             iov_pos += left;
             left = 0;
@@ -817,7 +817,7 @@ void cluster_client_t::slice_rw(cluster_op_t *op)
                 // First allocation
                 memset(op->bitmap_buf, 0, object_bitmap_size);
             }
-            op->part_bitmaps = op->bitmap_buf + object_bitmap_size;
+            op->part_bitmaps = (uint8_t*)op->bitmap_buf + object_bitmap_size;
             op->bitmap_buf_size = bitmap_mem;
         }
     }
@@ -839,7 +839,7 @@ void cluster_client_t::slice_rw(cluster_op_t *op)
             while (cur < end)
             {
                 unsigned bmp_loc = (cur - op->offset)/bs_bitmap_granularity;
-                bool skip = (((*(uint8_t*)(op->bitmap_buf + bmp_loc/8)) >> (bmp_loc%8)) & 0x1);
+                bool skip = (((*((uint8_t*)op->bitmap_buf + bmp_loc/8)) >> (bmp_loc%8)) & 0x1);
                 if (skip_prev != skip)
                 {
                     if (cur > prev)
@@ -944,7 +944,7 @@ bool cluster_client_t::try_send(cluster_op_t *op, int i)
                     .meta_revision = meta_rev,
                     .version = op->opcode == OSD_OP_WRITE || op->opcode == OSD_OP_DELETE ? op->version : 0,
                 } },
-                .bitmap = (op->opcode == OSD_OP_READ || op->opcode == OSD_OP_READ_BITMAP ? op->part_bitmaps + pg_bitmap_size*i : NULL),
+                .bitmap = (op->opcode == OSD_OP_READ || op->opcode == OSD_OP_READ_BITMAP ? (uint8_t*)op->part_bitmaps + pg_bitmap_size*i : NULL),
                 .bitmap_len = (unsigned)(op->opcode == OSD_OP_READ || op->opcode == OSD_OP_READ_BITMAP ? pg_bitmap_size : 0),
                 .callback = [this, part](osd_op_t *op_part)
                 {
@@ -1155,7 +1155,7 @@ void cluster_client_t::copy_part_bitmap(cluster_op_t *op, cluster_op_part_t *par
     if (!(object_offset & 0x7) && !(part_offset & 0x7) && (part_len >= 8))
     {
         // Copy bytes
-        mem_or(op->bitmap_buf + object_offset/8, part->op.bitmap + part_offset/8, part_len/8);
+        mem_or((uint8_t*)op->bitmap_buf + object_offset/8, (uint8_t*)part->op.bitmap + part_offset/8, part_len/8);
         object_offset += (part_len & ~0x7);
         part_offset += (part_len & ~0x7);
         part_len = (part_len & 0x7);
@@ -1163,8 +1163,8 @@ void cluster_client_t::copy_part_bitmap(cluster_op_t *op, cluster_op_part_t *par
     while (part_len > 0)
     {
         // Copy bits
-        (*(uint8_t*)(op->bitmap_buf + (object_offset >> 3))) |= (
-            (((*(uint8_t*)(part->op.bitmap + (part_offset >> 3))) >> (part_offset & 0x7)) & 0x1) << (object_offset & 0x7)
+        (*((uint8_t*)op->bitmap_buf + (object_offset >> 3))) |= (
+            (((*((uint8_t*)part->op.bitmap + (part_offset >> 3))) >> (part_offset & 0x7)) & 0x1) << (object_offset & 0x7)
         );
         part_offset++;
         object_offset++;

src/dump_journal.cpp

@@ -75,7 +75,7 @@ int main(int argc, char *argv[])
             uint64_t s;
             for (s = 0; s < self.journal_block; s += 8)
             {
-                if (*((uint64_t*)(data+s)) != 0)
+                if (*((uint64_t*)((uint8_t*)data+s)) != 0)
                     break;
             }
             if (s == self.journal_block)
@@ -139,7 +139,7 @@ int journal_dump_t::dump_block(void *buf)
     bool wrapped = false;
     while (pos < journal_block)
     {
-        journal_entry *je = (journal_entry*)(buf + pos);
+        journal_entry *je = (journal_entry*)((uint8_t*)buf + pos);
         if (je->magic != JOURNAL_MAGIC || je->type < JE_MIN || je->type > JE_MAX ||
             !all && started && je->crc32_prev != crc32_last)
         {

src/etcd_state_client.cpp

@@ -5,6 +5,7 @@
 #include "pg_states.h"
 #include "etcd_state_client.h"
 #ifndef __MOCK__
+#include "addr_util.h"
 #include "http_client.h"
 #include "base64.h"
 #endif
@@ -25,9 +26,14 @@ etcd_state_client_t::~etcd_state_client_t()
 #ifndef __MOCK__
     if (etcd_watch_ws)
     {
-        etcd_watch_ws->close();
+        http_close(etcd_watch_ws);
         etcd_watch_ws = NULL;
     }
+    if (keepalive_client)
+    {
+        http_close(keepalive_client);
+        keepalive_client = NULL;
+    }
 #endif
 }
 
@@ -48,12 +54,18 @@ etcd_kv_t etcd_state_client_t::parse_etcd_kv(const json11::Json & kv_json)
     return kv;
 }
 
-void etcd_state_client_t::etcd_txn(json11::Json txn, int timeout, std::function<void(std::string, json11::Json)> callback)
+void etcd_state_client_t::etcd_txn(json11::Json txn, int timeout, int retries, int interval, std::function<void(std::string, json11::Json)> callback)
 {
-    etcd_call("/kv/txn", txn, timeout, callback);
+    etcd_call("/kv/txn", txn, timeout, retries, interval, callback);
 }
 
-void etcd_state_client_t::etcd_call(std::string api, json11::Json payload, int timeout, std::function<void(std::string, json11::Json)> callback)
+void etcd_state_client_t::etcd_txn_slow(json11::Json txn, std::function<void(std::string, json11::Json)> callback)
+{
+    etcd_call("/kv/txn", txn, etcd_slow_timeout, max_etcd_attempts, 0, callback);
+}
+
+void etcd_state_client_t::etcd_call(std::string api, json11::Json payload, int timeout,
+    int retries, int interval, std::function<void(std::string, json11::Json)> callback)
 {
     if (!etcd_addresses.size() && !etcd_local.size())
     {
@@ -74,14 +86,49 @@ void etcd_state_client_t::etcd_call(std::string api, json11::Json payload, int t
         "Host: "+etcd_address+"\r\n"
         "Content-Type: application/json\r\n"
         "Content-Length: "+std::to_string(req.size())+"\r\n"
-        "Connection: close\r\n"
+        "Connection: keep-alive\r\n"
+        "Keep-Alive: timeout="+std::to_string(etcd_keepalive_timeout)+"\r\n"
         "\r\n"+req;
-    http_request_json(tfd, etcd_address, req, timeout, [this, cur_addr = selected_etcd_address, callback](std::string err, json11::Json data)
+    auto cb = [this, api, payload, timeout, retries, interval, callback,
+        cur_addr = selected_etcd_address](const http_response_t *response)
     {
-        if (err != "" && cur_addr == selected_etcd_address)
-            selected_etcd_address = "";
-        callback(err, data);
-    });
+        std::string err;
+        json11::Json data;
+        response->parse_json_response(err, data);
+        if (err != "")
+        {
+            if (cur_addr == selected_etcd_address)
+                selected_etcd_address = "";
+            if (retries > 0)
+            {
+                if (this->log_level > 0)
+                {
+                    printf(
+                        "Warning: etcd request failed: %s, retrying %d more times\n",
+                        err.c_str(), retries
+                    );
+                }
+                if (interval > 0)
+                {
+                    tfd->set_timer(interval, false, [this, api, payload, timeout, retries, interval, callback](int)
+                    {
+                        etcd_call(api, payload, timeout, retries-1, interval, callback);
+                    });
+                }
+                else
+                    etcd_call(api, payload, timeout, retries-1, interval, callback);
+            }
+            else
+                callback(err, data);
+        }
+        else
+            callback(err, data);
+    };
+    if (!keepalive_client)
+    {
+        keepalive_client = http_init(tfd);
+    }
+    http_request(keepalive_client, etcd_address, req, { .timeout = timeout, .keepalive = true }, cb);
 }
 
 void etcd_state_client_t::add_etcd_url(std::string addr)
@@ -155,6 +202,33 @@ void etcd_state_client_t::parse_config(const json11::Json & config)
         this->etcd_prefix = "/"+this->etcd_prefix;
     }
     this->log_level = config["log_level"].int64_value();
+    this->etcd_keepalive_timeout = config["etcd_keepalive_timeout"].uint64_value();
+    if (this->etcd_keepalive_timeout <= 0)
+    {
+        this->etcd_keepalive_timeout = config["etcd_report_interval"].uint64_value() * 2;
+        if (this->etcd_keepalive_timeout < 30)
+            this->etcd_keepalive_timeout = 30;
+    }
+    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->max_etcd_attempts = config["max_etcd_attempts"].uint64_value();
+    if (this->max_etcd_attempts <= 0)
+    {
+        this->max_etcd_attempts = 5;
+    }
+    this->etcd_slow_timeout = config["etcd_slow_timeout"].uint64_value();
+    if (this->etcd_slow_timeout <= 0)
+    {
+        this->etcd_slow_timeout = 5000;
+    }
+    this->etcd_quick_timeout = config["etcd_quick_timeout"].uint64_value();
+    if (this->etcd_quick_timeout <= 0)
+    {
+        this->etcd_quick_timeout = 1000;
+    }
 }
 
 void etcd_state_client_t::pick_next_etcd()
@@ -169,9 +243,16 @@ void etcd_state_client_t::pick_next_etcd()
         std::vector<int> ns;
         for (int i = 0; i < etcd_addresses.size(); i++)
             ns.push_back(i);
+        if (!rand_initialized)
+        {
+            timespec tv;
+            clock_gettime(CLOCK_REALTIME, &tv);
+            srand48(tv.tv_sec*1000000000 + tv.tv_nsec);
+            rand_initialized = true;
+        }
         while (ns.size())
         {
-            int i = rand() % ns.size();
+            int i = lrand48() % ns.size();
             addresses_to_try.push_back(etcd_addresses[ns[i]]);
             ns.erase(ns.begin()+i, ns.begin()+i+1);
         }
@@ -200,10 +281,12 @@ void etcd_state_client_t::start_etcd_watcher()
     ws_alive = 1;
     if (etcd_watch_ws)
     {
-        etcd_watch_ws->close();
+        http_close(etcd_watch_ws);
         etcd_watch_ws = NULL;
     }
-    etcd_watch_ws = open_websocket(tfd, etcd_address, etcd_api_path+"/watch", ETCD_SLOW_TIMEOUT,
+    if (this->log_level > 1)
+        printf("Trying to connect to etcd websocket at %s\n", etcd_address.c_str());
+    etcd_watch_ws = open_websocket(tfd, etcd_address, etcd_api_path+"/watch", etcd_slow_timeout,
         [this, cur_addr = selected_etcd_address](const http_response_t *msg)
     {
         if (msg->body.length())
@@ -219,6 +302,8 @@ void etcd_state_client_t::start_etcd_watcher()
             {
                 if (data["result"]["created"].bool_value())
                 {
+                    if (etcd_watches_initialised == 3 && this->log_level > 0)
+                        fprintf(stderr, "Successfully subscribed to etcd at %s\n", selected_etcd_address.c_str());
                     etcd_watches_initialised++;
                 }
                 if (data["result"]["canceled"].bool_value())
@@ -232,8 +317,11 @@ void etcd_state_client_t::start_etcd_watcher()
                         {
                             fprintf(stderr, "Revisions before %lu were compacted by etcd, reloading state\n",
                                 data["result"]["compact_revision"].uint64_value());
-                            etcd_watch_ws->close();
-                            etcd_watch_ws = NULL;
+                            if (etcd_watch_ws)
+                            {
+                                http_close(etcd_watch_ws);
+                                etcd_watch_ws = NULL;
+                            }
                             etcd_watch_revision = 0;
                             on_reload_hook();
                         }
@@ -284,13 +372,20 @@ void etcd_state_client_t::start_etcd_watcher()
         {
             if (cur_addr == selected_etcd_address)
             {
+                fprintf(stderr, "Disconnected from etcd %s\n", selected_etcd_address.c_str());
                 selected_etcd_address = "";
             }
-            etcd_watch_ws = NULL;
+            else
+                fprintf(stderr, "Disconnected from etcd\n");
+            if (etcd_watch_ws)
+            {
+                http_close(etcd_watch_ws);
+                etcd_watch_ws = NULL;
+            }
             if (etcd_watches_initialised == 0)
             {
-                // Connection not established, retry in <ETCD_QUICK_TIMEOUT>
-                tfd->set_timer(ETCD_QUICK_TIMEOUT, false, [this](int)
+                // Connection not established, retry in <etcd_quick_timeout>
+                tfd->set_timer(etcd_quick_timeout, false, [this](int)
                 {
                     start_etcd_watcher();
                 });
@@ -302,7 +397,7 @@ void etcd_state_client_t::start_etcd_watcher()
             }
         }
     });
-    etcd_watch_ws->post_message(WS_TEXT, json11::Json(json11::Json::object {
+    http_post_message(etcd_watch_ws, WS_TEXT, json11::Json(json11::Json::object {
         { "create_request", json11::Json::object {
             { "key", base64_encode(etcd_prefix+"/config/") },
             { "range_end", base64_encode(etcd_prefix+"/config0") },
@@ -311,7 +406,7 @@ void etcd_state_client_t::start_etcd_watcher()
             { "progress_notify", true },
         } }
     }).dump());
-    etcd_watch_ws->post_message(WS_TEXT, json11::Json(json11::Json::object {
+    http_post_message(etcd_watch_ws, WS_TEXT, json11::Json(json11::Json::object {
         { "create_request", json11::Json::object {
             { "key", base64_encode(etcd_prefix+"/osd/state/") },
             { "range_end", base64_encode(etcd_prefix+"/osd/state0") },
@@ -320,7 +415,7 @@ void etcd_state_client_t::start_etcd_watcher()
             { "progress_notify", true },
         } }
     }).dump());
-    etcd_watch_ws->post_message(WS_TEXT, json11::Json(json11::Json::object {
+    http_post_message(etcd_watch_ws, WS_TEXT, json11::Json(json11::Json::object {
         { "create_request", json11::Json::object {
             { "key", base64_encode(etcd_prefix+"/pg/state/") },
             { "range_end", base64_encode(etcd_prefix+"/pg/state0") },
@@ -329,7 +424,7 @@ void etcd_state_client_t::start_etcd_watcher()
             { "progress_notify", true },
         } }
     }).dump());
-    etcd_watch_ws->post_message(WS_TEXT, json11::Json(json11::Json::object {
+    http_post_message(etcd_watch_ws, WS_TEXT, json11::Json(json11::Json::object {
         { "create_request", json11::Json::object {
             { "key", base64_encode(etcd_prefix+"/pg/history/") },
             { "range_end", base64_encode(etcd_prefix+"/pg/history0") },
@@ -340,7 +435,7 @@ void etcd_state_client_t::start_etcd_watcher()
     }).dump());
     if (ws_keepalive_timer < 0)
     {
-        ws_keepalive_timer = tfd->set_timer(ETCD_KEEPALIVE_TIMEOUT, true, [this](int)
+        ws_keepalive_timer = tfd->set_timer(etcd_ws_keepalive_interval*1000, true, [this](int)
         {
             if (!etcd_watch_ws)
             {
@@ -348,14 +443,21 @@ void etcd_state_client_t::start_etcd_watcher()
             }
             else if (!ws_alive)
             {
-                etcd_watch_ws->close();
-                etcd_watch_ws = NULL;
+                if (this->log_level > 0)
+                {
+                    fprintf(stderr, "Websocket ping failed, disconnecting from etcd %s\n", selected_etcd_address.c_str());
+                }
+                if (etcd_watch_ws)
+                {
+                    http_close(etcd_watch_ws);
+                    etcd_watch_ws = NULL;
+                }
                 start_etcd_watcher();
             }
             else
             {
                 ws_alive = 0;
-                etcd_watch_ws->post_message(WS_TEXT, json11::Json(json11::Json::object {
+                http_post_message(etcd_watch_ws, WS_TEXT, json11::Json(json11::Json::object {
                     { "progress_request", json11::Json::object { } }
                 }).dump());
             }
@@ -367,12 +469,12 @@ void etcd_state_client_t::load_global_config()
 {
     etcd_call("/kv/range", json11::Json::object {
         { "key", base64_encode(etcd_prefix+"/config/global") }
-    }, ETCD_SLOW_TIMEOUT, [this](std::string err, json11::Json data)
+    }, etcd_slow_timeout, max_etcd_attempts, 0, [this](std::string err, json11::Json data)
     {
         if (err != "")
         {
             fprintf(stderr, "Error reading OSD configuration from etcd: %s\n", err.c_str());
-            tfd->set_timer(ETCD_SLOW_TIMEOUT, false, [this](int timer_id)
+            tfd->set_timer(etcd_slow_timeout, false, [this](int timer_id)
             {
                 load_global_config();
             });
@@ -440,12 +542,13 @@ void etcd_state_client_t::load_pgs()
     {
         req["compare"] = checks;
     }
-    etcd_txn(req, ETCD_SLOW_TIMEOUT, [this](std::string err, json11::Json data)
+    etcd_txn_slow(req, [this](std::string err, json11::Json data)
     {
         if (err != "")
         {
+            // Retry indefinitely
             fprintf(stderr, "Error loading PGs from etcd: %s\n", err.c_str());
-            tfd->set_timer(ETCD_SLOW_TIMEOUT, false, [this](int timer_id)
+            tfd->set_timer(etcd_slow_timeout, false, [this](int timer_id)
             {
                 load_pgs();
             });

src/etcd_state_client.h

@@ -12,11 +12,6 @@
 #define ETCD_PG_HISTORY_WATCH_ID 3
 #define ETCD_OSD_STATE_WATCH_ID 4
 
-#define MAX_ETCD_ATTEMPTS 5
-#define ETCD_SLOW_TIMEOUT 5000
-#define ETCD_QUICK_TIMEOUT 1000
-#define ETCD_KEEPALIVE_TIMEOUT 30000
-
 #define DEFAULT_BLOCK_SIZE 128*1024
 
 struct etcd_kv_t
@@ -71,7 +66,7 @@ struct inode_watch_t
     inode_config_t cfg;
 };
 
-struct websocket_t;
+struct http_co_t;
 
 struct etcd_state_client_t
 {
@@ -82,13 +77,20 @@ protected:
     std::string selected_etcd_address;
     std::vector<std::string> addresses_to_try;
     std::vector<inode_watch_t*> watches;
-    websocket_t *etcd_watch_ws = NULL;
+    http_co_t *etcd_watch_ws = NULL, *keepalive_client = NULL;
     int ws_keepalive_timer = -1;
     int ws_alive = 0;
+    bool rand_initialized = false;
     uint64_t bs_block_size = DEFAULT_BLOCK_SIZE;
     void add_etcd_url(std::string);
     void pick_next_etcd();
 public:
+    int etcd_keepalive_timeout = 30;
+    int etcd_ws_keepalive_interval = 30;
+    int max_etcd_attempts = 5;
+    int etcd_quick_timeout = 1000;
+    int etcd_slow_timeout = 5000;
+
     std::string etcd_prefix;
     int log_level = 0;
     timerfd_manager_t *tfd = NULL;
@@ -110,8 +112,9 @@ public:
 
     json11::Json::object serialize_inode_cfg(inode_config_t *cfg);
     etcd_kv_t parse_etcd_kv(const json11::Json & kv_json);
-    void etcd_call(std::string api, json11::Json payload, int timeout, std::function<void(std::string, json11::Json)> callback);
-    void etcd_txn(json11::Json txn, int timeout, std::function<void(std::string, json11::Json)> callback);
+    void etcd_call(std::string api, json11::Json payload, int timeout, int retries, int interval, std::function<void(std::string, json11::Json)> callback);
+    void etcd_txn(json11::Json txn, int timeout, int retries, int interval, std::function<void(std::string, json11::Json)> callback);
+    void etcd_txn_slow(json11::Json txn, std::function<void(std::string, json11::Json)> callback);
     void start_etcd_watcher();
     void load_global_config();
     void load_pgs();

src/fio_cluster.cpp

@@ -247,6 +247,12 @@ static int sec_setup(struct thread_data *td)
             vitastor_c_uring_wait_events(bsd->cli);
         }
         td->files[0]->real_file_size = vitastor_c_inode_get_size(bsd->watch);
+        if (!vitastor_c_inode_get_num(bsd->watch) ||
+            !td->files[0]->real_file_size)
+        {
+            td_verror(td, EINVAL, "image does not exist");
+            return 1;
+        }
     }
 
     bsd->trace = o->trace ? true : false;

src/fio_engine.cpp

@@ -26,9 +26,8 @@
 
 #include "blockstore.h"
 #include "epoll_manager.h"
-#include "fio_headers.h"
-
 #include "json11/json11.hpp"
+#include "fio_headers.h"
 
 struct bs_data
 {
@@ -150,7 +149,6 @@ static int bs_init(struct thread_data *td)
 static enum fio_q_status bs_queue(struct thread_data *td, struct io_u *io)
 {
     bs_data *bsd = (bs_data*)td->io_ops_data;
-    int n = bsd->op_n;
     if (io->ddir == DDIR_SYNC && bsd->last_sync)
     {
         return FIO_Q_COMPLETED;
@@ -178,7 +176,7 @@ static enum fio_q_status bs_queue(struct thread_data *td, struct io_u *io)
         op->version = UINT64_MAX; // last unstable
         op->offset = io->offset % bsd->bs->get_block_size();
         op->len = io->xfer_buflen;
-        op->callback = [io, n](blockstore_op_t *op)
+        op->callback = [io](blockstore_op_t *op)
         {
             io->error = op->retval < 0 ? -op->retval : 0;
             bs_data *bsd = (bs_data*)io->engine_data;
@@ -200,7 +198,7 @@ static enum fio_q_status bs_queue(struct thread_data *td, struct io_u *io)
         op->version = 0; // assign automatically
         op->offset = io->offset % bsd->bs->get_block_size();
         op->len = io->xfer_buflen;
-        op->callback = [io, n](blockstore_op_t *op)
+        op->callback = [io](blockstore_op_t *op)
         {
             io->error = op->retval < 0 ? -op->retval : 0;
             bs_data *bsd = (bs_data*)io->engine_data;
@@ -215,7 +213,7 @@ static enum fio_q_status bs_queue(struct thread_data *td, struct io_u *io)
         break;
     case DDIR_SYNC:
         op->opcode = BS_OP_SYNC_STAB_ALL;
-        op->callback = [io, n](blockstore_op_t *op)
+        op->callback = [io](blockstore_op_t *op)
         {
             bs_data *bsd = (bs_data*)io->engine_data;
             io->error = op->retval < 0 ? -op->retval : 0;
@@ -230,6 +228,7 @@ static enum fio_q_status bs_queue(struct thread_data *td, struct io_u *io)
         break;
     default:
         io->error = EINVAL;
+        delete op;
         return FIO_Q_COMPLETED;
     }
 

src/fio_sec_osd.cpp

@@ -28,16 +28,23 @@
 #include <vector>
 #include <unordered_map>
 
+#include "addr_util.h"
 #include "rw_blocking.h"
 #include "osd_ops.h"
 #include "fio_headers.h"
 
+struct op_buf_t
+{
+    osd_any_op_t buf;
+    io_u* fio_op;
+};
+
 struct sec_data
 {
     int connect_fd;
     /* block_size = 1 << block_order (128KB by default) */
     uint64_t block_order = 17, block_size = 1 << 17;
-    std::unordered_map<uint64_t, io_u*> queue;
+    std::unordered_map<uint64_t, op_buf_t*> queue;
     bool last_sync = false;
     /* The list of completed io_u structs. */
     std::vector<io_u*> completed;
@@ -52,6 +59,7 @@ struct sec_options
     int single_primary = 0;
     int trace = 0;
     int block_order = 17;
+    int zerocopy_send = 0;
 };
 
 static struct fio_option options[] = {
@@ -102,6 +110,16 @@ static struct fio_option options[] = {
         .category = FIO_OPT_C_ENGINE,
         .group  = FIO_OPT_G_FILENAME,
     },
+    {
+        .name   = "zerocopy_send",
+        .lname  = "Use zero-copy send",
+        .type   = FIO_OPT_BOOL,
+        .off1   = offsetof(struct sec_options, zerocopy_send),
+        .help   = "Use zero-copy send (MSG_ZEROCOPY)",
+        .def    = "0",
+        .category = FIO_OPT_C_ENGINE,
+        .group  = FIO_OPT_G_FILENAME,
+    },
     {
         .name = NULL,
     },
@@ -152,17 +170,14 @@ static int sec_init(struct thread_data *td)
     bsd->block_order = o->block_order == 0 ? 17 : o->block_order;
     bsd->block_size = 1 << o->block_order;
 
-    struct sockaddr_in addr;
-    int r;
-    if ((r = inet_pton(AF_INET, o->host ? o->host : "127.0.0.1", &addr.sin_addr)) != 1)
+    sockaddr addr;
+    if (!string_to_addr(std::string(o->host ? o->host : "127.0.0.1"), false, o->port > 0 ? o->port : 11203, &addr))
     {
-        fprintf(stderr, "server address: %s%s\n", o->host ? o->host : "127.0.0.1", r == 0 ? " is not valid" : ": no ipv4 support");
+        fprintf(stderr, "server address: %s is not valid\n", o->host ? o->host : "127.0.0.1");
         return 1;
     }
-    addr.sin_family = AF_INET;
-    addr.sin_port = htons(o->port ? o->port : 11203);
 
-    bsd->connect_fd = socket(AF_INET, SOCK_STREAM, 0);
+    bsd->connect_fd = socket(addr.sa_family, SOCK_STREAM, 0);
     if (bsd->connect_fd < 0)
     {
         perror("socket");
@@ -175,6 +190,14 @@ static int sec_init(struct thread_data *td)
     }
     int one = 1;
     setsockopt(bsd->connect_fd, SOL_TCP, TCP_NODELAY, &one, sizeof(one));
+    if (o->zerocopy_send)
+    {
+        if (setsockopt(bsd->connect_fd, SOL_SOCKET, SO_ZEROCOPY, &one, sizeof(one)) < 0)
+        {
+            perror("setsockopt zerocopy");
+            return 1;
+        }
+    }
 
     // FIXME: read config (block size) from OSD
 
@@ -195,7 +218,9 @@ static enum fio_q_status sec_queue(struct thread_data *td, struct io_u *io)
     }
 
     io->engine_data = bsd;
-    osd_any_op_t op = { 0 };
+    op_buf_t *op_buf = new op_buf_t;
+    op_buf->fio_op = io;
+    osd_any_op_t &op = op_buf->buf;
 
     op.hdr.magic = SECONDARY_OSD_OP_MAGIC;
     op.hdr.id = n;
@@ -259,6 +284,7 @@ static enum fio_q_status sec_queue(struct thread_data *td, struct io_u *io)
         break;
     default:
         io->error = EINVAL;
+        delete op_buf;
         return FIO_Q_COMPLETED;
     }
 
@@ -271,19 +297,18 @@ static enum fio_q_status sec_queue(struct thread_data *td, struct io_u *io)
     io->error = 0;
     bsd->inflight++;
     bsd->op_n++;
-    bsd->queue[n] = io;
+    bsd->queue[n] = op_buf;
 
     iovec iov[2] = { { .iov_base = op.buf, .iov_len = OSD_PACKET_SIZE } };
     int iovcnt = 1, wtotal = OSD_PACKET_SIZE;
     if (io->ddir == DDIR_WRITE)
     {
-        iov[1] = { .iov_base = io->xfer_buf, .iov_len = io->xfer_buflen };
+        iov[iovcnt++] = { .iov_base = io->xfer_buf, .iov_len = io->xfer_buflen };
         wtotal += io->xfer_buflen;
-        iovcnt++;
     }
-    if (writev_blocking(bsd->connect_fd, iov, iovcnt) != wtotal)
+    if (sendv_blocking(bsd->connect_fd, iov, iovcnt, opt->zerocopy_send ? MSG_ZEROCOPY : 0) != wtotal)
     {
-        perror("writev");
+        perror("sendmsg");
         exit(1);
     }
 
@@ -312,7 +337,8 @@ static int sec_getevents(struct thread_data *td, unsigned int min, unsigned int
             fprintf(stderr, "bad reply: op id %lx missing in local queue\n", reply.hdr.id);
             exit(1);
         }
-        io_u* io = it->second;
+        io_u* io = it->second->fio_op;
+        delete it->second;
         bsd->queue.erase(it);
         if (io->ddir == DDIR_READ)
         {
@@ -321,7 +347,23 @@ static int sec_getevents(struct thread_data *td, unsigned int min, unsigned int
                 fprintf(stderr, "Short read: retval = %ld instead of %llu\n", reply.hdr.retval, io->xfer_buflen);
                 exit(1);
             }
-            read_blocking(bsd->connect_fd, io->xfer_buf, io->xfer_buflen);
+            // Support bitmap
+            uint64_t bitmap = 0;
+            int iovcnt = 0;
+            iovec iov[2];
+            if (reply.sec_rw.attr_len > 0)
+            {
+                if (reply.sec_rw.attr_len <= 8)
+                    iov[iovcnt++] = { .iov_base = &bitmap, .iov_len = reply.sec_rw.attr_len };
+                else
+                    iov[iovcnt++] = { .iov_base = (void*)(bitmap = (uint64_t)malloc(reply.sec_rw.attr_len)), .iov_len = reply.sec_rw.attr_len };
+            }
+            iov[iovcnt++] = { .iov_base = io->xfer_buf, .iov_len = io->xfer_buflen };
+            readv_blocking(bsd->connect_fd, iov, iovcnt);
+            if (reply.sec_rw.attr_len > 8)
+            {
+                free((void*)bitmap);
+            }
         }
         else if (io->ddir == DDIR_WRITE)
         {

src/http_client.cpp

@@ -4,9 +4,7 @@
 #include <netinet/tcp.h>
 #include <sys/epoll.h>
 
-#include <net/if.h>
 #include <arpa/inet.h>
-#include <ifaddrs.h>
 
 #include <ctype.h>
 #include <unistd.h>
@@ -15,21 +13,22 @@
 
 #include <stdexcept>
 
+#include "addr_util.h"
 #include "json11/json11.hpp"
 #include "http_client.h"
 #include "timerfd_manager.h"
 
 #define READ_BUFFER_SIZE 9000
 
-static int extract_port(std::string & host);
 static std::string trim(const std::string & in);
 static std::string ws_format_frame(int type, uint64_t size);
 static bool ws_parse_frame(std::string & buf, int & type, std::string & res);
+static void parse_http_headers(std::string & res, http_response_t *parsed);
 
-// FIXME: Use keepalive
 struct http_co_t
 {
     timerfd_manager_t *tfd;
+    std::function<void(const http_response_t*)> response_callback;
 
     int request_timeout = 0;
     std::string host;
@@ -37,11 +36,12 @@ struct http_co_t
     std::string ws_outbox;
     std::string response;
     bool want_streaming;
+    bool keepalive;
 
-    http_response_t parsed;
-    uint64_t target_response_size = 0;
+    std::vector<std::function<void()>> keepalive_queue;
 
     int state = 0;
+    std::string connected_host;
     int peer_fd = -1;
     int timeout_id = -1;
     int epoll_events = 0;
@@ -49,10 +49,8 @@ struct http_co_t
     std::vector<char> rbuf;
     iovec read_iov, send_iov;
     msghdr read_msg = { 0 }, send_msg = { 0 };
-
-    std::function<void(const http_response_t*)> callback;
-
-    websocket_t ws;
+    http_response_t parsed;
+    uint64_t target_response_size = 0;
 
     int onstack = 0;
     bool ended = false;
@@ -61,66 +59,40 @@ struct http_co_t
     inline void stackin() { onstack++; }
     inline void stackout() { onstack--; if (!onstack && ended) end(); }
     inline void end() { ended = true; if (!onstack) { delete this; } }
+    void run_cb_and_clear();
     void start_connection();
+    void close_connection();
     void handle_events();
     void handle_connect_result();
     void submit_read();
     void submit_send();
     bool handle_read();
     void post_message(int type, const std::string & msg);
+    void send_request(const std::string & host, const std::string & request,
+        const http_options_t & options, std::function<void(const http_response_t *response)> response_callback);
 };
 
+#define HTTP_CO_CLOSED 0
 #define HTTP_CO_CONNECTING 1
 #define HTTP_CO_SENDING_REQUEST 2
 #define HTTP_CO_REQUEST_SENT 3
 #define HTTP_CO_HEADERS_RECEIVED 4
 #define HTTP_CO_WEBSOCKET 5
 #define HTTP_CO_CHUNKED 6
+#define HTTP_CO_KEEPALIVE 7
 
 #define DEFAULT_TIMEOUT 5000
 
-void http_request(timerfd_manager_t *tfd, const std::string & host, const std::string & request,
-    const http_options_t & options, std::function<void(const http_response_t *response)> callback)
+http_co_t *http_init(timerfd_manager_t *tfd)
 {
     http_co_t *handler = new http_co_t();
-    handler->request_timeout = options.timeout < 0 ? 0 : (options.timeout == 0 ? DEFAULT_TIMEOUT : options.timeout);
-    handler->want_streaming = options.want_streaming;
     handler->tfd = tfd;
-    handler->host = host;
-    handler->request = request;
-    handler->callback = callback;
-    handler->ws.co = handler;
-    handler->start_connection();
+    handler->state = HTTP_CO_CLOSED;
+    return handler;
 }
 
-void http_request_json(timerfd_manager_t *tfd, const std::string & host, const std::string & request,
-    int timeout, std::function<void(std::string, json11::Json r)> callback)
-{
-    http_request(tfd, host, request, { .timeout = timeout }, [callback](const http_response_t* res)
-    {
-        if (res->error_code != 0)
-        {
-            callback("Error code: "+std::to_string(res->error_code)+" ("+std::string(strerror(res->error_code))+")", json11::Json());
-            return;
-        }
-        if (res->status_code != 200)
-        {
-            callback("HTTP "+std::to_string(res->status_code)+" "+res->status_line+" body: "+trim(res->body), json11::Json());
-            return;
-        }
-        std::string json_err;
-        json11::Json data = json11::Json::parse(res->body, json_err);
-        if (json_err != "")
-        {
-            callback("Bad JSON: "+json_err+" (response: "+trim(res->body)+")", json11::Json());
-            return;
-        }
-        callback(std::string(), data);
-    });
-}
-
-websocket_t* open_websocket(timerfd_manager_t *tfd, const std::string & host, const std::string & path,
-    int timeout, std::function<void(const http_response_t *msg)> callback)
+http_co_t* open_websocket(timerfd_manager_t *tfd, const std::string & host, const std::string & path,
+    int timeout, std::function<void(const http_response_t *msg)> response_callback)
 {
     std::string request = "GET "+path+" HTTP/1.1\r\n"
         "Host: "+host+"\r\n"
@@ -130,28 +102,154 @@ websocket_t* open_websocket(timerfd_manager_t *tfd, const std::string & host, co
         "Sec-WebSocket-Version: 13\r\n"
         "\r\n";
     http_co_t *handler = new http_co_t();
+    handler->tfd = tfd;
+    handler->state = HTTP_CO_CLOSED;
+    handler->host = host;
     handler->request_timeout = timeout < 0 ? -1 : (timeout == 0 ? DEFAULT_TIMEOUT : timeout);
     handler->want_streaming = false;
-    handler->tfd = tfd;
-    handler->host = host;
+    handler->keepalive = false;
     handler->request = request;
-    handler->callback = callback;
-    handler->ws.co = handler;
+    handler->response_callback = response_callback;
     handler->start_connection();
-    return &handler->ws;
+    return handler;
 }
 
-void websocket_t::post_message(int type, const std::string & msg)
+void http_request(http_co_t *handler, const std::string & host, const std::string & request,
+    const http_options_t & options, std::function<void(const http_response_t *response)> response_callback)
 {
-    co->post_message(type, msg);
+    handler->send_request(host, request, options, response_callback);
 }
 
-void websocket_t::close()
+void http_co_t::run_cb_and_clear()
 {
-    co->end();
+    parsed.eof = true;
+    std::function<void(const http_response_t*)> cb;
+    cb.swap(response_callback);
+    // Call callback after clearing it because otherwise we may hit reenterability problems
+    if (cb != NULL)
+        cb(&parsed);
+}
+
+void http_co_t::send_request(const std::string & host, const std::string & request,
+    const http_options_t & options, std::function<void(const http_response_t *response)> response_callback)
+{
+    stackin();
+    if (state == HTTP_CO_WEBSOCKET)
+    {
+        stackout();
+        throw std::runtime_error("Attempt to send HTTP request into a websocket or chunked stream");
+    }
+    else if (state != HTTP_CO_KEEPALIVE && state != HTTP_CO_CLOSED)
+    {
+        keepalive_queue.push_back([this, host, request, options, response_callback]()
+        {
+            this->send_request(host, request, options, response_callback);
+        });
+        stackout();
+        return;
+    }
+    if (state == HTTP_CO_KEEPALIVE && connected_host != host)
+    {
+        close_connection();
+    }
+    this->request_timeout = options.timeout < 0 ? 0 : (options.timeout == 0 ? DEFAULT_TIMEOUT : options.timeout);
+    this->want_streaming = options.want_streaming;
+    this->keepalive = options.keepalive;
+    this->host = host;
+    this->request = request;
+    this->response = "";
+    this->sent = 0;
+    this->response_callback = response_callback;
+    this->parsed = {};
+    if (request_timeout > 0)
+    {
+        timeout_id = tfd->set_timer(request_timeout, false, [this](int timer_id)
+        {
+            stackin();
+            close_connection();
+            parsed = { .error = "HTTP request timed out" };
+            run_cb_and_clear();
+            stackout();
+        });
+    }
+    if (state == HTTP_CO_KEEPALIVE)
+    {
+        state = HTTP_CO_SENDING_REQUEST;
+        submit_send();
+    }
+    else
+    {
+        start_connection();
+    }
+    stackout();
+}
+
+void http_post_message(http_co_t *handler, int type, const std::string & msg)
+{
+    handler->post_message(type, msg);
+}
+
+void http_co_t::post_message(int type, const std::string & msg)
+{
+    stackin();
+    if (state == HTTP_CO_WEBSOCKET)
+    {
+        request += ws_format_frame(type, msg.size());
+        request += msg;
+        submit_send();
+    }
+    else if (state == HTTP_CO_KEEPALIVE || state == HTTP_CO_CHUNKED)
+    {
+        throw std::runtime_error("Attempt to send websocket message on a regular HTTP connection");
+    }
+    else
+    {
+        ws_outbox += ws_format_frame(type, msg.size());
+        ws_outbox += msg;
+    }
+    stackout();
+}
+
+void http_close(http_co_t *handler)
+{
+    handler->end();
+}
+
+void http_response_t::parse_json_response(std::string & error, json11::Json & r) const
+{
+    if (this->error != "")
+    {
+        error = this->error;
+        r = json11::Json();
+    }
+    else if (status_code != 200)
+    {
+        error = "HTTP "+std::to_string(status_code)+" "+status_line+" body: "+trim(body);
+        r = json11::Json();
+    }
+    else
+    {
+        std::string json_err;
+        json11::Json data = json11::Json::parse(body, json_err);
+        if (json_err != "")
+        {
+            error = "Bad JSON: "+json_err+" (response: "+trim(body)+")";
+            r = json11::Json();
+        }
+        else
+        {
+            error = "";
+            r = data;
+        }
+    }
 }
 
 http_co_t::~http_co_t()
+{
+    close_connection();
+}
+
+void http_co_t::close_connection()
 {
     if (timeout_id >= 0)
     {
@@ -164,67 +262,41 @@ http_co_t::~http_co_t()
         close(peer_fd);
         peer_fd = -1;
     }
-    if (parsed.headers["transfer-encoding"] == "chunked")
-    {
-        int prev = 0, pos = 0;
-        while ((pos = response.find("\r\n", prev)) >= prev)
-        {
-            uint64_t len = strtoull(response.c_str()+prev, NULL, 16);
-            parsed.body += response.substr(pos+2, len);
-            prev = pos+2+len+2;
-        }
-    }
-    else
-    {
-        std::swap(parsed.body, response);
-    }
-    parsed.eof = true;
-    callback(&parsed);
+    state = HTTP_CO_CLOSED;
+    connected_host = "";
+    response = "";
+    epoll_events = 0;
 }
 
 void http_co_t::start_connection()
 {
     stackin();
-    int port = extract_port(host);
-    struct sockaddr_in addr;
-    int r;
-    if ((r = inet_pton(AF_INET, host.c_str(), &addr.sin_addr)) != 1)
+    struct sockaddr addr;
+    if (!string_to_addr(host.c_str(), 1, 80, &addr))
     {
-        parsed.error_code = ENXIO;
+        parsed = { .error = "Invalid address: "+host };
+        run_cb_and_clear();
         stackout();
-        end();
         return;
     }
-    addr.sin_family = AF_INET;
-    addr.sin_port = htons(port ? port : 80);
-    peer_fd = socket(AF_INET, SOCK_STREAM, 0);
+    peer_fd = socket(addr.sa_family, SOCK_STREAM, 0);
     if (peer_fd < 0)
     {
-        parsed.error_code = errno;
+        parsed = { .error = std::string("socket: ")+strerror(errno) };
+        run_cb_and_clear();
         stackout();
-        end();
         return;
     }
     fcntl(peer_fd, F_SETFL, fcntl(peer_fd, F_GETFL, 0) | O_NONBLOCK);
-    if (request_timeout > 0)
-    {
-        timeout_id = tfd->set_timer(request_timeout, false, [this](int timer_id)
-        {
-            if (response.length() == 0)
-            {
-                parsed.error_code = ETIME;
-            }
-            end();
-        });
-    }
     epoll_events = 0;
     // Finally call connect
-    r = ::connect(peer_fd, (sockaddr*)&addr, sizeof(addr));
+    int r = ::connect(peer_fd, (sockaddr*)&addr, sizeof(addr));
     if (r < 0 && errno != EINPROGRESS)
     {
-        parsed.error_code = errno;
+        close_connection();
+        parsed = { .error = std::string("connect: ")+strerror(errno) };
+        run_cb_and_clear();
         stackout();
-        end();
         return;
     }
     tfd->set_fd_handler(peer_fd, true, [this](int peer_fd, int epoll_events)
@@ -232,6 +304,7 @@ void http_co_t::start_connection()
         this->epoll_events |= epoll_events;
         handle_events();
     });
+    connected_host = host;
     state = HTTP_CO_CONNECTING;
     stackout();
 }
@@ -254,7 +327,8 @@ void http_co_t::handle_events()
             }
             else if (epoll_events & (EPOLLRDHUP|EPOLLERR))
             {
-                end();
+                close_connection();
+                run_cb_and_clear();
                 break;
             }
         }
@@ -273,9 +347,10 @@ void http_co_t::handle_connect_result()
     }
     if (result != 0)
     {
-        parsed.error_code = result;
+        close_connection();
+        parsed = { .error = std::string("connect: ")+strerror(result) };
+        run_cb_and_clear();
         stackout();
-        end();
         return;
     }
     int one = 1;
@@ -290,6 +365,51 @@ void http_co_t::handle_connect_result()
     stackout();
 }
 
+void http_co_t::submit_send()
+{
+    stackin();
+    int res;
+again:
+    if (sent < request.size())
+    {
+        send_iov = (iovec){ .iov_base = (void*)(request.c_str()+sent), .iov_len = request.size()-sent };
+        send_msg.msg_iov = &send_iov;
+        send_msg.msg_iovlen = 1;
+        res = sendmsg(peer_fd, &send_msg, MSG_NOSIGNAL);
+        if (res < 0)
+        {
+            res = -errno;
+        }
+        if (res == -EAGAIN || res == -EINTR)
+        {
+            res = 0;
+        }
+        else if (res < 0)
+        {
+            close_connection();
+            parsed = { .error = std::string("sendmsg: ")+strerror(errno) };
+            run_cb_and_clear();
+            stackout();
+            return;
+        }
+        sent += res;
+        if (state == HTTP_CO_SENDING_REQUEST)
+        {
+            if (sent >= request.size())
+                state = HTTP_CO_REQUEST_SENT;
+            else
+                goto again;
+        }
+        else if (state == HTTP_CO_WEBSOCKET)
+        {
+            request = request.substr(sent);
+            sent = 0;
+            goto again;
+        }
+    }
+    stackout();
+}
+
 void http_co_t::submit_read()
 {
     stackin();
@@ -306,16 +426,18 @@ void http_co_t::submit_read()
     {
         res = -errno;
     }
-    if (res == -EAGAIN)
+    if (res == -EAGAIN || res == -EINTR)
     {
         epoll_events = epoll_events & ~EPOLLIN;
     }
     else if (res <= 0)
     {
         // < 0 means error, 0 means EOF
-        if (!res)
-            epoll_events = epoll_events & ~EPOLLIN;
-        end();
+        epoll_events = epoll_events & ~EPOLLIN;
+        close_connection();
+        if (res < 0)
+            parsed = { .error = std::string("recvmsg: ")+strerror(-res) };
+        run_cb_and_clear();
     }
     else
     {
@@ -325,51 +447,6 @@ void http_co_t::submit_read()
     stackout();
 }
 
-void http_co_t::submit_send()
-{
-    stackin();
-    int res;
-again:
-    if (sent < request.size())
-    {
-        send_iov = (iovec){ .iov_base = (void*)(request.c_str()+sent), .iov_len = request.size()-sent };
-        send_msg.msg_iov = &send_iov;
-        send_msg.msg_iovlen = 1;
-        res = sendmsg(peer_fd, &send_msg, MSG_NOSIGNAL);
-        if (res < 0)
-        {
-            res = -errno;
-        }
-        if (res == -EAGAIN)
-        {
-            res = 0;
-        }
-        else if (res < 0)
-        {
-            stackout();
-            end();
-            return;
-        }
-        sent += res;
-        if (state == HTTP_CO_SENDING_REQUEST)
-        {
-            if (sent >= request.size())
-            {
-                state = HTTP_CO_REQUEST_SENT;
-            }
-            else
-                goto again;
-        }
-        else if (state == HTTP_CO_WEBSOCKET)
-        {
-            request = request.substr(sent);
-            sent = 0;
-            goto again;
-        }
-    }
-    stackout();
-}
-
 bool http_co_t::handle_read()
 {
     stackin();
@@ -380,6 +457,7 @@ bool http_co_t::handle_read()
         {
             if (timeout_id >= 0)
             {
+                // Timeout is cleared when headers are received
                 tfd->clear_timer(timeout_id);
                 timeout_id = -1;
             }
@@ -407,20 +485,26 @@ bool http_co_t::handle_read()
                 if (!target_response_size)
                 {
                     // Sorry, unsupported response
+                    close_connection();
+                    parsed = { .error = "Response has neither Connection: close, nor Transfer-Encoding: chunked nor Content-Length headers" };
+                    run_cb_and_clear();
                     stackout();
-                    end();
                     return false;
                 }
             }
+            else
+            {
+                keepalive = false;
+            }
         }
     }
     if (state == HTTP_CO_HEADERS_RECEIVED && target_response_size > 0 && response.size() >= target_response_size)
     {
-        stackout();
-        end();
-        return false;
+        std::swap(parsed.body, response);
+        response_callback(&parsed);
+        parsed.eof = true;
     }
-    if (state == HTTP_CO_CHUNKED && response.size() > 0)
+    else if (state == HTTP_CO_CHUNKED && response.size() > 0)
     {
         int prev = 0, pos = 0;
         while ((pos = response.find("\r\n", prev)) >= prev)
@@ -443,55 +527,49 @@ bool http_co_t::handle_read()
         {
             response = response.substr(prev);
         }
-        if (parsed.eof)
+        if (want_streaming)
         {
-            stackout();
-            end();
-            return false;
-        }
-        if (want_streaming && parsed.body.size() > 0)
-        {
-            if (!ended)
-            {
-                // Don't deliver additional events after close()
-                callback(&parsed);
-            }
+            // Streaming response
+            response_callback(&parsed);
             parsed.body = "";
         }
+        if (parsed.eof && !want_streaming)
+        {
+            // Normal response
+            response_callback(&parsed);
+        }
     }
-    if (state == HTTP_CO_WEBSOCKET && response.size() > 0)
+    else if (state == HTTP_CO_WEBSOCKET && response.size() > 0)
     {
         while (ws_parse_frame(response, parsed.ws_msg_type, parsed.body))
         {
-            if (!ended)
-            {
-                // Don't deliver additional events after close()
-                callback(&parsed);
-            }
+            response_callback(&parsed);
             parsed.body = "";
         }
     }
+    if (parsed.eof)
+    {
+        response_callback = NULL;
+        parsed = {};
+        if (!keepalive)
+        {
+            close_connection();
+        }
+        else
+        {
+            state = HTTP_CO_KEEPALIVE;
+            if (keepalive_queue.size() > 0)
+            {
+                auto next = keepalive_queue[0];
+                keepalive_queue.erase(keepalive_queue.begin(), keepalive_queue.begin()+1);
+                next();
+            }
+        }
+    }
     stackout();
     return true;
 }
 
-void http_co_t::post_message(int type, const std::string & msg)
-{
-    stackin();
-    if (state == HTTP_CO_WEBSOCKET)
-    {
-        request += ws_format_frame(type, msg.size());
-        request += msg;
-        submit_send();
-    }
-    else
-    {
-        ws_outbox += ws_format_frame(type, msg.size());
-        ws_outbox += msg;
-    }
-    stackout();
-}
-
 uint64_t stoull_full(const std::string & str, int base)
 {
     if (isspace(str[0]))
@@ -507,7 +585,7 @@ uint64_t stoull_full(const std::string & str, int base)
     return r;
 }
 
-void parse_http_headers(std::string & res, http_response_t *parsed)
+static void parse_http_headers(std::string & res, http_response_t *parsed)
 {
     int pos = res.find("\r\n");
     pos = pos < 0 ? res.length() : pos+2;
@@ -556,13 +634,13 @@ static std::string ws_format_frame(int type, uint64_t size)
         res[p++] = size | /*mask*/0x80;
     else if (size < 65536)
     {
-        res[p++] = 126 | /*mask*/0x80;
+        res[p++] = (char)(126 | /*mask*/0x80);
         res[p++] = (size >> 8) & 0xFF;
         res[p++] = (size >> 0) & 0xFF;
     }
     else
     {
-        res[p++] = 127 | /*mask*/0x80;
+        res[p++] = (char)(127 | /*mask*/0x80);
         res[p++] = (size >> 56) & 0xFF;
         res[p++] = (size >> 48) & 0xFF;
         res[p++] = (size >> 40) & 0xFF;
@@ -629,152 +707,6 @@ static bool ws_parse_frame(std::string & buf, int & type, std::string & res)
     return true;
 }
 
-static bool cidr_match(const in_addr &addr, const in_addr &net, uint8_t bits)
-{
-    if (bits == 0)
-    {
-        // C99 6.5.7 (3): u32 << 32 is undefined behaviour
-        return true;
-    }
-    return !((addr.s_addr ^ net.s_addr) & htonl(0xFFFFFFFFu << (32 - bits)));
-}
-
-static bool cidr6_match(const in6_addr &address, const in6_addr &network, uint8_t bits)
-{
-    const uint32_t *a = address.s6_addr32;
-    const uint32_t *n = network.s6_addr32;
-    int bits_whole, bits_incomplete;
-    bits_whole = bits >> 5;         // number of whole u32
-    bits_incomplete = bits & 0x1F;  // number of bits in incomplete u32
-    if (bits_whole && memcmp(a, n, bits_whole << 2))
-        return false;
-    if (bits_incomplete)
-    {
-        uint32_t mask = htonl((0xFFFFFFFFu) << (32 - bits_incomplete));
-        if ((a[bits_whole] ^ n[bits_whole]) & mask)
-            return false;
-    }
-    return true;
-}
-
-struct addr_mask_t
-{
-    sa_family_t family;
-    in_addr ipv4;
-    in6_addr ipv6;
-    uint8_t bits;
-};
-
-std::vector<std::string> getifaddr_list(json11::Json mask_cfg, bool include_v6)
-{
-    std::vector<addr_mask_t> masks;
-    if (mask_cfg.is_string())
-    {
-        mask_cfg = json11::Json::array{ mask_cfg };
-    }
-    for (auto mask_json: mask_cfg.array_items())
-    {
-        std::string mask = mask_json.string_value();
-        unsigned bits = 0;
-        int p = mask.find('/');
-        if (p != std::string::npos)
-        {
-            char null_byte = 0;
-            if (sscanf(mask.c_str()+p+1, "%u%c", &bits, &null_byte) != 1 || bits > 128)
-            {
-                throw std::runtime_error((include_v6 ? "Invalid IPv4 address mask: " : "Invalid IP address mask: ") + mask);
-            }
-            mask = mask.substr(0, p);
-        }
-        in_addr ipv4;
-        in6_addr ipv6;
-        if (inet_pton(AF_INET, mask.c_str(), &ipv4) == 1)
-        {
-            if (bits > 32)
-            {
-                throw std::runtime_error((include_v6 ? "Invalid IPv4 address mask: " : "Invalid IP address mask: ") + mask);
-            }
-            masks.push_back((addr_mask_t){ .family = AF_INET, .ipv4 = ipv4, .bits = (uint8_t)bits });
-        }
-        else if (include_v6 && inet_pton(AF_INET6, mask.c_str(), &ipv6) == 1)
-        {
-            masks.push_back((addr_mask_t){ .family = AF_INET6, .ipv6 = ipv6, .bits = (uint8_t)bits });
-        }
-        else
-        {
-            throw std::runtime_error((include_v6 ? "Invalid IPv4 address mask: " : "Invalid IP address mask: ") + mask);
-        }
-    }
-    std::vector<std::string> addresses;
-    ifaddrs *list, *ifa;
-    if (getifaddrs(&list) == -1)
-    {
-        throw std::runtime_error(std::string("getifaddrs: ") + strerror(errno));
-    }
-    for (ifa = list; ifa != NULL; ifa = ifa->ifa_next)
-    {
-        if (!ifa->ifa_addr)
-        {
-            continue;
-        }
-        int family = ifa->ifa_addr->sa_family;
-        if ((family == AF_INET || family == AF_INET6 && include_v6) &&
-            (ifa->ifa_flags & (IFF_UP | IFF_RUNNING | IFF_LOOPBACK)) == (IFF_UP | IFF_RUNNING))
-        {
-            void *addr_ptr;
-            if (family == AF_INET)
-            {
-                addr_ptr = &((sockaddr_in *)ifa->ifa_addr)->sin_addr;
-            }
-            else
-            {
-                addr_ptr = &((sockaddr_in6 *)ifa->ifa_addr)->sin6_addr;
-            }
-            if (masks.size() > 0)
-            {
-                int i;
-                for (i = 0; i < masks.size(); i++)
-                {
-                    if (masks[i].family == family && (family == AF_INET
-                        ? cidr_match(*(in_addr*)addr_ptr, masks[i].ipv4, masks[i].bits)
-                        : cidr6_match(*(in6_addr*)addr_ptr, masks[i].ipv6, masks[i].bits)))
-                    {
-                        break;
-                    }
-                }
-                if (i >= masks.size())
-                {
-                    continue;
-                }
-            }
-            char addr[INET6_ADDRSTRLEN];
-            if (!inet_ntop(family, addr_ptr, addr, INET6_ADDRSTRLEN))
-            {
-                throw std::runtime_error(std::string("inet_ntop: ") + strerror(errno));
-            }
-            addresses.push_back(std::string(addr));
-        }
-    }
-    freeifaddrs(list);
-    return addresses;
-}
-
-static int extract_port(std::string & host)
-{
-    int port = 0;
-    int pos = 0;
-    if ((pos = host.find(':')) >= 0)
-    {
-        port = strtoull(host.c_str() + pos + 1, NULL, 10);
-        if (port >= 0x10000)
-        {
-            port = 0;
-        }
-        host = host.substr(0, pos);
-    }
-    return port;
-}
-
 std::string strtolower(const std::string & in)
 {
     std::string s = in;

src/http_client.h

@@ -21,41 +21,34 @@ struct http_options_t
 {
     int timeout;
     bool want_streaming;
+    bool keepalive;
 };
 
 struct http_response_t
 {
+    std::string error;
+
     bool eof = false;
-    int error_code = 0;
     int status_code = 0;
     std::string status_line;
     std::map<std::string, std::string> headers;
     int ws_msg_type = -1;
     std::string body;
+
+    void parse_json_response(std::string & error, json11::Json & r) const;
 };
 
+// Opened websocket or keepalive HTTP connection
 struct http_co_t;
 
-struct websocket_t
-{
-    http_co_t *co;
-    void post_message(int type, const std::string & msg);
-    void close();
-};
-
-void parse_http_headers(std::string & res, http_response_t *parsed);
-
-std::vector<std::string> getifaddr_list(json11::Json mask_cfg = json11::Json(), bool include_v6 = false);
+http_co_t* http_init(timerfd_manager_t *tfd);
+http_co_t* open_websocket(timerfd_manager_t *tfd, const std::string & host, const std::string & path,
+    int timeout, std::function<void(const http_response_t *msg)> on_message);
+void http_request(http_co_t *handler, const std::string & host, const std::string & request,
+    const http_options_t & options, std::function<void(const http_response_t *response)> response_callback);
+void http_post_message(http_co_t *handler, int type, const std::string & msg);
+void http_close(http_co_t *co);
 
+// Utils
 uint64_t stoull_full(const std::string & str, int base = 10);
-
 std::string strtolower(const std::string & in);
-
-void http_request(timerfd_manager_t *tfd, const std::string & host, const std::string & request,
-    const http_options_t & options, std::function<void(const http_response_t *response)> callback);
-
-void http_request_json(timerfd_manager_t *tfd, const std::string & host, const std::string & request,
-    int timeout, std::function<void(std::string, json11::Json r)> callback);
-
-websocket_t* open_websocket(timerfd_manager_t *tfd, const std::string & host, const std::string & path,
-    int timeout, std::function<void(const http_response_t *msg)> callback);

src/messenger.cpp

@@ -4,10 +4,12 @@
 #include <unistd.h>
 #include <fcntl.h>
 #include <sys/socket.h>
+#include <sys/stat.h>
 #include <sys/epoll.h>
 #include <netinet/tcp.h>
 #include <stdexcept>
 
+#include "addr_util.h"
 #include "messenger.h"
 
 void osd_messenger_t::init()
@@ -220,23 +222,20 @@ void osd_messenger_t::try_connect_peer(uint64_t peer_osd)
 void osd_messenger_t::try_connect_peer_addr(osd_num_t peer_osd, const char *peer_host, int peer_port)
 {
     assert(peer_osd != this->osd_num);
-    struct sockaddr_in addr;
-    int r;
-    if ((r = inet_pton(AF_INET, peer_host, &addr.sin_addr)) != 1)
+    struct sockaddr addr;
+    if (!string_to_addr(peer_host, 0, peer_port, &addr))
     {
         on_connect_peer(peer_osd, -EINVAL);
         return;
     }
-    addr.sin_family = AF_INET;
-    addr.sin_port = htons(peer_port ? peer_port : 11203);
-    int peer_fd = socket(AF_INET, SOCK_STREAM, 0);
+    int peer_fd = socket(addr.sa_family, SOCK_STREAM, 0);
     if (peer_fd < 0)
     {
         on_connect_peer(peer_osd, -errno);
         return;
     }
     fcntl(peer_fd, F_SETFL, fcntl(peer_fd, F_GETFL, 0) | O_NONBLOCK);
-    r = connect(peer_fd, (sockaddr*)&addr, sizeof(addr));
+    int r = connect(peer_fd, (sockaddr*)&addr, sizeof(addr));
     if (r < 0 && errno != EINPROGRESS)
     {
         close(peer_fd);
@@ -485,21 +484,20 @@ void osd_messenger_t::check_peer_config(osd_client_t *cl)
 void osd_messenger_t::accept_connections(int listen_fd)
 {
     // Accept new connections
-    sockaddr_in addr;
+    sockaddr addr;
     socklen_t peer_addr_size = sizeof(addr);
     int peer_fd;
-    while ((peer_fd = accept(listen_fd, (sockaddr*)&addr, &peer_addr_size)) >= 0)
+    while ((peer_fd = accept(listen_fd, &addr, &peer_addr_size)) >= 0)
     {
         assert(peer_fd != 0);
-        char peer_str[256];
-        fprintf(stderr, "[OSD %lu] new client %d: connection from %s port %d\n", this->osd_num, peer_fd,
-            inet_ntop(AF_INET, &addr.sin_addr, peer_str, 256), ntohs(addr.sin_port));
+        fprintf(stderr, "[OSD %lu] new client %d: connection from %s\n", this->osd_num, peer_fd,
+            addr_to_string(addr).c_str());
         fcntl(peer_fd, F_SETFL, fcntl(peer_fd, F_GETFL, 0) | O_NONBLOCK);
         int one = 1;
         setsockopt(peer_fd, SOL_TCP, TCP_NODELAY, &one, sizeof(one));
         clients[peer_fd] = new osd_client_t();
         clients[peer_fd]->peer_addr = addr;
-        clients[peer_fd]->peer_port = ntohs(addr.sin_port);
+        clients[peer_fd]->peer_port = ntohs(((sockaddr_in*)&addr)->sin_port);
         clients[peer_fd]->peer_fd = peer_fd;
         clients[peer_fd]->peer_state = PEER_CONNECTED;
         clients[peer_fd]->in_buf = malloc_or_die(receive_buffer_size);
@@ -547,7 +545,7 @@ json11::Json osd_messenger_t::read_config(const json11::Json & config)
     int done = 0;
     while (done < st.st_size)
     {
-        int r = read(fd, (void*)buf.data()+done, st.st_size-done);
+        int r = read(fd, (uint8_t*)buf.data()+done, st.st_size-done);
         if (r < 0)
         {
             fprintf(stderr, "Error reading %s: %s\n", config_path, strerror(errno));

src/messenger.h

@@ -49,7 +49,7 @@ struct osd_client_t
 {
     int refs = 0;
 
-    sockaddr_in peer_addr;
+    sockaddr peer_addr;
     int peer_port;
     int peer_fd;
     int peer_state;

src/msgr_op.h

@@ -141,7 +141,7 @@ struct osd_op_buf_list_t
             else
             {
                 iov.iov_len -= result;
-                iov.iov_base += result;
+                iov.iov_base = (uint8_t*)iov.iov_base + result;
                 break;
             }
         }

src/msgr_rdma.cpp

@@ -58,7 +58,9 @@ msgr_rdma_context_t *msgr_rdma_context_t::create(const char *ib_devname, uint8_t
     msgr_rdma_context_t *ctx = new msgr_rdma_context_t();
     ctx->mtu = mtu;
 
-    srand48(time(NULL));
+    timespec tv;
+    clock_gettime(CLOCK_REALTIME, &tv);
+    srand48(tv.tv_sec*1000000000 + tv.tv_nsec);
     dev_list = ibv_get_device_list(NULL);
     if (!dev_list)
     {
@@ -389,7 +391,7 @@ bool osd_messenger_t::try_send_rdma(osd_client_t *cl)
         uint32_t len = (uint32_t)(op_size+iov.iov_len-rc->send_buf_pos < rc->max_msg
             ? iov.iov_len-rc->send_buf_pos : rc->max_msg-op_size);
         sge[op_sge++] = {
-            .addr = (uintptr_t)(iov.iov_base+rc->send_buf_pos),
+            .addr = (uintptr_t)((uint8_t*)iov.iov_base+rc->send_buf_pos),
             .length = len,
             .lkey = rc->ctx->mr->lkey,
         };
@@ -519,7 +521,7 @@ void osd_messenger_t::handle_rdma_events()
                     }
                     if (cl->rdma_conn->send_buf_pos > 0)
                     {
-                        cl->send_list[0].iov_base += cl->rdma_conn->send_buf_pos;
+                        cl->send_list[0].iov_base = (uint8_t*)cl->send_list[0].iov_base + cl->rdma_conn->send_buf_pos;
                         cl->send_list[0].iov_len -= cl->rdma_conn->send_buf_pos;
                         cl->rdma_conn->send_buf_pos = 0;
                     }

src/msgr_receive.cpp

@@ -67,7 +67,7 @@ bool osd_messenger_t::handle_read(int result, osd_client_t *cl)
         }
         return false;
     }
-    if (result <= 0 && result != -EAGAIN)
+    if (result <= 0 && result != -EAGAIN && result != -EINTR)
     {
         // this is a client socket, so don't panic on error. just disconnect it
         if (result != 0)
@@ -77,7 +77,7 @@ bool osd_messenger_t::handle_read(int result, osd_client_t *cl)
         stop_client(cl->peer_fd);
         return false;
     }
-    if (result == -EAGAIN || result < cl->read_iov.iov_len)
+    if (result == -EAGAIN || result == -EINTR || result < cl->read_iov.iov_len)
     {
         cl->read_ready--;
         if (cl->read_ready > 0)
@@ -142,13 +142,13 @@ bool osd_messenger_t::handle_read_buffer(osd_client_t *cl, void *curbuf, int rem
                 memcpy(cur->iov_base, curbuf, remain);
                 cl->read_remaining -= remain;
                 cur->iov_len -= remain;
-                cur->iov_base += remain;
+                cur->iov_base = (uint8_t*)cur->iov_base + remain;
                 remain = 0;
             }
             else
             {
                 memcpy(cur->iov_base, curbuf, cur->iov_len);
-                curbuf += cur->iov_len;
+                curbuf = (uint8_t*)curbuf + cur->iov_len;
                 cl->read_remaining -= cur->iov_len;
                 remain -= cur->iov_len;
                 cur->iov_len = 0;
@@ -390,7 +390,7 @@ void osd_messenger_t::handle_reply_ready(osd_op_t *op)
         (tv_end.tv_sec - op->tv_begin.tv_sec)*1000000 +
         (tv_end.tv_nsec - op->tv_begin.tv_nsec)/1000
     );
-    set_immediate.push_back([this, op]()
+    set_immediate.push_back([op]()
     {
         // Copy lambda to be unaffected by `delete op`
         std::function<void(osd_op_t*)>(op->callback)(op);

src/msgr_send.cpp

@@ -224,7 +224,7 @@ void osd_messenger_t::handle_send(int result, osd_client_t *cl)
         }
         return;
     }
-    if (result < 0 && result != -EAGAIN)
+    if (result < 0 && result != -EAGAIN && result != -EINTR)
     {
         // this is a client socket, so don't panic. just disconnect it
         fprintf(stderr, "Client %d socket write error: %d (%s). Disconnecting client\n", cl->peer_fd, -result, strerror(-result));
@@ -250,7 +250,7 @@ void osd_messenger_t::handle_send(int result, osd_client_t *cl)
             else
             {
                 iov.iov_len -= result;
-                iov.iov_base += result;
+                iov.iov_base = (uint8_t*)iov.iov_base + result;
                 break;
             }
         }

src/nbd_proxy.cpp

@@ -30,6 +30,9 @@ protected:
     std::string image_name;
     uint64_t inode = 0;
     uint64_t device_size = 0;
+    int nbd_timeout = 30;
+    int nbd_max_devices = 64;
+    int nbd_max_part = 3;
     inode_watch_t *watch = NULL;
 
     ring_loop_t *ringloop = NULL;
@@ -117,9 +120,18 @@ public:
             "Vitastor NBD proxy\n"
             "(c) Vitaliy Filippov, 2020-2021 (VNPL-1.1)\n\n"
             "USAGE:\n"
-            "  %s map [--etcd_address <etcd_address>] (--image <image> | --pool <pool> --inode <inode> --size <size in bytes>)\n"
+            "  %s map [OPTIONS] (--image <image> | --pool <pool> --inode <inode> --size <size in bytes>)\n"
             "  %s unmap /dev/nbd0\n"
-            "  %s ls [--json]\n",
+            "  %s ls [--json]\n"
+            "OPTIONS:\n"
+            "  All usual Vitastor config options like --etcd_address <etcd_address> plus NBD-specific:\n"
+            "  --nbd_timeout 30\n"
+            "    timeout in seconds after which the kernel will stop the device\n"
+            "    you can set it to 0, but beware that you won't be able to stop the device at all\n"
+            "    if vitastor-nbd process dies\n"
+            "  --nbd_max_devices 64 --nbd_max_part 3\n"
+            "    options for the \"nbd\" kernel module when modprobing it (nbds_max and max_part).\n"
+            "    note that maximum allowed (nbds_max)*(1+max_part) is 256.\n",
             exe_name, exe_name, exe_name
         );
         exit(0);
@@ -174,6 +186,18 @@ public:
                 exit(1);
             }
         }
+        if (cfg["nbd_max_devices"].is_number() || cfg["nbd_max_devices"].is_string())
+        {
+            nbd_max_devices = cfg["nbd_max_devices"].uint64_value();
+        }
+        if (cfg["nbd_max_part"].is_number() || cfg["nbd_max_part"].is_string())
+        {
+            nbd_max_part = cfg["nbd_max_part"].uint64_value();
+        }
+        if (cfg["nbd_timeout"].is_number() || cfg["nbd_timeout"].is_string())
+        {
+            nbd_timeout = cfg["nbd_timeout"].uint64_value();
+        }
         // Create client
         ringloop = new ring_loop_t(512);
         epmgr = new epoll_manager_t(ringloop);
@@ -190,6 +214,12 @@ public:
             }
             watch = cli->st_cli.watch_inode(image_name);
             device_size = watch->cfg.size;
+            if (!watch->cfg.num || !device_size)
+            {
+                // Image does not exist
+                fprintf(stderr, "Image %s does not exist\n", image_name.c_str());
+                exit(1);
+            }
         }
         // Initialize NBD
         int sockfd[2];
@@ -204,7 +234,7 @@ public:
         bool bg = cfg["foreground"].is_null();
         if (!cfg["dev_num"].is_null())
         {
-            if (run_nbd(sockfd, cfg["dev_num"].int64_value(), device_size, NBD_FLAG_SEND_FLUSH, 30, bg) < 0)
+            if (run_nbd(sockfd, cfg["dev_num"].int64_value(), device_size, NBD_FLAG_SEND_FLUSH, nbd_timeout, bg) < 0)
             {
                 perror("run_nbd");
                 exit(1);
@@ -278,7 +308,7 @@ public:
         stop = false;
         cluster_op_t *close_sync = new cluster_op_t;
         close_sync->opcode = OSD_OP_SYNC;
-        close_sync->callback = [this, &stop](cluster_op_t *op)
+        close_sync->callback = [&stop](cluster_op_t *op)
         {
             stop = true;
             delete op;
@@ -301,7 +331,10 @@ public:
             return;
         }
         int r;
-        if ((r = system("modprobe nbd")) != 0)
+        // Kernel built-in default is 16 devices with up to 16 partitions per device which is a big shit
+        // 64 also isn't too high, but the possible maximum is nbds_max=256 max_part=0 and it won't reserve
+        // any block device minor numbers for partitions
+        if ((r = system(("modprobe nbd nbds_max="+std::to_string(nbd_max_devices)+" max_part="+std::to_string(nbd_max_part)).c_str())) != 0)
         {
             if (r < 0)
                 perror("Failed to load NBD kernel module");
@@ -553,7 +586,7 @@ protected:
             }
             else
             {
-                send_list[to_eat].iov_base += result;
+                send_list[to_eat].iov_base = (uint8_t*)send_list[to_eat].iov_base + result;
                 send_list[to_eat].iov_len -= result;
                 break;
             }
@@ -627,8 +660,8 @@ protected:
                 memcpy(cur_buf, b, inc);
                 cur_left -= inc;
                 result -= inc;
-                cur_buf += inc;
-                b += inc;
+                cur_buf = (uint8_t*)cur_buf + inc;
+                b = (uint8_t*)b + inc;
             }
             else
             {
@@ -667,7 +700,7 @@ protected:
                 op->offset = be64toh(cur_req.from);
                 op->len = be32toh(cur_req.len);
                 buf = malloc_or_die(sizeof(nbd_reply) + op->len);
-                op->iov.push_back(buf + sizeof(nbd_reply), op->len);
+                op->iov.push_back((uint8_t*)buf + sizeof(nbd_reply), op->len);
             }
             else if (req_type == NBD_CMD_FLUSH)
             {
@@ -695,7 +728,7 @@ protected:
             if (req_type == NBD_CMD_WRITE)
             {
                 cur_op = op;
-                cur_buf = buf + sizeof(nbd_reply);
+                cur_buf = (uint8_t*)buf + sizeof(nbd_reply);
                 cur_left = op->len;
                 read_state = CL_READ_DATA;
             }
@@ -734,5 +767,6 @@ int main(int narg, const char *args[])
     exe_name = args[0];
     nbd_proxy *p = new nbd_proxy();
     p->exec(nbd_proxy::parse_args(narg, args));
+    delete p;
     return 0;
 }

src/osd.cpp

@@ -7,6 +7,7 @@
 #include <netinet/tcp.h>
 #include <arpa/inet.h>
 
+#include "addr_util.h"
 #include "blockstore_impl.h"
 #include "osd_primary.h"
 #include "osd.h"
@@ -156,14 +157,6 @@ void osd_t::parse_config(const json11::Json & config)
 
 void osd_t::bind_socket()
 {
-    listen_fd = socket(AF_INET, SOCK_STREAM, 0);
-    if (listen_fd < 0)
-    {
-        throw std::runtime_error(std::string("socket: ") + strerror(errno));
-    }
-    int enable = 1;
-    setsockopt(listen_fd, SOL_SOCKET, SO_REUSEADDR, &enable, sizeof(enable));
-
     if (config["osd_network"].is_string() ||
         config["osd_network"].is_array())
     {
@@ -173,7 +166,7 @@ void osd_t::bind_socket()
         else
             for (auto v: config["osd_network"].array_items())
                 mask.push_back(v.string_value());
-        auto matched_addrs = getifaddr_list(mask, false);
+        auto matched_addrs = getifaddr_list(mask);
         if (matched_addrs.size() > 1)
         {
             fprintf(stderr, "More than 1 address matches requested network(s): %s\n", json11::Json(matched_addrs).dump().c_str());
@@ -192,17 +185,21 @@ void osd_t::bind_socket()
 
     // FIXME Support multiple listening sockets
 
-    sockaddr_in addr;
-    int r;
-    if ((r = inet_pton(AF_INET, bind_address.c_str(), &addr.sin_addr)) != 1)
+    sockaddr addr;
+    if (!string_to_addr(bind_address, 0, bind_port, &addr))
     {
-        close(listen_fd);
-        throw std::runtime_error("bind address "+bind_address+(r == 0 ? " is not valid" : ": no ipv4 support"));
+        throw std::runtime_error("bind address "+bind_address+" is not valid");
     }
-    addr.sin_family = AF_INET;
 
-    addr.sin_port = htons(bind_port);
-    if (bind(listen_fd, (sockaddr*)&addr, sizeof(addr)) < 0)
+    listen_fd = socket(addr.sa_family, SOCK_STREAM, 0);
+    if (listen_fd < 0)
+    {
+        throw std::runtime_error(std::string("socket: ") + strerror(errno));
+    }
+    int enable = 1;
+    setsockopt(listen_fd, SOL_SOCKET, SO_REUSEADDR, &enable, sizeof(enable));
+
+    if (bind(listen_fd, &addr, sizeof(addr)) < 0)
     {
         close(listen_fd);
         throw std::runtime_error(std::string("bind: ") + strerror(errno));
@@ -215,7 +212,7 @@ void osd_t::bind_socket()
             close(listen_fd);
             throw std::runtime_error(std::string("getsockname: ") + strerror(errno));
         }
-        listening_port = ntohs(addr.sin_port);
+        listening_port = ntohs(((sockaddr_in*)&addr)->sin_port);
     }
     else
     {
@@ -334,8 +331,8 @@ void osd_t::exec_op(osd_op_t *cur_op)
 
 void osd_t::reset_stats()
 {
-    msgr.stats = { 0 };
-    prev_stats = { 0 };
+    msgr.stats = {};
+    prev_stats = {};
     memset(recovery_stat_count, 0, sizeof(recovery_stat_count));
     memset(recovery_stat_bytes, 0, sizeof(recovery_stat_bytes));
 }
@@ -450,7 +447,7 @@ void osd_t::print_slow()
                 {
                     for (uint64_t i = 0; i < op->req.sec_stab.len; i += sizeof(obj_ver_id))
                     {
-                        obj_ver_id *ov = (obj_ver_id*)(op->buf + i);
+                        obj_ver_id *ov = (obj_ver_id*)((uint8_t*)op->buf + i);
                         bufprintf(i == 0 ? " %lx:%lx v%lu" : ", %lx:%lx v%lu", ov->oid.inode, ov->oid.stripe, ov->version);
                     }
                 }

src/osd.h

@@ -102,7 +102,7 @@ class osd_t
     bool no_rebalance = false;
     bool no_recovery = false;
     std::string bind_address;
-    int bind_port, listen_backlog;
+    int bind_port, listen_backlog = 128;
     // FIXME: Implement client queue depth limit
     int client_queue_depth = 128;
     bool allow_test_ops = false;
@@ -166,8 +166,8 @@ class osd_t
     osd_op_stats_t prev_stats;
     std::map<uint64_t, inode_stats_t> inode_stats;
     const char* recovery_stat_names[2] = { "degraded", "misplaced" };
-    uint64_t recovery_stat_count[2][2] = { 0 };
-    uint64_t recovery_stat_bytes[2][2] = { 0 };
+    uint64_t recovery_stat_count[2][2] = {};
+    uint64_t recovery_stat_bytes[2][2] = {};
 
     // cluster connection
     void parse_config(const json11::Json & config);

src/osd_cluster.cpp

@@ -6,6 +6,7 @@
 #include "etcd_state_client.h"
 #include "http_client.h"
 #include "osd_rmw.h"
+#include "addr_util.h"
 
 // Startup sequence:
 //   Start etcd watcher -> Load global OSD configuration -> Bind socket -> Acquire lease -> Report&lock OSD state
@@ -276,14 +277,14 @@ void osd_t::report_statistics()
             } }
         });
     }
-    st_cli.etcd_txn(json11::Json::object { { "success", txn } }, ETCD_SLOW_TIMEOUT, [this](std::string err, json11::Json res)
+    st_cli.etcd_txn_slow(json11::Json::object { { "success", txn } }, [this](std::string err, json11::Json res)
     {
         etcd_reporting_stats = false;
         if (err != "")
         {
             printf("[OSD %lu] Error reporting state to etcd: %s\n", this->osd_num, err.c_str());
             // Retry indefinitely
-            tfd->set_timer(ETCD_SLOW_TIMEOUT, false, [this](int timer_id)
+            tfd->set_timer(st_cli.etcd_slow_timeout, false, [this](int timer_id)
             {
                 report_statistics();
             });
@@ -354,13 +355,13 @@ void osd_t::acquire_lease()
 {
     // Maximum lease TTL is (report interval) + retries * (timeout + repeat interval)
     st_cli.etcd_call("/lease/grant", json11::Json::object {
-        { "TTL", etcd_report_interval+(MAX_ETCD_ATTEMPTS*(2*ETCD_QUICK_TIMEOUT)+999)/1000 }
-    }, ETCD_QUICK_TIMEOUT, [this](std::string err, json11::Json data)
+        { "TTL", etcd_report_interval+(st_cli.max_etcd_attempts*(2*st_cli.etcd_quick_timeout)+999)/1000 }
+    }, st_cli.etcd_quick_timeout, 0, 0, [this](std::string err, json11::Json data)
     {
         if (err != "" || data["ID"].string_value() == "")
         {
-            printf("Error acquiring a lease from etcd: %s\n", err.c_str());
-            tfd->set_timer(ETCD_QUICK_TIMEOUT, false, [this](int timer_id)
+            printf("Error acquiring a lease from etcd: %s, retrying\n", err.c_str());
+            tfd->set_timer(st_cli.etcd_quick_timeout, false, [this](int timer_id)
             {
                 acquire_lease();
             });
@@ -407,19 +408,19 @@ void osd_t::create_osd_state()
                 } }
             },
         } },
-    }, ETCD_QUICK_TIMEOUT, [this](std::string err, json11::Json data)
+    }, st_cli.etcd_quick_timeout, 0, 0, [this](std::string err, json11::Json data)
     {
         if (err != "")
         {
             etcd_failed_attempts++;
             printf("Error creating OSD state key: %s\n", err.c_str());
-            if (etcd_failed_attempts > MAX_ETCD_ATTEMPTS)
+            if (etcd_failed_attempts > st_cli.max_etcd_attempts)
             {
                 // Die
                 throw std::runtime_error("Cluster connection failed");
             }
             // Retry
-            tfd->set_timer(ETCD_QUICK_TIMEOUT, false, [this](int timer_id)
+            tfd->set_timer(st_cli.etcd_quick_timeout, false, [this](int timer_id)
             {
                 create_osd_state();
             });
@@ -451,7 +452,7 @@ void osd_t::renew_lease()
 {
     st_cli.etcd_call("/lease/keepalive", json11::Json::object {
         { "ID", etcd_lease_id }
-    }, ETCD_QUICK_TIMEOUT, [this](std::string err, json11::Json data)
+    }, st_cli.etcd_quick_timeout, 0, 0, [this](std::string err, json11::Json data)
     {
         if (err == "" && data["result"]["TTL"].string_value() == "")
         {
@@ -462,13 +463,13 @@ void osd_t::renew_lease()
         {
             etcd_failed_attempts++;
             printf("Error renewing etcd lease: %s\n", err.c_str());
-            if (etcd_failed_attempts > MAX_ETCD_ATTEMPTS)
+            if (etcd_failed_attempts > st_cli.max_etcd_attempts)
             {
                 // Die
                 throw std::runtime_error("Cluster connection failed");
             }
             // Retry
-            tfd->set_timer(ETCD_QUICK_TIMEOUT, false, [this](int timer_id)
+            tfd->set_timer(st_cli.etcd_quick_timeout, false, [this](int timer_id)
             {
                 renew_lease();
             });
@@ -487,7 +488,7 @@ void osd_t::force_stop(int exitcode)
     {
         st_cli.etcd_call("/kv/lease/revoke", json11::Json::object {
             { "ID", etcd_lease_id }
-        }, ETCD_QUICK_TIMEOUT, [this, exitcode](std::string err, json11::Json data)
+        }, st_cli.etcd_quick_timeout, st_cli.max_etcd_attempts, 0, [this, exitcode](std::string err, json11::Json data)
         {
             if (err != "")
             {
@@ -825,7 +826,7 @@ void osd_t::report_pg_states()
     etcd_reporting_pg_state = true;
     st_cli.etcd_txn(json11::Json::object {
         { "compare", checks }, { "success", success }, { "failure", failure }
-    }, ETCD_QUICK_TIMEOUT, [this, reporting_pgs](std::string err, json11::Json data)
+    }, st_cli.etcd_quick_timeout, 0, 0, [this, reporting_pgs](std::string err, json11::Json data)
     {
         etcd_reporting_pg_state = false;
         if (!data["succeeded"].bool_value())
@@ -857,10 +858,13 @@ void osd_t::report_pg_states()
                         if (null_byte == 0)
                         {
                             auto pg_it = pgs.find({ .pool_id = pool_id, .pg_num = pg_num });
-                            if (pg_it != pgs.end() && pg_it->second.state != PG_OFFLINE && pg_it->second.state != PG_STARTING)
+                            if (pg_it != pgs.end() && pg_it->second.state != PG_OFFLINE && pg_it->second.state != PG_STARTING &&
+                                kv.value["primary"].uint64_value() != 0 &&
+                                kv.value["primary"].uint64_value() != this->osd_num)
                             {
-                                // Live PG state update failed
-                                printf("Failed to report state of pool %u PG %u which is live. Race condition detected, exiting\n", pool_id, pg_num);
+                                // PG is somehow captured by another OSD
+                                printf("BUG: OSD %lu captured our PG %u/%u. Race condition detected, exiting\n",
+                                    kv.value["primary"].uint64_value(), pool_id, pg_num);
                                 force_stop(1);
                                 return;
                             }

src/osd_peering.cpp

@@ -27,9 +27,9 @@ void osd_t::handle_peers()
                     misplaced_objects += p.second.misplaced_objects.size();
                     // FIXME: degraded objects may currently include misplaced, too! Report them separately?
                     degraded_objects += p.second.degraded_objects.size();
-                    if ((p.second.state & (PG_ACTIVE | PG_HAS_UNCLEAN)) == (PG_ACTIVE | PG_HAS_UNCLEAN))
+                    if (p.second.state & PG_HAS_UNCLEAN)
                         peering_state = peering_state | OSD_FLUSHING_PGS;
-                    else if (p.second.state & PG_ACTIVE)
+                    else if (p.second.state & PG_HAS_DEGRADED)
                         peering_state = peering_state | OSD_RECOVERING;
                 }
                 else
@@ -176,6 +176,17 @@ void osd_t::start_pg_peering(pg_t & pg)
             msgr.stop_client(peer_fd);
         }
     }
+    // Try to connect with current peers if they're up, but we don't have connections to them
+    // Otherwise we may erroneously decide that the pg is incomplete :-)
+    for (auto pg_osd: pg.all_peers)
+    {
+        if (pg_osd != this->osd_num &&
+            msgr.osd_peer_fds.find(pg_osd) == msgr.osd_peer_fds.end() &&
+            msgr.wanted_peers.find(pg_osd) == msgr.wanted_peers.end())
+        {
+            msgr.connect_peer(pg_osd, st_cli.peer_states[pg_osd]);
+        }
+    }
     // Calculate current write OSD set
     pg.pg_cursize = 0;
     pg.cur_set.resize(pg.target_set.size());
@@ -194,6 +205,20 @@ void osd_t::start_pg_peering(pg_t & pg)
             });
         }
     }
+    if (pg.pg_cursize < pg.pg_minsize)
+    {
+        pg.state = PG_INCOMPLETE;
+        report_pg_state(pg);
+        return;
+    }
+    std::set<osd_num_t> cur_peers;
+    for (auto pg_osd: pg.all_peers)
+    {
+        if (pg_osd == this->osd_num || msgr.osd_peer_fds.find(pg_osd) != msgr.osd_peer_fds.end())
+        {
+            cur_peers.insert(pg_osd);
+        }
+    }
     if (pg.target_history.size())
     {
         // Refuse to start PG if no peers are available from any of the historical OSD sets
@@ -222,24 +247,6 @@ void osd_t::start_pg_peering(pg_t & pg)
             }
         }
     }
-    if (pg.pg_cursize < pg.pg_minsize)
-    {
-        pg.state = PG_INCOMPLETE;
-        report_pg_state(pg);
-        return;
-    }
-    std::set<osd_num_t> cur_peers;
-    for (auto pg_osd: pg.all_peers)
-    {
-        if (pg_osd == this->osd_num || msgr.osd_peer_fds.find(pg_osd) != msgr.osd_peer_fds.end())
-        {
-            cur_peers.insert(pg_osd);
-        }
-        else if (msgr.wanted_peers.find(pg_osd) == msgr.wanted_peers.end())
-        {
-            msgr.connect_peer(pg_osd, st_cli.peer_states[pg_osd]);
-        }
-    }
     pg.cur_peers.insert(pg.cur_peers.begin(), cur_peers.begin(), cur_peers.end());
     if (pg.peering_state)
     {

src/osd_primary.cpp

@@ -96,11 +96,11 @@ bool osd_t::prepare_primary_rw(osd_op_t *cur_op)
             (pool_cfg.scheme == POOL_SCHEME_REPLICATED ? 0 : pg_it->second.pg_size)
         )
     );
-    void *data_buf = ((void*)op_data) + sizeof(osd_primary_op_data_t);
+    void *data_buf = (uint8_t*)op_data + sizeof(osd_primary_op_data_t);
     op_data->pg_num = pg_num;
     op_data->oid = oid;
     op_data->stripes = (osd_rmw_stripe_t*)data_buf;
-    data_buf += sizeof(osd_rmw_stripe_t) * stripe_count;
+    data_buf = (uint8_t*)data_buf + sizeof(osd_rmw_stripe_t) * stripe_count;
     op_data->scheme = pool_cfg.scheme;
     op_data->pg_data_size = pg_data_size;
     op_data->pg_size = pg_it->second.pg_size;
@@ -110,17 +110,17 @@ bool osd_t::prepare_primary_rw(osd_op_t *cur_op)
     for (int i = 0; i < stripe_count; i++)
     {
         op_data->stripes[i].bmp_buf = data_buf;
-        data_buf += clean_entry_bitmap_size;
+        data_buf = (uint8_t*)data_buf + clean_entry_bitmap_size;
     }
     op_data->chain_size = chain_size;
     if (chain_size > 0)
     {
         op_data->read_chain = (inode_t*)data_buf;
-        data_buf += sizeof(inode_t) * chain_size;
+        data_buf = (uint8_t*)data_buf + sizeof(inode_t) * chain_size;
         op_data->snapshot_bitmaps = data_buf;
-        data_buf += chain_size * stripe_count * clean_entry_bitmap_size;
+        data_buf = (uint8_t*)data_buf + chain_size * stripe_count * clean_entry_bitmap_size;
         op_data->missing_flags = (uint8_t*)data_buf;
-        data_buf += chain_size * (pool_cfg.scheme == POOL_SCHEME_REPLICATED ? 0 : pg_it->second.pg_size);
+        data_buf = (uint8_t*)data_buf + chain_size * (pool_cfg.scheme == POOL_SCHEME_REPLICATED ? 0 : pg_it->second.pg_size);
         // Copy chain
         int chain_num = 0;
         op_data->read_chain[chain_num++] = cur_op->req.rw.inode;
@@ -248,7 +248,7 @@ resume_2:
             {
                 // Send buffer in parts to avoid copying
                 cur_op->iov.push_back(
-                    stripes[role].read_buf + (stripes[role].req_start - stripes[role].read_start),
+                    (uint8_t*)stripes[role].read_buf + (stripes[role].req_start - stripes[role].read_start),
                     stripes[role].req_end - stripes[role].req_start
                 );
             }

src/osd_primary_chain.cpp

@@ -66,7 +66,7 @@ int osd_t::read_bitmaps(osd_op_t *cur_op, pg_t & pg, int base_state)
             auto read_version = (vo_it != pg.ver_override.end() ? vo_it->second : UINT64_MAX);
             // Read bitmap synchronously from the local database
             bs->read_bitmap(
-                cur_oid, read_version, op_data->snapshot_bitmaps + chain_num*clean_entry_bitmap_size,
+                cur_oid, read_version, (uint8_t*)op_data->snapshot_bitmaps + chain_num*clean_entry_bitmap_size,
                 !chain_num ? &cur_op->reply.rw.version : NULL
             );
         }
@@ -96,12 +96,15 @@ resume_1:
                 {
                     if (op_data->missing_flags[chain_num*pg.pg_size + i])
                     {
-                        osd_rmw_stripe_t local_stripes[pg.pg_size] = { 0 };
+                        osd_rmw_stripe_t local_stripes[pg.pg_size];
                         for (i = 0; i < pg.pg_size; i++)
                         {
-                            local_stripes[i].missing = op_data->missing_flags[chain_num*pg.pg_size + i] && true;
-                            local_stripes[i].bmp_buf = op_data->snapshot_bitmaps + (chain_num*pg.pg_size + i)*clean_entry_bitmap_size;
-                            local_stripes[i].read_start = local_stripes[i].read_end = 1;
+                            local_stripes[i] = (osd_rmw_stripe_t){
+                                .bmp_buf = (uint8_t*)op_data->snapshot_bitmaps + (chain_num*pg.pg_size + i)*clean_entry_bitmap_size,
+                                .read_start = 1,
+                                .read_end = 1,
+                                .missing = op_data->missing_flags[chain_num*pg.pg_size + i] && true,
+                            };
                         }
                         if (pg.scheme == POOL_SCHEME_XOR)
                         {
@@ -146,7 +149,7 @@ int osd_t::collect_bitmap_requests(osd_op_t *cur_op, pg_t & pg, std::vector<bitm
                 .osd_num = read_target,
                 .oid = cur_oid,
                 .version = target_version,
-                .bmp_buf = op_data->snapshot_bitmaps + chain_num*clean_entry_bitmap_size,
+                .bmp_buf = (uint8_t*)op_data->snapshot_bitmaps + chain_num*clean_entry_bitmap_size,
             });
         }
         else
@@ -185,7 +188,7 @@ int osd_t::collect_bitmap_requests(osd_op_t *cur_op, pg_t & pg, std::vector<bitm
                             .stripe = cur_oid.stripe | i,
                         },
                         .version = target_version,
-                        .bmp_buf = op_data->snapshot_bitmaps + (chain_num*pg.pg_size + i)*clean_entry_bitmap_size,
+                        .bmp_buf = (uint8_t*)op_data->snapshot_bitmaps + (chain_num*pg.pg_size + i)*clean_entry_bitmap_size,
                     });
                     found++;
                 }
@@ -204,6 +207,7 @@ int osd_t::submit_bitmap_subops(osd_op_t *cur_op, pg_t & pg)
     std::vector<bitmap_request_t> *bitmap_requests = new std::vector<bitmap_request_t>();
     if (collect_bitmap_requests(cur_op, pg, *bitmap_requests) < 0)
     {
+        delete bitmap_requests;
         return -1;
     }
     op_data->n_subops = 0;
@@ -266,15 +270,15 @@ int osd_t::submit_bitmap_subops(osd_op_t *cur_op, pg_t & pg)
                     int requested_count = subop->req.sec_read_bmp.len / sizeof(obj_ver_id);
                     if (subop->reply.hdr.retval == requested_count * (8 + clean_entry_bitmap_size))
                     {
-                        void *cur_buf = subop->buf + 8;
+                        void *cur_buf = (uint8_t*)subop->buf + 8;
                         for (int j = prev; j <= i; j++)
                         {
                             memcpy((*bitmap_requests)[j].bmp_buf, cur_buf, clean_entry_bitmap_size);
                             if ((*bitmap_requests)[j].oid.inode == cur_op->req.rw.inode)
                             {
-                                memcpy(&cur_op->reply.rw.version, cur_buf-8, 8);
+                                memcpy(&cur_op->reply.rw.version, (uint8_t*)cur_buf-8, 8);
                             }
-                            cur_buf += 8 + clean_entry_bitmap_size;
+                            cur_buf = (uint8_t*)cur_buf + 8 + clean_entry_bitmap_size;
                         }
                     }
                     if ((cur_op->op_data->errors + cur_op->op_data->done + 1) >= cur_op->op_data->n_subops)
@@ -363,7 +367,7 @@ int osd_t::submit_chained_read_requests(pg_t & pg, osd_op_t *cur_op)
         + sizeof(osd_rmw_stripe_t) * stripe_count * op_data->chain_size
     );
     osd_rmw_stripe_t *chain_stripes = (osd_rmw_stripe_t*)(
-        ((void*)op_data->chain_reads) + sizeof(osd_chain_read_t) * op_data->chain_read_count
+        (uint8_t*)op_data->chain_reads + sizeof(osd_chain_read_t) * op_data->chain_read_count
     );
     // Now process each subrequest as a separate read, including reconstruction if needed
     // Prepare reads
@@ -425,8 +429,8 @@ int osd_t::submit_chained_read_requests(pg_t & pg, osd_op_t *cur_op)
             if (stripes[role].read_end > 0)
             {
                 stripes[role].read_buf = cur_buf;
-                stripes[role].bmp_buf = op_data->snapshot_bitmaps + (chain_reads[cri].chain_pos*stripe_count + role)*clean_entry_bitmap_size;
-                cur_buf += stripes[role].read_end - stripes[role].read_start;
+                stripes[role].bmp_buf = (uint8_t*)op_data->snapshot_bitmaps + (chain_reads[cri].chain_pos*stripe_count + role)*clean_entry_bitmap_size;
+                cur_buf = (uint8_t*)cur_buf + stripes[role].read_end - stripes[role].read_start;
             }
         }
     }
@@ -474,7 +478,7 @@ void osd_t::send_chained_read_results(pg_t & pg, osd_op_t *cur_op)
     osd_primary_op_data_t *op_data = cur_op->op_data;
     int stripe_count = (pg.scheme == POOL_SCHEME_REPLICATED ? 1 : pg.pg_size);
     osd_rmw_stripe_t *chain_stripes = (osd_rmw_stripe_t*)(
-        ((void*)op_data->chain_reads) + sizeof(osd_chain_read_t) * op_data->chain_read_count
+        (uint8_t*)op_data->chain_reads + sizeof(osd_chain_read_t) * op_data->chain_read_count
     );
     // Reconstruct parts if needed
     if (op_data->degraded)
@@ -544,7 +548,7 @@ void osd_t::send_chained_read_results(pg_t & pg, osd_op_t *cur_op)
                             role_end = bs_block_size;
                         assert(stripes[role].read_buf);
                         cur_op->iov.push_back(
-                            stripes[role].read_buf + (role_start - stripes[role].read_start),
+                            (uint8_t*)stripes[role].read_buf + (role_start - stripes[role].read_start),
                             role_end - role_start
                         );
                         sent += role_end - role_start;

src/osd_primary_sync.cpp

@@ -86,7 +86,7 @@ resume_2:
             sizeof(obj_ver_osd_t)*this->copies_to_delete_after_sync_count
         );
         op_data->dirty_pgs = (pool_pg_num_t*)dirty_buf;
-        op_data->dirty_osds = (osd_num_t*)(dirty_buf + sizeof(pool_pg_num_t)*dirty_pgs.size());
+        op_data->dirty_osds = (osd_num_t*)((uint8_t*)dirty_buf + sizeof(pool_pg_num_t)*dirty_pgs.size());
         op_data->dirty_pg_count = dirty_pgs.size();
         op_data->dirty_osd_count = dirty_osds.size();
         if (this->copies_to_delete_after_sync_count)

src/osd_primary_write.cpp

@@ -113,7 +113,7 @@ resume_3:
             op_data->stripes[0].write_end != bs_block_size))
         {
             memcpy(
-                op_data->stripes[0].read_buf + op_data->stripes[0].req_start,
+                (uint8_t*)op_data->stripes[0].read_buf + op_data->stripes[0].req_start,
                 op_data->stripes[0].write_buf,
                 op_data->stripes[0].req_end - op_data->stripes[0].req_start
             );

src/osd_rmw.cpp

@@ -103,8 +103,8 @@ void reconstruct_stripes_xor(osd_rmw_stripe_t *stripes, int pg_size, uint32_t bi
                         assert(stripes[role].read_start >= stripes[prev].read_start &&
                             stripes[role].read_start >= stripes[other].read_start);
                         memxor(
-                            stripes[prev].read_buf + (stripes[role].read_start - stripes[prev].read_start),
-                            stripes[other].read_buf + (stripes[role].read_start - stripes[other].read_start),
+                            (uint8_t*)stripes[prev].read_buf + (stripes[role].read_start - stripes[prev].read_start),
+                            (uint8_t*)stripes[other].read_buf + (stripes[role].read_start - stripes[other].read_start),
                             stripes[role].read_buf, stripes[role].read_end - stripes[role].read_start
                         );
                         memxor(stripes[prev].bmp_buf, stripes[other].bmp_buf, stripes[role].bmp_buf, bitmap_size);
@@ -115,7 +115,7 @@ void reconstruct_stripes_xor(osd_rmw_stripe_t *stripes, int pg_size, uint32_t bi
                         assert(stripes[role].read_start >= stripes[other].read_start);
                         memxor(
                             stripes[role].read_buf,
-                            stripes[other].read_buf + (stripes[role].read_start - stripes[other].read_start),
+                            (uint8_t*)stripes[other].read_buf + (stripes[role].read_start - stripes[other].read_start),
                             stripes[role].read_buf, stripes[role].read_end - stripes[role].read_start
                         );
                         memxor(stripes[role].bmp_buf, stripes[other].bmp_buf, stripes[role].bmp_buf, bitmap_size);
@@ -202,10 +202,9 @@ reed_sol_matrix_t* get_jerasure_matrix(int pg_size, int pg_minsize)
 int* get_jerasure_decoding_matrix(osd_rmw_stripe_t *stripes, int pg_size, int pg_minsize)
 {
     int edd = 0;
-    int erased[pg_size] = { 0 };
+    int erased[pg_size];
     for (int i = 0; i < pg_size; i++)
-        if (stripes[i].read_end == 0 || stripes[i].missing)
-            erased[i] = 1;
+        erased[i] = (stripes[i].read_end == 0 || stripes[i].missing ? 1 : 0);
     for (int i = 0; i < pg_minsize; i++)
         if (stripes[i].read_end != 0 && stripes[i].missing)
             edd++;
@@ -241,7 +240,9 @@ void reconstruct_stripes_jerasure(osd_rmw_stripe_t *stripes, int pg_size, int pg
         return;
     }
     int *decoding_matrix = dm_ids + pg_minsize;
-    char *data_ptrs[pg_size] = { 0 };
+    char *data_ptrs[pg_size];
+    for (int role = 0; role < pg_size; role++)
+        data_ptrs[role] = NULL;
     for (int role = 0; role < pg_minsize; role++)
     {
         if (stripes[role].read_end != 0 && stripes[role].missing)
@@ -254,7 +255,7 @@ void reconstruct_stripes_jerasure(osd_rmw_stripe_t *stripes, int pg_size, int pg
                     {
                         assert(stripes[other].read_start <= stripes[role].read_start);
                         assert(stripes[other].read_end >= stripes[role].read_end);
-                        data_ptrs[other] = (char*)(stripes[other].read_buf + (stripes[role].read_start - stripes[other].read_start));
+                        data_ptrs[other] = (char*)stripes[other].read_buf + (stripes[role].read_start - stripes[other].read_start);
                     }
                 }
                 data_ptrs[role] = (char*)stripes[role].read_buf;
@@ -330,7 +331,7 @@ void* alloc_read_buffer(osd_rmw_stripe_t *stripes, int read_pg_size, uint64_t ad
     {
         if (stripes[role].read_end != 0)
         {
-            stripes[role].read_buf = buf + buf_pos;
+            stripes[role].read_buf = (uint8_t*)buf + buf_pos;
             buf_pos += stripes[role].read_end - stripes[role].read_start;
         }
     }
@@ -446,12 +447,12 @@ void* calc_rmw(void *request_buf, osd_rmw_stripe_t *stripes, uint64_t *read_osd_
     {
         if (stripes[role].req_end != 0)
         {
-            stripes[role].write_buf = request_buf + in_pos;
+            stripes[role].write_buf = (uint8_t*)request_buf + in_pos;
             in_pos += stripes[role].req_end - stripes[role].req_start;
         }
         else if (role >= pg_minsize && write_osd_set[role] != 0 && end != 0)
         {
-            stripes[role].write_buf = rmw_buf + buf_pos;
+            stripes[role].write_buf = (uint8_t*)rmw_buf + buf_pos;
             buf_pos += end - start;
         }
     }
@@ -476,13 +477,13 @@ static void get_old_new_buffers(osd_rmw_stripe_t & stripe, uint32_t wr_start, ui
     if (ne && (!oe || ns <= os))
     {
         // NEW or NEW->OLD
-        bufs[nbufs++] = { .buf = stripe.write_buf + ns - stripe.req_start, .len = ne-ns };
+        bufs[nbufs++] = { .buf = (uint8_t*)stripe.write_buf + ns - stripe.req_start, .len = ne-ns };
         if (os < ne)
             os = ne;
         if (oe > os)
         {
             // NEW->OLD
-            bufs[nbufs++] = { .buf = stripe.read_buf + os - stripe.read_start, .len = oe-os };
+            bufs[nbufs++] = { .buf = (uint8_t*)stripe.read_buf + os - stripe.read_start, .len = oe-os };
         }
     }
     else if (oe)
@@ -491,18 +492,18 @@ static void get_old_new_buffers(osd_rmw_stripe_t & stripe, uint32_t wr_start, ui
         if (ne)
         {
             // OLD->NEW or OLD->NEW->OLD
-            bufs[nbufs++] = { .buf = stripe.read_buf + os - stripe.read_start, .len = ns-os };
-            bufs[nbufs++] = { .buf = stripe.write_buf + ns - stripe.req_start, .len = ne-ns };
+            bufs[nbufs++] = { .buf = (uint8_t*)stripe.read_buf + os - stripe.read_start, .len = ns-os };
+            bufs[nbufs++] = { .buf = (uint8_t*)stripe.write_buf + ns - stripe.req_start, .len = ne-ns };
             if (oe > ne)
             {
                 // OLD->NEW->OLD
-                bufs[nbufs++] = { .buf = stripe.read_buf + ne - stripe.read_start, .len = oe-ne };
+                bufs[nbufs++] = { .buf = (uint8_t*)stripe.read_buf + ne - stripe.read_start, .len = oe-ne };
             }
         }
         else
         {
             // OLD
-            bufs[nbufs++] = { .buf = stripe.read_buf + os - stripe.read_start, .len = oe-os };
+            bufs[nbufs++] = { .buf = (uint8_t*)stripe.read_buf + os - stripe.read_start, .len = oe-os };
         }
     }
 }
@@ -517,7 +518,7 @@ static void xor_multiple_buffers(buf_len_t *xor1, int n1, buf_len_t *xor2, int n
     {
         // We know for sure that ranges overlap
         uint32_t end = std::min(end1, end2);
-        memxor(xor1[i1].buf + pos-start1, xor2[i2].buf + pos-start2, dest+pos, end-pos);
+        memxor((uint8_t*)xor1[i1].buf + pos-start1, (uint8_t*)xor2[i2].buf + pos-start2, (uint8_t*)dest+pos, end-pos);
         pos = end;
         if (pos >= end1)
         {
@@ -586,7 +587,7 @@ static void calc_rmw_parity_copy_mod(osd_rmw_stripe_t *stripes, int pg_size, int
             {
                 // Copy modified chunk into the read buffer to write it back
                 memcpy(
-                    stripes[role].read_buf + stripes[role].req_start,
+                    (uint8_t*)stripes[role].read_buf + stripes[role].req_start,
                     stripes[role].write_buf,
                     stripes[role].req_end - stripes[role].req_start
                 );
@@ -609,7 +610,7 @@ static void calc_rmw_parity_copy_parity(osd_rmw_stripe_t *stripes, int pg_size,
             {
                 // Copy new parity into the read buffer to write it back
                 memcpy(
-                    stripes[role].read_buf + start,
+                    (uint8_t*)stripes[role].read_buf + start,
                     stripes[role].write_buf,
                     end - start
                 );
@@ -698,9 +699,15 @@ void calc_rmw_parity_jerasure(osd_rmw_stripe_t *stripes, int pg_size, int pg_min
         {
             // Calculate new coding chunks
             buf_len_t bufs[pg_size][3];
-            int nbuf[pg_size] = { 0 }, curbuf[pg_size] = { 0 };
+            int nbuf[pg_size], curbuf[pg_size];
             uint32_t positions[pg_size];
-            void *data_ptrs[pg_size] = { 0 };
+            void *data_ptrs[pg_size];
+            for (int i = 0; i < pg_size; i++)
+            {
+                data_ptrs[i] = NULL;
+                nbuf[i] = 0;
+                curbuf[i] = 0;
+            }
             for (int i = 0; i < pg_minsize; i++)
             {
                 get_old_new_buffers(stripes[i], start, end, bufs[i], nbuf[i]);
@@ -719,7 +726,7 @@ void calc_rmw_parity_jerasure(osd_rmw_stripe_t *stripes, int pg_size, int pg_min
                 {
                     assert(curbuf[i] < nbuf[i]);
                     assert(bufs[i][curbuf[i]].buf);
-                    data_ptrs[i] = bufs[i][curbuf[i]].buf + pos-positions[i];
+                    data_ptrs[i] = (uint8_t*)bufs[i][curbuf[i]].buf + pos-positions[i];
                     uint32_t this_end = bufs[i][curbuf[i]].len + positions[i];
                     if (next_end > this_end)
                         next_end = this_end;

src/osd_rmw_test.cpp

@@ -90,7 +90,7 @@ void dump_stripes(osd_rmw_stripe_t *stripes, int pg_size)
 void test1()
 {
     osd_num_t osd_set[3] = { 1, 0, 3 };
-    osd_rmw_stripe_t stripes[3] = { 0 };
+    osd_rmw_stripe_t stripes[3] = {};
     // Test 1.1
     split_stripes(2, 128*1024, 128*1024-4096, 8192, stripes);
     assert(stripes[0].req_start == 128*1024-4096 && stripes[0].req_end == 128*1024);
@@ -129,7 +129,7 @@ void test4()
     const uint32_t bmp = 4;
     unsigned bitmaps[3] = { 0 };
     osd_num_t osd_set[3] = { 1, 0, 3 };
-    osd_rmw_stripe_t stripes[3] = { 0 };
+    osd_rmw_stripe_t stripes[3] = {};
     // Test 4.1
     split_stripes(2, 128*1024, 128*1024-4096, 8192, stripes);
     for (int i = 0; i < 3; i++)
@@ -142,11 +142,11 @@ void test4()
     assert(stripes[0].write_start == 128*1024-4096 && stripes[0].write_end == 128*1024);
     assert(stripes[1].write_start == 0 && stripes[1].write_end == 4096);
     assert(stripes[2].write_start == 0 && stripes[2].write_end == 128*1024);
-    assert(stripes[0].read_buf == rmw_buf+128*1024);
-    assert(stripes[1].read_buf == rmw_buf+128*1024*2);
-    assert(stripes[2].read_buf == rmw_buf+128*1024*3-4096);
+    assert(stripes[0].read_buf == (uint8_t*)rmw_buf+128*1024);
+    assert(stripes[1].read_buf == (uint8_t*)rmw_buf+128*1024*2);
+    assert(stripes[2].read_buf == (uint8_t*)rmw_buf+128*1024*3-4096);
     assert(stripes[0].write_buf == write_buf);
-    assert(stripes[1].write_buf == write_buf+4096);
+    assert(stripes[1].write_buf == (uint8_t*)write_buf+4096);
     assert(stripes[2].write_buf == rmw_buf);
     // Test 4.2
     set_pattern(write_buf, 8192, PATTERN0);
@@ -183,7 +183,7 @@ void test4()
 void test5()
 {
     osd_num_t osd_set[3] = { 1, 0, 3 };
-    osd_rmw_stripe_t stripes[3] = { 0 };
+    osd_rmw_stripe_t stripes[3] = {};
     // Test 5.1
     split_stripes(2, 128*1024, 0, 64*1024*3, stripes);
     assert(stripes[0].req_start == 0 && stripes[0].req_end == 128*1024);
@@ -198,11 +198,11 @@ void test5()
     assert(stripes[0].write_start == 0 && stripes[0].write_end == 128*1024);
     assert(stripes[1].write_start == 0 && stripes[1].write_end == 64*1024);
     assert(stripes[2].write_start == 0 && stripes[2].write_end == 128*1024);
-    assert(stripes[0].read_buf == rmw_buf+128*1024);
-    assert(stripes[1].read_buf == rmw_buf+64*3*1024);
-    assert(stripes[2].read_buf == rmw_buf+64*4*1024);
+    assert(stripes[0].read_buf == (uint8_t*)rmw_buf+128*1024);
+    assert(stripes[1].read_buf == (uint8_t*)rmw_buf+64*3*1024);
+    assert(stripes[2].read_buf == (uint8_t*)rmw_buf+64*4*1024);
     assert(stripes[0].write_buf == write_buf);
-    assert(stripes[1].write_buf == write_buf+128*1024);
+    assert(stripes[1].write_buf == (uint8_t*)write_buf+128*1024);
     assert(stripes[2].write_buf == rmw_buf);
     free(rmw_buf);
     free(write_buf);
@@ -224,7 +224,7 @@ void test5()
 void test6()
 {
     osd_num_t osd_set[3] = { 1, 2, 3 };
-    osd_rmw_stripe_t stripes[3] = { 0 };
+    osd_rmw_stripe_t stripes[3] = {};
     // Test 6.1
     split_stripes(2, 128*1024, 0, 64*1024*3, stripes);
     void *write_buf = malloc(64*1024*3);
@@ -236,10 +236,10 @@ void test6()
     assert(stripes[1].write_start == 0 && stripes[1].write_end == 64*1024);
     assert(stripes[2].write_start == 0 && stripes[2].write_end == 128*1024);
     assert(stripes[0].read_buf == 0);
-    assert(stripes[1].read_buf == rmw_buf+128*1024);
+    assert(stripes[1].read_buf == (uint8_t*)rmw_buf+128*1024);
     assert(stripes[2].read_buf == 0);
     assert(stripes[0].write_buf == write_buf);
-    assert(stripes[1].write_buf == write_buf+128*1024);
+    assert(stripes[1].write_buf == (uint8_t*)write_buf+128*1024);
     assert(stripes[2].write_buf == rmw_buf);
     free(rmw_buf);
     free(write_buf);
@@ -267,7 +267,7 @@ void test7()
 {
     osd_num_t osd_set[3] = { 1, 0, 3 };
     osd_num_t write_osd_set[3] = { 1, 2, 3 };
-    osd_rmw_stripe_t stripes[3] = { 0 };
+    osd_rmw_stripe_t stripes[3] = {};
     // Test 7.1
     split_stripes(2, 128*1024, 128*1024-4096, 8192, stripes);
     void *write_buf = malloc(8192);
@@ -278,11 +278,11 @@ void test7()
     assert(stripes[0].write_start == 128*1024-4096 && stripes[0].write_end == 128*1024);
     assert(stripes[1].write_start == 0 && stripes[1].write_end == 4096);
     assert(stripes[2].write_start == 0 && stripes[2].write_end == 128*1024);
-    assert(stripes[0].read_buf == rmw_buf+128*1024);
-    assert(stripes[1].read_buf == rmw_buf+128*1024*2);
-    assert(stripes[2].read_buf == rmw_buf+128*1024*3);
+    assert(stripes[0].read_buf == (uint8_t*)rmw_buf+128*1024);
+    assert(stripes[1].read_buf == (uint8_t*)rmw_buf+128*1024*2);
+    assert(stripes[2].read_buf == (uint8_t*)rmw_buf+128*1024*3);
     assert(stripes[0].write_buf == write_buf);
-    assert(stripes[1].write_buf == write_buf+4096);
+    assert(stripes[1].write_buf == (uint8_t*)write_buf+4096);
     assert(stripes[2].write_buf == rmw_buf);
     // Test 7.2
     set_pattern(write_buf, 8192, PATTERN0);
@@ -320,7 +320,7 @@ void test8()
 {
     osd_num_t osd_set[3] = { 0, 2, 3 };
     osd_num_t write_osd_set[3] = { 1, 2, 3 };
-    osd_rmw_stripe_t stripes[3] = { 0 };
+    osd_rmw_stripe_t stripes[3] = {};
     // Test 8.1
     split_stripes(2, 128*1024, 0, 128*1024+4096, stripes);
     void *write_buf = malloc(128*1024+4096);
@@ -332,10 +332,10 @@ void test8()
     assert(stripes[1].write_start == 0 && stripes[1].write_end == 4096);
     assert(stripes[2].write_start == 0 && stripes[2].write_end == 128*1024);
     assert(stripes[0].read_buf == NULL);
-    assert(stripes[1].read_buf == rmw_buf+128*1024);
+    assert(stripes[1].read_buf == (uint8_t*)rmw_buf+128*1024);
     assert(stripes[2].read_buf == NULL);
     assert(stripes[0].write_buf == write_buf);
-    assert(stripes[1].write_buf == write_buf+128*1024);
+    assert(stripes[1].write_buf == (uint8_t*)write_buf+128*1024);
     assert(stripes[2].write_buf == rmw_buf);
     // Test 8.2
     set_pattern(write_buf, 128*1024+4096, PATTERN0);
@@ -345,7 +345,7 @@ void test8()
     assert(stripes[1].write_start == 0 && stripes[1].write_end == 4096);     // recheck again
     assert(stripes[2].write_start == 0 && stripes[2].write_end == 128*1024); // recheck again
     assert(stripes[0].write_buf == write_buf);                               // recheck again
-    assert(stripes[1].write_buf == write_buf+128*1024);                      // recheck again
+    assert(stripes[1].write_buf == (uint8_t*)write_buf+128*1024);                      // recheck again
     assert(stripes[2].write_buf == rmw_buf);                                 // recheck again
     check_pattern(stripes[2].write_buf, 4096, 0); // new parity
     check_pattern(stripes[2].write_buf+4096, 128*1024-4096, PATTERN0^PATTERN1); // new parity
@@ -375,7 +375,7 @@ void test9()
 {
     osd_num_t osd_set[3] = { 0, 2, 3 };
     osd_num_t write_osd_set[3] = { 1, 2, 3 };
-    osd_rmw_stripe_t stripes[3] = { 0 };
+    osd_rmw_stripe_t stripes[3] = {};
     // Test 9.0
     split_stripes(2, 128*1024, 64*1024, 0, stripes);
     assert(stripes[0].req_start == 0 && stripes[0].req_end == 0);
@@ -391,8 +391,8 @@ void test9()
     assert(stripes[1].write_start == 0 && stripes[1].write_end == 0);
     assert(stripes[2].write_start == 0 && stripes[2].write_end == 0);
     assert(stripes[0].read_buf == rmw_buf);
-    assert(stripes[1].read_buf == rmw_buf+128*1024);
-    assert(stripes[2].read_buf == rmw_buf+128*1024*2);
+    assert(stripes[1].read_buf == (uint8_t*)rmw_buf+128*1024);
+    assert(stripes[2].read_buf == (uint8_t*)rmw_buf+128*1024*2);
     assert(stripes[0].write_buf == NULL);
     assert(stripes[1].write_buf == NULL);
     assert(stripes[2].write_buf == NULL);
@@ -430,7 +430,7 @@ void test10()
 {
     osd_num_t osd_set[3] = { 1, 0, 0 };
     osd_num_t write_osd_set[3] = { 1, 2, 3 };
-    osd_rmw_stripe_t stripes[3] = { 0 };
+    osd_rmw_stripe_t stripes[3] = {};
     // Test 10.0
     split_stripes(2, 128*1024, 0, 256*1024, stripes);
     assert(stripes[0].req_start == 0 && stripes[0].req_end == 128*1024);
@@ -450,7 +450,7 @@ void test10()
     assert(stripes[1].read_buf == NULL);
     assert(stripes[2].read_buf == NULL);
     assert(stripes[0].write_buf == write_buf);
-    assert(stripes[1].write_buf == write_buf+128*1024);
+    assert(stripes[1].write_buf == (uint8_t*)write_buf+128*1024);
     assert(stripes[2].write_buf == rmw_buf);
     // Test 10.2
     set_pattern(stripes[0].write_buf, 128*1024, PATTERN1);
@@ -460,7 +460,7 @@ void test10()
     assert(stripes[1].write_start == 0 && stripes[1].write_end == 128*1024);
     assert(stripes[2].write_start == 0 && stripes[2].write_end == 128*1024);
     assert(stripes[0].write_buf == write_buf);
-    assert(stripes[1].write_buf == write_buf+128*1024);
+    assert(stripes[1].write_buf == (uint8_t*)write_buf+128*1024);
     assert(stripes[2].write_buf == rmw_buf);
     check_pattern(stripes[2].write_buf, 128*1024, PATTERN1^PATTERN2);
     free(rmw_buf);
@@ -486,7 +486,7 @@ void test11()
 {
     osd_num_t osd_set[3] = { 1, 0, 0 };
     osd_num_t write_osd_set[3] = { 1, 2, 3 };
-    osd_rmw_stripe_t stripes[3] = { 0 };
+    osd_rmw_stripe_t stripes[3] = {};
     // Test 11.0
     split_stripes(2, 128*1024, 128*1024, 256*1024, stripes);
     assert(stripes[0].req_start == 0 && stripes[0].req_end == 0);
@@ -502,7 +502,7 @@ void test11()
     assert(stripes[0].write_start == 0 && stripes[0].write_end == 0);
     assert(stripes[1].write_start == 0 && stripes[1].write_end == 128*1024);
     assert(stripes[2].write_start == 0 && stripes[2].write_end == 128*1024);
-    assert(stripes[0].read_buf == rmw_buf+128*1024);
+    assert(stripes[0].read_buf == (uint8_t*)rmw_buf+128*1024);
     assert(stripes[1].read_buf == NULL);
     assert(stripes[2].read_buf == NULL);
     assert(stripes[0].write_buf == NULL);
@@ -542,7 +542,7 @@ void test12()
 {
     osd_num_t osd_set[3] = { 1, 2, 0 };
     osd_num_t write_osd_set[3] = { 1, 2, 3 };
-    osd_rmw_stripe_t stripes[3] = { 0 };
+    osd_rmw_stripe_t stripes[3] = {};
     // Test 12.0
     split_stripes(2, 128*1024, 0, 0, stripes);
     assert(stripes[0].req_start == 0 && stripes[0].req_end == 0);
@@ -557,8 +557,8 @@ void test12()
     assert(stripes[0].write_start == 0 && stripes[0].write_end == 0);
     assert(stripes[1].write_start == 0 && stripes[1].write_end == 0);
     assert(stripes[2].write_start == 0 && stripes[2].write_end == 128*1024);
-    assert(stripes[0].read_buf == rmw_buf+128*1024);
-    assert(stripes[1].read_buf == rmw_buf+2*128*1024);
+    assert(stripes[0].read_buf == (uint8_t*)rmw_buf+128*1024);
+    assert(stripes[1].read_buf == (uint8_t*)rmw_buf+2*128*1024);
     assert(stripes[2].read_buf == NULL);
     assert(stripes[0].write_buf == NULL);
     assert(stripes[1].write_buf == NULL);
@@ -597,7 +597,7 @@ void test13()
     use_jerasure(4, 2, true);
     osd_num_t osd_set[4] = { 1, 2, 0, 0 };
     osd_num_t write_osd_set[4] = { 1, 2, 3, 4 };
-    osd_rmw_stripe_t stripes[4] = { 0 };
+    osd_rmw_stripe_t stripes[4] = {};
     // Test 13.0
     void *write_buf = malloc_or_die(8192);
     split_stripes(2, 128*1024, 128*1024-4096, 8192, stripes);
@@ -616,14 +616,14 @@ void test13()
     assert(stripes[1].write_start == 0 && stripes[1].write_end == 4096);
     assert(stripes[2].write_start == 0 && stripes[2].write_end == 128*1024);
     assert(stripes[3].write_start == 0 && stripes[3].write_end == 128*1024);
-    assert(stripes[0].read_buf == rmw_buf+2*128*1024);
-    assert(stripes[1].read_buf == rmw_buf+3*128*1024-4096);
+    assert(stripes[0].read_buf == (uint8_t*)rmw_buf+2*128*1024);
+    assert(stripes[1].read_buf == (uint8_t*)rmw_buf+3*128*1024-4096);
     assert(stripes[2].read_buf == NULL);
     assert(stripes[3].read_buf == NULL);
     assert(stripes[0].write_buf == write_buf);
-    assert(stripes[1].write_buf == write_buf+4096);
+    assert(stripes[1].write_buf == (uint8_t*)write_buf+4096);
     assert(stripes[2].write_buf == rmw_buf);
-    assert(stripes[3].write_buf == rmw_buf+128*1024);
+    assert(stripes[3].write_buf == (uint8_t*)rmw_buf+128*1024);
     // Test 13.2 - encode
     set_pattern(write_buf, 8192, PATTERN3);
     set_pattern(stripes[0].read_buf, 128*1024-4096, PATTERN1);
@@ -634,9 +634,9 @@ void test13()
     assert(stripes[2].write_start == 0 && stripes[2].write_end == 128*1024);
     assert(stripes[3].write_start == 0 && stripes[3].write_end == 128*1024);
     assert(stripes[0].write_buf == write_buf);
-    assert(stripes[1].write_buf == write_buf+4096);
+    assert(stripes[1].write_buf == (uint8_t*)write_buf+4096);
     assert(stripes[2].write_buf == rmw_buf);
-    assert(stripes[3].write_buf == rmw_buf+128*1024);
+    assert(stripes[3].write_buf == (uint8_t*)rmw_buf+128*1024);
     // Test 13.3 - full decode and verify
     osd_num_t read_osd_set[4] = { 0, 0, 3, 4 };
     memset(stripes, 0, sizeof(stripes));
@@ -658,11 +658,11 @@ void test13()
     void *read_buf = alloc_read_buffer(stripes, 4, 0);
     assert(read_buf);
     assert(stripes[0].read_buf == read_buf);
-    assert(stripes[1].read_buf == read_buf+128*1024);
-    assert(stripes[2].read_buf == read_buf+2*128*1024);
-    assert(stripes[3].read_buf == read_buf+3*128*1024);
-    memcpy(read_buf+2*128*1024, rmw_buf, 128*1024);
-    memcpy(read_buf+3*128*1024, rmw_buf+128*1024, 128*1024);
+    assert(stripes[1].read_buf == (uint8_t*)read_buf+128*1024);
+    assert(stripes[2].read_buf == (uint8_t*)read_buf+2*128*1024);
+    assert(stripes[3].read_buf == (uint8_t*)read_buf+3*128*1024);
+    memcpy((uint8_t*)read_buf+2*128*1024, rmw_buf, 128*1024);
+    memcpy((uint8_t*)read_buf+3*128*1024, (uint8_t*)rmw_buf+128*1024, 128*1024);
     reconstruct_stripes_jerasure(stripes, 4, 2, 0);
     check_pattern(stripes[0].read_buf, 128*1024-4096, PATTERN1);
     check_pattern(stripes[0].read_buf+128*1024-4096, 4096, PATTERN3);
@@ -690,10 +690,10 @@ void test13()
     assert(read_buf);
     assert(stripes[0].read_buf == read_buf);
     assert(stripes[1].read_buf == NULL);
-    assert(stripes[2].read_buf == read_buf+128*1024);
-    assert(stripes[3].read_buf == read_buf+2*128*1024);
-    memcpy(read_buf+128*1024, rmw_buf, 128*1024);
-    memcpy(read_buf+2*128*1024, rmw_buf+128*1024, 128*1024);
+    assert(stripes[2].read_buf == (uint8_t*)read_buf+128*1024);
+    assert(stripes[3].read_buf == (uint8_t*)read_buf+2*128*1024);
+    memcpy((uint8_t*)read_buf+128*1024, rmw_buf, 128*1024);
+    memcpy((uint8_t*)read_buf+2*128*1024, (uint8_t*)rmw_buf+128*1024, 128*1024);
     reconstruct_stripes_jerasure(stripes, 4, 2, 0);
     check_pattern(stripes[0].read_buf, 128*1024-4096, PATTERN1);
     check_pattern(stripes[0].read_buf+128*1024-4096, 4096, PATTERN3);
@@ -725,7 +725,7 @@ void test14()
     use_jerasure(3, 2, true);
     osd_num_t osd_set[3] = { 1, 2, 0 };
     osd_num_t write_osd_set[3] = { 1, 2, 3 };
-    osd_rmw_stripe_t stripes[3] = { 0 };
+    osd_rmw_stripe_t stripes[3] = {};
     unsigned bitmaps[3] = { 0 };
     // Test 13.0
     void *write_buf = malloc_or_die(8192);
@@ -744,11 +744,11 @@ void test14()
     assert(stripes[0].write_start == 128*1024-4096 && stripes[0].write_end == 128*1024);
     assert(stripes[1].write_start == 0 && stripes[1].write_end == 4096);
     assert(stripes[2].write_start == 0 && stripes[2].write_end == 128*1024);
-    assert(stripes[0].read_buf == rmw_buf+128*1024);
-    assert(stripes[1].read_buf == rmw_buf+2*128*1024-4096);
+    assert(stripes[0].read_buf == (uint8_t*)rmw_buf+128*1024);
+    assert(stripes[1].read_buf == (uint8_t*)rmw_buf+2*128*1024-4096);
     assert(stripes[2].read_buf == NULL);
     assert(stripes[0].write_buf == write_buf);
-    assert(stripes[1].write_buf == write_buf+4096);
+    assert(stripes[1].write_buf == (uint8_t*)write_buf+4096);
     assert(stripes[2].write_buf == rmw_buf);
     // Test 13.2 - encode
     set_pattern(write_buf, 8192, PATTERN3);
@@ -765,7 +765,7 @@ void test14()
     assert(stripes[1].write_start == 0 && stripes[1].write_end == 4096);
     assert(stripes[2].write_start == 0 && stripes[2].write_end == 128*1024);
     assert(stripes[0].write_buf == write_buf);
-    assert(stripes[1].write_buf == write_buf+4096);
+    assert(stripes[1].write_buf == (uint8_t*)write_buf+4096);
     assert(stripes[2].write_buf == rmw_buf);
     // Test 13.3 - decode and verify
     osd_num_t read_osd_set[4] = { 0, 2, 3 };
@@ -788,8 +788,8 @@ void test14()
         stripes[i].bmp_buf = bitmaps+i;
     assert(read_buf);
     assert(stripes[0].read_buf == read_buf);
-    assert(stripes[1].read_buf == read_buf+128*1024);
-    assert(stripes[2].read_buf == read_buf+2*128*1024);
+    assert(stripes[1].read_buf == (uint8_t*)read_buf+128*1024);
+    assert(stripes[2].read_buf == (uint8_t*)read_buf+2*128*1024);
     set_pattern(stripes[1].read_buf, 4096, PATTERN3);
     set_pattern(stripes[1].read_buf+4096, 128*1024-4096, PATTERN2);
     memcpy(stripes[2].read_buf, rmw_buf, 128*1024);

src/osd_secondary.cpp

@@ -54,8 +54,8 @@ void osd_t::exec_secondary(osd_op_t *cur_op)
             void *cur_buf = reply_buf;
             for (int i = 0; i < n; i++)
             {
-                bs->read_bitmap(ov[i].oid, ov[i].version, cur_buf + sizeof(uint64_t), (uint64_t*)cur_buf);
-                cur_buf += (8 + clean_entry_bitmap_size);
+                bs->read_bitmap(ov[i].oid, ov[i].version, (uint8_t*)cur_buf + sizeof(uint64_t), (uint64_t*)cur_buf);
+                cur_buf = (uint8_t*)cur_buf + (8 + clean_entry_bitmap_size);
             }
             free(cur_op->buf);
             cur_op->buf = reply_buf;
@@ -159,7 +159,7 @@ void osd_t::exec_show_config(osd_op_t *cur_op)
         { "readonly", readonly },
         { "immediate_commit", (immediate_commit == IMMEDIATE_ALL ? "all" :
             (immediate_commit == IMMEDIATE_SMALL ? "small" : "none")) },
-        { "lease_timeout", etcd_report_interval+(MAX_ETCD_ATTEMPTS*(2*ETCD_QUICK_TIMEOUT)+999)/1000 },
+        { "lease_timeout", etcd_report_interval+(st_cli.max_etcd_attempts*(2*st_cli.etcd_quick_timeout)+999)/1000 },
     };
 #ifdef WITH_RDMA
     if (msgr.is_rdma_enabled())

src/osd_test.cpp

@@ -16,6 +16,7 @@
 
 #include <stdexcept>
 
+#include "addr_util.h"
 #include "osd_ops.h"
 #include "rw_blocking.h"
 #include "test_pattern.h"
@@ -133,17 +134,14 @@ int main(int narg, char *args[])
 
 int connect_osd(const char *osd_address, int osd_port)
 {
-    struct sockaddr_in addr;
-    int r;
-    if ((r = inet_pton(AF_INET, osd_address, &addr.sin_addr)) != 1)
+    struct sockaddr addr;
+    if (!string_to_addr(osd_address, 0, osd_port, &addr))
     {
-        fprintf(stderr, "server address: %s%s\n", osd_address, r == 0 ? " is not valid" : ": no ipv4 support");
+        fprintf(stderr, "server address: %s is not valid\n", osd_address);
         return -1;
     }
-    addr.sin_family = AF_INET;
-    addr.sin_port = htons(osd_port);
 
-    int connect_fd = socket(AF_INET, SOCK_STREAM, 0);
+    int connect_fd = socket(addr.sa_family, SOCK_STREAM, 0);
     if (connect_fd < 0)
     {
         perror("socket");

src/rw_blocking.cpp

@@ -3,7 +3,10 @@
 
 #include <errno.h>
 #include <stdlib.h>
+#include <stdint.h>
 #include <stdio.h>
+#include <sys/types.h>
+#include <sys/socket.h>
 
 #include "rw_blocking.h"
 
@@ -20,7 +23,7 @@ int read_blocking(int fd, void *read_buf, size_t remaining)
                 // EOF
                 return done;
             }
-            else if (errno != EAGAIN && errno != EPIPE)
+            else if (errno != EINTR && errno != EAGAIN && errno != EPIPE)
             {
                 perror("read");
                 exit(1);
@@ -28,7 +31,7 @@ int read_blocking(int fd, void *read_buf, size_t remaining)
             continue;
         }
         done += r;
-        read_buf += r;
+        read_buf = (uint8_t*)read_buf + r;
     }
     return done;
 }
@@ -41,7 +44,7 @@ int write_blocking(int fd, void *write_buf, size_t remaining)
         size_t r = write(fd, write_buf, remaining-done);
         if (r < 0)
         {
-            if (errno != EAGAIN && errno != EPIPE)
+            if (errno != EINTR && errno != EAGAIN && errno != EPIPE)
             {
                 perror("write");
                 exit(1);
@@ -49,7 +52,7 @@ int write_blocking(int fd, void *write_buf, size_t remaining)
             continue;
         }
         done += r;
-        write_buf += r;
+        write_buf = (uint8_t*)write_buf + r;
     }
     return done;
 }
@@ -60,30 +63,31 @@ int readv_blocking(int fd, iovec *iov, int iovcnt)
     int done = 0;
     while (v < iovcnt)
     {
-        ssize_t r = readv(fd, iov, iovcnt);
+        ssize_t r = readv(fd, iov+v, iovcnt-v);
         if (r < 0)
         {
-            if (errno != EAGAIN && errno != EPIPE)
+            if (errno != EINTR && errno != EAGAIN && errno != EPIPE)
             {
                 perror("writev");
                 exit(1);
             }
             continue;
         }
+        done += r;
         while (v < iovcnt)
         {
             if (iov[v].iov_len > r)
             {
                 iov[v].iov_len -= r;
-                iov[v].iov_base += r;
+                iov[v].iov_base = (uint8_t*)iov[v].iov_base + r;
                 break;
             }
             else
             {
+                r -= iov[v].iov_len;
                 v++;
             }
         }
-        done += r;
     }
     return done;
 }
@@ -94,30 +98,69 @@ int writev_blocking(int fd, iovec *iov, int iovcnt)
     int done = 0;
     while (v < iovcnt)
     {
-        ssize_t r = writev(fd, iov, iovcnt);
+        ssize_t r = writev(fd, iov+v, iovcnt-v);
         if (r < 0)
         {
-            if (errno != EAGAIN && errno != EPIPE)
+            if (errno != EINTR && errno != EAGAIN && errno != EPIPE)
             {
                 perror("writev");
                 exit(1);
             }
             continue;
         }
+        done += r;
         while (v < iovcnt)
         {
             if (iov[v].iov_len > r)
             {
                 iov[v].iov_len -= r;
-                iov[v].iov_base += r;
+                iov[v].iov_base = (uint8_t*)iov[v].iov_base + r;
                 break;
             }
             else
             {
+                r -= iov[v].iov_len;
+                v++;
+            }
+        }
+    }
+    return done;
+}
+
+int sendv_blocking(int fd, iovec *iov, int iovcnt, int flags)
+{
+    struct msghdr msg = { 0 };
+    int v = 0;
+    int done = 0;
+    while (v < iovcnt)
+    {
+        msg.msg_iov = iov+v;
+        msg.msg_iovlen = iovcnt-v;
+        ssize_t r = sendmsg(fd, &msg, flags);
+        if (r < 0)
+        {
+            if (errno != EINTR && errno != EAGAIN && errno != EPIPE)
+            {
+                perror("sendmsg");
+                exit(1);
+            }
+            continue;
+        }
+        done += r;
+        while (v < iovcnt)
+        {
+            if (iov[v].iov_len > r)
+            {
+                iov[v].iov_len -= r;
+                iov[v].iov_base = (uint8_t*)iov[v].iov_base + r;
+                break;
+            }
+            else
+            {
+                r -= iov[v].iov_len;
                 v++;
             }
         }
-        done += r;
     }
     return done;
 }

src/rw_blocking.h

@@ -10,3 +10,4 @@ int read_blocking(int fd, void *read_buf, size_t remaining);
 int write_blocking(int fd, void *write_buf, size_t remaining);
 int readv_blocking(int fd, iovec *iov, int iovcnt);
 int writev_blocking(int fd, iovec *iov, int iovcnt);
+int sendv_blocking(int fd, iovec *iov, int iovcnt, int flags);

src/stub_bench.cpp

@@ -21,6 +21,7 @@
 
 #include <stdexcept>
 
+#include "addr_util.h"
 #include "rw_blocking.h"
 #include "osd_ops.h"
 
@@ -66,16 +67,14 @@ int main(int narg, char *args[])
 
 int connect_stub(const char *server_address, int server_port)
 {
-    struct sockaddr_in addr;
-    int r;
-    if ((r = inet_pton(AF_INET, server_address, &addr.sin_addr)) != 1)
+    struct sockaddr addr;
+    if (!string_to_addr(server_address, 0, server_port, &addr))
     {
-        fprintf(stderr, "server address: %s%s\n", server_address, r == 0 ? " is not valid" : ": no ipv4 support");
+        fprintf(stderr, "server address: %s is not valid\n", server_address);
         return -1;
     }
-    addr.sin_family = AF_INET;
-    addr.sin_port = htons(server_port);
-    int connect_fd = socket(AF_INET, SOCK_STREAM, 0);
+
+    int connect_fd = socket(addr.sa_family, SOCK_STREAM, 0);
     if (connect_fd < 0)
     {
         perror("socket");

src/stub_osd.cpp

@@ -37,10 +37,11 @@
 
 #include <stdexcept>
 
+#include "addr_util.h"
 #include "rw_blocking.h"
 #include "osd_ops.h"
 
-int bind_stub(const char *bind_address, int bind_port);
+int bind_stub(std::string bind_address, int bind_port);
 
 void run_stub(int peer_fd);
 
@@ -48,13 +49,13 @@ int main(int narg, char *args[])
 {
     int listen_fd = bind_stub("0.0.0.0", 11203);
     // Accept new connections
-    sockaddr_in addr;
+    sockaddr addr;
     socklen_t peer_addr_size = sizeof(addr);
     int peer_fd;
     while (1)
     {
         printf("stub_osd: waiting for 1 client\n");
-        peer_fd = accept(listen_fd, (sockaddr*)&addr, &peer_addr_size);
+        peer_fd = accept(listen_fd, &addr, &peer_addr_size);
         if (peer_fd == -1)
         {
             if (errno == EAGAIN)
@@ -62,9 +63,8 @@ int main(int narg, char *args[])
             else
                 throw std::runtime_error(std::string("accept: ") + strerror(errno));
         }
-        char peer_str[256];
-        printf("stub_osd: new client %d: connection from %s port %d\n", peer_fd,
-            inet_ntop(AF_INET, &addr.sin_addr, peer_str, 256), ntohs(addr.sin_port));
+        printf("stub_osd: new client %d: connection from %s\n", peer_fd,
+            addr_to_string(addr).c_str());
         int one = 1;
         setsockopt(peer_fd, SOL_TCP, TCP_NODELAY, &one, sizeof(one));
         run_stub(peer_fd);
@@ -76,11 +76,17 @@ int main(int narg, char *args[])
     return 0;
 }
 
-int bind_stub(const char *bind_address, int bind_port)
+int bind_stub(std::string bind_address, int bind_port)
 {
     int listen_backlog = 128;
 
-    int listen_fd = socket(AF_INET, SOCK_STREAM, 0);
+    sockaddr addr;
+    if (!string_to_addr(bind_address, 0, bind_port, &addr))
+    {
+        throw std::runtime_error("bind address "+bind_address+" is not valid");
+    }
+
+    int listen_fd = socket(addr.sa_family, SOCK_STREAM, 0);
     if (listen_fd < 0)
     {
         throw std::runtime_error(std::string("socket: ") + strerror(errno));
@@ -88,17 +94,7 @@ int bind_stub(const char *bind_address, int bind_port)
     int enable = 1;
     setsockopt(listen_fd, SOL_SOCKET, SO_REUSEADDR, &enable, sizeof(enable));
 
-    sockaddr_in addr;
-    int r;
-    if ((r = inet_pton(AF_INET, bind_address, &addr.sin_addr)) != 1)
-    {
-        close(listen_fd);
-        throw std::runtime_error("bind address "+std::string(bind_address)+(r == 0 ? " is not valid" : ": no ipv4 support"));
-    }
-    addr.sin_family = AF_INET;
-    addr.sin_port = htons(bind_port);
-
-    if (bind(listen_fd, (sockaddr*)&addr, sizeof(addr)) < 0)
+    if (bind(listen_fd, &addr, sizeof(addr)) < 0)
     {
         close(listen_fd);
         throw std::runtime_error(std::string("bind: ") + strerror(errno));
@@ -116,7 +112,7 @@ int bind_stub(const char *bind_address, int bind_port)
 void run_stub(int peer_fd)
 {
     osd_any_op_t op;
-    osd_any_reply_t reply;
+    osd_any_reply_t reply = {};
     void *buf = NULL;
     while (1)
     {
@@ -139,7 +135,7 @@ void run_stub(int peer_fd)
             buf = malloc(op.sec_rw.len);
             r = write_blocking(peer_fd, reply.buf, OSD_PACKET_SIZE);
             if (r == OSD_PACKET_SIZE)
-                r = write_blocking(peer_fd, &buf, op.sec_rw.len);
+                r = write_blocking(peer_fd, buf, op.sec_rw.len);
             free(buf);
             if (r < op.sec_rw.len)
                 break;
@@ -170,5 +166,4 @@ void run_stub(int peer_fd)
             break;
         }
     }
-    free(buf);
 }

src/stub_uring_osd.cpp

@@ -20,11 +20,12 @@
 
 #include <stdexcept>
 
+#include "addr_util.h"
 #include "ringloop.h"
 #include "epoll_manager.h"
 #include "messenger.h"
 
-int bind_stub(const char *bind_address, int bind_port);
+int bind_stub(std::string bind_address, int bind_port);
 
 void stub_exec_op(osd_messenger_t *msgr, osd_op_t *op);
 
@@ -39,6 +40,8 @@ int main(int narg, char *args[])
     msgr->ringloop = ringloop;
     msgr->repeer_pgs = [](osd_num_t) {};
     msgr->exec_op = [msgr](osd_op_t *op) { stub_exec_op(msgr, op); };
+    json11::Json config = json11::Json::object { { "log_level", 1 } };
+    msgr->parse_config(config);
     // Accept new connections
     int listen_fd = bind_stub("0.0.0.0", 11203);
     epmgr->set_fd_handler(listen_fd, false, [listen_fd, msgr](int fd, int events)
@@ -64,11 +67,17 @@ int main(int narg, char *args[])
     return 0;
 }
 
-int bind_stub(const char *bind_address, int bind_port)
+int bind_stub(std::string bind_address, int bind_port)
 {
     int listen_backlog = 128;
 
-    int listen_fd = socket(AF_INET, SOCK_STREAM, 0);
+    sockaddr addr;
+    if (!string_to_addr(bind_address, 0, bind_port, &addr))
+    {
+        throw std::runtime_error("bind address "+bind_address+" is not valid");
+    }
+
+    int listen_fd = socket(addr.sa_family, SOCK_STREAM, 0);
     if (listen_fd < 0)
     {
         throw std::runtime_error(std::string("socket: ") + strerror(errno));
@@ -76,17 +85,7 @@ int bind_stub(const char *bind_address, int bind_port)
     int enable = 1;
     setsockopt(listen_fd, SOL_SOCKET, SO_REUSEADDR, &enable, sizeof(enable));
 
-    sockaddr_in addr;
-    int r;
-    if ((r = inet_pton(AF_INET, bind_address, &addr.sin_addr)) != 1)
-    {
-        close(listen_fd);
-        throw std::runtime_error("bind address "+std::string(bind_address)+(r == 0 ? " is not valid" : ": no ipv4 support"));
-    }
-    addr.sin_family = AF_INET;
-    addr.sin_port = htons(bind_port);
-
-    if (bind(listen_fd, (sockaddr*)&addr, sizeof(addr)) < 0)
+    if (bind(listen_fd, &addr, sizeof(addr)) < 0)
     {
         close(listen_fd);
         throw std::runtime_error(std::string("bind: ") + strerror(errno));
@@ -111,8 +110,10 @@ void stub_exec_op(osd_messenger_t *msgr, osd_op_t *op)
     if (op->req.hdr.opcode == OSD_OP_SEC_READ)
     {
         op->reply.hdr.retval = op->req.sec_rw.len;
-        op->buf = malloc(op->req.sec_rw.len);
+        op->buf = memalign_or_die(MEM_ALIGNMENT, op->req.sec_rw.len);
         op->iov.push_back(op->buf, op->req.sec_rw.len);
+        op->reply.sec_rw.attr_len = 4;
+        op->bitmap = op->buf;
     }
     else if (op->req.hdr.opcode == OSD_OP_SEC_WRITE || op->req.hdr.opcode == OSD_OP_SEC_WRITE_STABLE)
     {

src/test_pattern.h

@@ -11,5 +11,5 @@
 #define PATTERN2 0xffe3bad5f578a78e
 #define PATTERN3 0x426bd7854eb08509
 
-#define set_pattern(buf, len, pattern) for (uint64_t i = 0; i < len; i += 8) { *(uint64_t*)((void*)buf + i) = pattern; }
-#define check_pattern(buf, len, pattern) { uint64_t bad = UINT64_MAX; for (uint64_t i = 0; i < len; i += 8) { if ((*(uint64_t*)(buf + i)) != (pattern)) { bad = i; break; } } if (bad != UINT64_MAX) { printf("mismatch at %lx\n", bad); } assert(bad == UINT64_MAX); }
+#define set_pattern(buf, len, pattern) for (uint64_t i = 0; i < len; i += 8) { *(uint64_t*)((uint8_t*)buf + i) = pattern; }
+#define check_pattern(buf, len, pattern) { uint64_t bad = UINT64_MAX; for (uint64_t i = 0; i < len; i += 8) { if ((*(uint64_t*)((uint8_t*)buf + i)) != (pattern)) { bad = i; break; } } if (bad != UINT64_MAX) { printf("mismatch at %lx\n", bad); } assert(bad == UINT64_MAX); }

src/timerfd_manager.cpp

@@ -94,7 +94,7 @@ again:
     if (!timers.size())
     {
         nearest = -1;
-        itimerspec exp = { 0 };
+        itimerspec exp = {};
         if (timerfd_settime(timerfd, 0, &exp, NULL))
         {
             throw std::runtime_error(std::string("timerfd_settime: ") + strerror(errno));

src/vitastor.pc.in

@@ -6,7 +6,7 @@ includedir=${prefix}/@CMAKE_INSTALL_INCLUDEDIR@
 
 Name: Vitastor
 Description: Vitastor client library
-Version: 0.6.10
+Version: 0.6.12
 Libs: -L${libdir} -lvitastor_client
 Cflags: -I${includedir}
 

tests/common.sh

@@ -23,19 +23,39 @@ trap 'kill -9 $(jobs -p)' EXIT
 ETCD=${ETCD:-etcd}
 ETCD_IP=${ETCD_IP:-127.0.0.1}
 ETCD_PORT=${ETCD_PORT:-12379}
+ETCD_COUNT=${ETCD_COUNT:-1}
 
 if [ "$KEEP_DATA" = "" ]; then
     rm -rf ./testdata
     mkdir -p ./testdata
 fi
 
-$ETCD -name etcd_test --data-dir ./testdata/etcd \
-    --advertise-client-urls http://$ETCD_IP:$ETCD_PORT --listen-client-urls http://$ETCD_IP:$ETCD_PORT \
-    --initial-advertise-peer-urls http://$ETCD_IP:$((ETCD_PORT+1)) --listen-peer-urls http://$ETCD_IP:$((ETCD_PORT+1)) \
-    --max-txn-ops=100000 --auto-compaction-retention=10 --auto-compaction-mode=revision &>./testdata/etcd.log &
-ETCD_PID=$!
-ETCD_URL=$ETCD_IP:$ETCD_PORT/v3
-ETCDCTL="${ETCD}ctl --endpoints=http://$ETCD_URL"
+ETCD_URL="http://$ETCD_IP:$ETCD_PORT"
+ETCD_CLUSTER="etcd1=http://$ETCD_IP:$((ETCD_PORT+1))"
+for i in $(seq 2 $ETCD_COUNT); do
+    ETCD_URL="$ETCD_URL,http://$ETCD_IP:$((ETCD_PORT+2*i-2))"
+    ETCD_CLUSTER="$ETCD_CLUSTER,etcd$i=http://$ETCD_IP:$((ETCD_PORT+2*i-1))"
+done
+ETCDCTL="${ETCD}ctl --endpoints=$ETCD_URL"
+
+start_etcd()
+{
+    local i=$1
+    $ETCD -name etcd$i --data-dir ./testdata/etcd$i \
+        --advertise-client-urls http://$ETCD_IP:$((ETCD_PORT+2*i-2)) --listen-client-urls http://$ETCD_IP:$((ETCD_PORT+2*i-2)) \
+        --initial-advertise-peer-urls http://$ETCD_IP:$((ETCD_PORT+2*i-1)) --listen-peer-urls http://$ETCD_IP:$((ETCD_PORT+2*i-1)) \
+        --initial-cluster-token vitastor-tests-etcd --initial-cluster-state new \
+        --initial-cluster "$ETCD_CLUSTER" \
+        --max-txn-ops=100000 --auto-compaction-retention=10 --auto-compaction-mode=revision &>./testdata/etcd$i.log &
+    eval ETCD${i}_PID=$!
+}
+
+for i in $(seq 1 $ETCD_COUNT); do
+    start_etcd $i
+done
+if [ $ETCD_COUNT -gt 1 ]; then
+    sleep 1
+fi
 
 echo leak:fio >> testdata/lsan-suppress.txt
 echo leak:tcmalloc >> testdata/lsan-suppress.txt

tests/run_3osds.sh

@@ -5,6 +5,7 @@
 OSD_SIZE=${OSD_SIZE:-1024}
 PG_COUNT=${PG_COUNT:-1}
 PG_SIZE=${PG_SIZE:-3}
+PG_MINSIZE=${PG_MINSIZE:-2}
 OSD_COUNT=${OSD_COUNT:-3}
 SCHEME=${SCHEME:-ec}
 
@@ -17,7 +18,7 @@ done
 cd mon
 npm install
 cd ..
-node mon/mon-main.js --etcd_url http://$ETCD_URL --etcd_prefix "/vitastor" &>./testdata/mon.log &
+node mon/mon-main.js --etcd_url $ETCD_URL --etcd_prefix "/vitastor" &>./testdata/mon.log &
 MON_PID=$!
 
 if [ -n "$GLOBAL_CONF" ]; then
@@ -25,12 +26,12 @@ if [ -n "$GLOBAL_CONF" ]; then
 fi
 
 if [ "$SCHEME" = "replicated" ]; then
-    $ETCDCTL put /vitastor/config/pools '{"1":{"name":"testpool","scheme":"replicated","pg_size":'$PG_SIZE',"pg_minsize":'$((PG_SIZE-1))',"pg_count":'$PG_COUNT',"failure_domain":"osd"}}'
+    $ETCDCTL put /vitastor/config/pools '{"1":{"name":"testpool","scheme":"replicated","pg_size":'$PG_SIZE',"pg_minsize":'$PG_MINSIZE',"pg_count":'$PG_COUNT',"failure_domain":"osd"}}'
 else
-    $ETCDCTL put /vitastor/config/pools '{"1":{"name":"testpool","scheme":"xor","pg_size":'$PG_SIZE',"pg_minsize":'$((PG_SIZE-1))',"parity_chunks":1,"pg_count":'$PG_COUNT',"failure_domain":"osd"}}'
+    $ETCDCTL put /vitastor/config/pools '{"1":{"name":"testpool","scheme":"xor","pg_size":'$PG_SIZE',"pg_minsize":'$PG_MINSIZE',"parity_chunks":1,"pg_count":'$PG_COUNT',"failure_domain":"osd"}}'
 fi
 
-sleep 2
+sleep 3
 
 if ! ($ETCDCTL get /vitastor/config/pgs --print-value-only | jq -s -e '(. | length) != 0 and ([ .[0].items["1"][] | select(((.osd_set | select(. != 0) | sort | unique) | length) == '$PG_SIZE') ] | length) == '$PG_COUNT); then
     format_error "FAILED: $PG_COUNT PG(s) NOT CONFIGURED"

tests/test_add_osd.sh

@@ -0,0 +1,36 @@
+#!/bin/bash -ex
+
+PG_COUNT=16
+
+. `dirname $0`/run_3osds.sh
+
+LD_PRELOAD=libasan.so.5 \
+    fio -thread -name=test -ioengine=build/src/libfio_vitastor.so -bs=4M -direct=1 -iodepth=1 -end_fsync=1 \
+        -rw=write -etcd=$ETCD_URL -pool=1 -inode=1 -size=128M -cluster_log_level=10
+
+for i in 4; do
+    dd if=/dev/zero of=./testdata/test_osd$i.bin bs=1024 count=1 seek=$((OSD_SIZE*1024-1))
+    build/src/vitastor-osd --osd_num $i --bind_address 127.0.0.1 $OSD_ARGS --etcd_address $ETCD_URL $(build/src/vitastor-cli simple-offsets --format options ./testdata/test_osd$i.bin 2>/dev/null) &>./testdata/osd$i.log &
+    eval OSD${i}_PID=$!
+done
+
+sleep 2
+
+for i in {1..10}; do
+    ($ETCDCTL get /vitastor/config/pgs --print-value-only |\
+        jq -s -e '([ .[0].items["1"] | map(.osd_set)[][] ] | sort | unique == ["1","2","3","4"])') && \
+        ($ETCDCTL get --prefix /vitastor/pg/state/ --print-value-only | jq -s -e '([ .[] | select(.state == ["active"]) ] | length) == '$PG_COUNT'') && \
+        break
+    sleep 1
+done
+
+if ! ($ETCDCTL get /vitastor/config/pgs --print-value-only |\
+    jq -s -e '([ .[0].items["1"] | map(.osd_set)[][] ] | sort | unique == ["1","2","3","4"])'); then
+    format_error "FAILED: OSD NOT ADDED INTO DISTRIBUTION"
+fi
+
+if ! ($ETCDCTL get --prefix /vitastor/pg/state/ --print-value-only | jq -s -e '([ .[] | select(.state == ["active"]) ] | length) == '$PG_COUNT''); then
+    format_error "FAILED: $PG_COUNT PGS NOT ACTIVE"
+fi
+
+format_green OK

tests/test_change_pg_count.sh

@@ -22,7 +22,7 @@ done
 cd mon
 npm install
 cd ..
-node mon/mon-main.js --etcd_url http://$ETCD_URL --etcd_prefix "/vitastor" --verbose 1 &>./testdata/mon.log &
+node mon/mon-main.js --etcd_url $ETCD_URL --etcd_prefix "/vitastor" --verbose 1 &>./testdata/mon.log &
 MON_PID=$!
 
 $ETCDCTL put /vitastor/config/pools '{"1":{"name":"testpool",'$POOLCFG',"pg_count":16,"failure_domain":"osd"}}'

tests/test_change_pg_size.sh

@@ -14,7 +14,7 @@ done
 cd mon
 npm install
 cd ..
-node mon/mon-main.js --etcd_url http://$ETCD_URL --etcd_prefix "/vitastor" &>./testdata/mon.log &
+node mon/mon-main.js --etcd_url $ETCD_URL --etcd_prefix "/vitastor" &>./testdata/mon.log &
 MON_PID=$!
 
 $ETCDCTL put /vitastor/config/pools '{"1":{"name":"testpool","scheme":"replicated","pg_size":3,"pg_minsize":2,"pg_count":16,"failure_domain":"osd"}}'

tests/test_etcd_fail.sh

@@ -0,0 +1,33 @@
+#!/bin/bash -ex
+
+# Run 5 etcds and kill them while operating
+ETCD_COUNT=5
+
+. `dirname $0`/run_3osds.sh
+
+LD_PRELOAD=libasan.so.5 \
+    fio -thread -name=test -ioengine=build/src/libfio_vitastor.so -bs=4M -direct=1 -iodepth=1 -fsync=1 -rw=randwrite \
+        -etcd=$ETCD_URL -pool=1 -inode=1 -size=128M -cluster_log_level=10
+
+kill_etcds()
+{
+    sleep 5
+    kill $ETCD1_PID $ETCD2_PID
+    sleep 5
+    kill $ETCD3_PID
+    start_etcd 1
+    sleep 5
+    kill $ETCD4_PID
+    start_etcd 2
+    sleep 5
+    kill $ETCD5_PID
+    start_etcd 3
+}
+
+kill_etcds &
+
+LD_PRELOAD=libasan.so.5 \
+    fio -thread -name=test -ioengine=build/src/libfio_vitastor.so -bs=4k -direct=1 -iodepth=1 -fsync=1 -rw=randwrite \
+        -etcd=$ETCD_URL -pool=1 -inode=1 -size=128M -cluster_log_level=10 -runtime=30
+
+format_green OK

tests/test_failure_domain.sh

@@ -18,7 +18,7 @@ $ETCDCTL put /vitastor/config/pools '{"1":{"name":"testpool","scheme":"replicate
 cd mon
 npm install
 cd ..
-node mon/mon-main.js --etcd_url http://$ETCD_URL --etcd_prefix "/vitastor" &>./testdata/mon.log &
+node mon/mon-main.js --etcd_url $ETCD_URL --etcd_prefix "/vitastor" &>./testdata/mon.log &
 MON_PID=$!
 
 sleep 2

tests/test_interrupted_rebalance.sh

@@ -22,7 +22,7 @@ done
 cd mon
 npm install
 cd ..
-node mon/mon-main.js --etcd_url http://$ETCD_URL --etcd_prefix "/vitastor" --verbose 1 &>./testdata/mon.log &
+node mon/mon-main.js --etcd_url $ETCD_URL --etcd_prefix "/vitastor" --verbose 1 &>./testdata/mon.log &
 MON_PID=$!
 
 $ETCDCTL put /vitastor/config/pools '{"1":{"name":"testpool","scheme":"replicated","pg_size":2,"pg_minsize":1,"pg_count":32,"failure_domain":"osd"}}'

tests/test_minsize_1.sh

@@ -0,0 +1,17 @@
+#!/bin/bash -ex
+
+PG_MINSIZE=1
+SCHEME=replicated
+
+. `dirname $0`/run_3osds.sh
+
+kill -INT $OSD1_PID
+kill -INT $OSD2_PID
+
+sleep 5
+
+if ! ($ETCDCTL get /vitastor/pg/state/1/ --prefix --print-value-only | jq -s -e '[ .[] | select(.state == ["active", "degraded"]) ] | length == '$PG_COUNT); then
+    format_error "FAILED: $PG_COUNT PG(s) NOT ACTIVE+DEGRADED"
+fi
+
+format_green OK

tests/test_vm_cont.sh

@@ -14,7 +14,7 @@ for i in $(seq 1 $OSD_COUNT); do
     eval OSD${i}_PID=$!
 done
 
-node mon/mon-main.js --etcd_url http://$ETCD_URL --etcd_prefix "/vitastor" &>./testdata/mon.log &
+node mon/mon-main.js --etcd_url $ETCD_URL --etcd_prefix "/vitastor" &>./testdata/mon.log &
 MON_PID=$!
 
 sleep 3