Merge branch 'valkey-io:unstable' into clang-format

.github/workflows/ci.yml

@@ -95,16 +95,14 @@ jobs:
     - name: make
       run: make SERVER_CFLAGS='-Werror' MALLOC=libc
 
-  build-centos7-jemalloc:
+  build-almalinux8-jemalloc:
     runs-on: ubuntu-latest
-    container: centos:7
+    container: almalinux:8
     steps:
-    # on centos7, actions/checkout@v4 does not work, so we use v3
-    # ref. https://github.com/actions/checkout/issues/1487
-    - uses: actions/checkout@f43a0e5ff2bd294095638e18286ca9a3d1956744 # v3.6.0
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # v4.1.1
 
     - name: make
       run: |
-        yum -y install gcc make
-        make SERVER_CFLAGS='-Werror'
+        dnf -y install epel-release gcc make procps-ng which
+        make -j SERVER_CFLAGS='-Werror'
 

.github/workflows/daily.yml

@@ -11,7 +11,7 @@ on:
     inputs:
       skipjobs:
         description: 'jobs to skip (delete the ones you wanna keep, do not leave empty)'
-        default: 'valgrind,sanitizer,tls,freebsd,macos,alpine,32bit,iothreads,ubuntu,centos,malloc,specific,fortify,reply-schema'
+        default: 'valgrind,sanitizer,tls,freebsd,macos,alpine,32bit,iothreads,ubuntu,rpm-distros,malloc,specific,fortify,reply-schema'
       skiptests:
         description: 'tests to skip (delete the ones you wanna keep, do not leave empty)'
         default: 'valkey,modules,sentinel,cluster,unittest'
@@ -672,13 +672,34 @@ jobs:
         if: true && !contains(github.event.inputs.skiptests, 'unittest')
         run: ./src/valkey-unit-tests --accurate
 
-  test-centos7-jemalloc:
-    runs-on: ubuntu-latest
+  test-rpm-distros-jemalloc:
     if: |
-      (github.event_name == 'workflow_dispatch' || (github.event_name != 'workflow_dispatch' && github.repository == 'valkey-io/valkey')) &&
-      !contains(github.event.inputs.skipjobs, 'centos')
-    container: centos:7
+       (github.event_name == 'workflow_dispatch' || (github.event_name != 'workflow_dispatch' && github.repository == 'valkey-io/valkey')) &&
+       !contains(github.event.inputs.skipjobs, 'rpm-distros')
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - name: test-almalinux8-jemalloc
+            container: almalinux:8
+            install_epel: true
+          - name: test-almalinux9-jemalloc
+            container: almalinux:8
+            install_epel: true
+          - name: test-centosstream9-jemalloc
+            container: quay.io/centos/centos:stream9
+            install_epel: true
+          - name: test-fedoralatest-jemalloc
+            container: fedora:latest
+          - name: test-fedorarawhide-jemalloc
+            container: fedora:rawhide
+
+    name: ${{ matrix.name }}
+    runs-on: ubuntu-latest
+
+    container: ${{ matrix.container }}
     timeout-minutes: 14400
+
     steps:
     - name: prep
       if: github.event_name == 'workflow_dispatch'
@@ -689,18 +710,19 @@ jobs:
         echo "skiptests: ${{github.event.inputs.skiptests}}"
         echo "test_args: ${{github.event.inputs.test_args}}"
         echo "cluster_test_args: ${{github.event.inputs.cluster_test_args}}"
-    # On centos7 actions/checkout@v4 does not work, so we use v3
-    # ref. https://github.com/actions/checkout/issues/1487
-    - uses: actions/checkout@f43a0e5ff2bd294095638e18286ca9a3d1956744 # v3.6.0
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # v4.1.1
       with:
         repository: ${{ env.GITHUB_REPOSITORY }}
         ref: ${{ env.GITHUB_HEAD_REF }}
+    - name: Install EPEL
+      if: matrix.install_epel
+      run: dnf -y install epel-release
     - name: make
       run: |
-        yum -y install gcc make
-        make SERVER_CFLAGS='-Werror'
+        dnf -y install gcc make procps-ng which /usr/bin/kill
+        make -j SERVER_CFLAGS='-Werror'
     - name: testprep
-      run: yum -y install which tcl tclx
+      run: dnf -y install tcl tcltls
     - name: test
       if: true && !contains(github.event.inputs.skiptests, 'valkey')
       run: ./runtest --accurate --verbose --dump-logs ${{github.event.inputs.test_args}}
@@ -714,13 +736,34 @@ jobs:
       if: true && !contains(github.event.inputs.skiptests, 'cluster')
       run: ./runtest-cluster ${{github.event.inputs.cluster_test_args}}
 
-  test-centos7-tls-module:
-    runs-on: ubuntu-latest
+  test-rpm-distros-tls-module:
     if: |
-      (github.event_name == 'workflow_dispatch' || (github.event_name != 'workflow_dispatch' && github.repository == 'valkey-io/valkey')) &&
-      !contains(github.event.inputs.skipjobs, 'tls')
-    container: centos:7
+       (github.event_name == 'workflow_dispatch' || (github.event_name != 'workflow_dispatch' && github.repository == 'valkey-io/valkey')) &&
+       !contains(github.event.inputs.skipjobs, 'tls')
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - name: test-almalinux8-tls-module
+            container: almalinux:8
+            install_epel: true
+          - name: test-almalinux9-tls-module
+            container: almalinux:8
+            install_epel: true
+          - name: test-centosstream9-tls-module
+            container: quay.io/centos/centos:stream9
+            install_epel: true
+          - name: test-fedoralatest-tls-module
+            container: fedora:latest
+          - name: test-fedorarawhide-tls-module
+            container: fedora:rawhide
+
+    name: ${{ matrix.name }}
+    runs-on: ubuntu-latest
+
+    container: ${{ matrix.container }}
     timeout-minutes: 14400
+
     steps:
     - name: prep
       if: github.event_name == 'workflow_dispatch'
@@ -731,20 +774,20 @@ jobs:
         echo "skiptests: ${{github.event.inputs.skiptests}}"
         echo "test_args: ${{github.event.inputs.test_args}}"
         echo "cluster_test_args: ${{github.event.inputs.cluster_test_args}}"
-    # On centos7 actions/checkout@v4 does not work, so we use v3
-    # ref. https://github.com/actions/checkout/issues/1487
-    - uses: actions/checkout@f43a0e5ff2bd294095638e18286ca9a3d1956744 # v3.6.0
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # v4.1.1
       with:
         repository: ${{ env.GITHUB_REPOSITORY }}
         ref: ${{ env.GITHUB_HEAD_REF }}
+    - name: Install EPEL
+      if: matrix.install_epel
+      run: dnf -y install epel-release
     - name: make
       run: |
-        yum -y install centos-release-scl epel-release
-        yum -y install devtoolset-7 openssl-devel openssl
-        scl enable devtoolset-7 "make BUILD_TLS=module SERVER_CFLAGS='-Werror'"
+        dnf -y install make gcc openssl-devel openssl procps-ng which /usr/bin/kill
+        make -j BUILD_TLS=module SERVER_CFLAGS='-Werror'
     - name: testprep
       run: |
-        yum -y install tcl tcltls tclx
+        dnf -y install tcl tcltls
         ./utils/gen-test-certs.sh
     - name: test
       if: true && !contains(github.event.inputs.skiptests, 'valkey')
@@ -763,13 +806,34 @@ jobs:
       run: |
         ./runtest-cluster --tls-module ${{github.event.inputs.cluster_test_args}}
 
-  test-centos7-tls-module-no-tls:
-    runs-on: ubuntu-latest
+  test-rpm-distros-tls-module-no-tls:
     if: |
-      (github.event_name == 'workflow_dispatch' || (github.event_name != 'workflow_dispatch' && github.repository == 'valkey-io/valkey')) &&
-      !contains(github.event.inputs.skipjobs, 'tls')
-    container: centos:7
+       (github.event_name == 'workflow_dispatch' || (github.event_name != 'workflow_dispatch' && github.repository == 'valkey-io/valkey')) &&
+       !contains(github.event.inputs.skipjobs, 'tls')
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - name: test-almalinux8-tls-module-no-tls
+            container: almalinux:8
+            install_epel: true
+          - name: test-almalinux9-tls-module-no-tls
+            container: almalinux:8
+            install_epel: true
+          - name: test-centosstream9-tls-module-no-tls
+            container: quay.io/centos/centos:stream9
+            install_epel: true
+          - name: test-fedoralatest-tls-module-no-tls
+            container: fedora:latest
+          - name: test-fedorarawhide-tls-module-no-tls
+            container: fedora:rawhide
+
+    name: ${{ matrix.name }}
+    runs-on: ubuntu-latest
+
+    container: ${{ matrix.container }}
     timeout-minutes: 14400
+
     steps:
     - name: prep
       if: github.event_name == 'workflow_dispatch'
@@ -780,20 +844,20 @@ jobs:
         echo "skiptests: ${{github.event.inputs.skiptests}}"
         echo "test_args: ${{github.event.inputs.test_args}}"
         echo "cluster_test_args: ${{github.event.inputs.cluster_test_args}}"
-    # On centos7 actions/checkout@v4 does not work, so we use v3
-    # ref. https://github.com/actions/checkout/issues/1487
-    - uses: actions/checkout@f43a0e5ff2bd294095638e18286ca9a3d1956744 # v3.6.0
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # v4.1.1
       with:
         repository: ${{ env.GITHUB_REPOSITORY }}
         ref: ${{ env.GITHUB_HEAD_REF }}
+    - name: Install EPEL
+      if: matrix.install_epel
+      run: dnf -y install epel-release
     - name: make
       run: |
-        yum -y install centos-release-scl epel-release
-        yum -y install devtoolset-7 openssl-devel openssl
-        scl enable devtoolset-7 "make BUILD_TLS=module SERVER_CFLAGS='-Werror'"
+        dnf -y install make gcc openssl-devel openssl procps-ng which /usr/bin/kill
+        make -j BUILD_TLS=module SERVER_CFLAGS='-Werror'
     - name: testprep
       run: |
-        yum -y install tcl tcltls tclx
+        dnf -y install tcl tcltls
         ./utils/gen-test-certs.sh
     - name: test
       if: true && !contains(github.event.inputs.skiptests, 'valkey')
@@ -1074,7 +1138,7 @@ jobs:
   notify-about-job-results:
     runs-on: ubuntu-latest
     if: always() && github.event_name != 'workflow_dispatch' && github.repository == 'valkey-io/valkey'
-    needs: [test-ubuntu-jemalloc, test-ubuntu-jemalloc-fortify, test-ubuntu-libc-malloc, test-ubuntu-no-malloc-usable-size, test-ubuntu-32bit, test-ubuntu-tls, test-ubuntu-tls-no-tls, test-ubuntu-io-threads, test-ubuntu-reclaim-cache, test-valgrind-test, test-valgrind-misc, test-valgrind-no-malloc-usable-size-test, test-valgrind-no-malloc-usable-size-misc, test-sanitizer-address, test-sanitizer-undefined, test-centos7-jemalloc, test-centos7-tls-module, test-centos7-tls-module-no-tls, test-macos-latest, test-macos-latest-sentinel, test-macos-latest-cluster, build-macos, test-freebsd, test-alpine-jemalloc, test-alpine-libc-malloc, reply-schemas-validator]
+    needs: [test-ubuntu-jemalloc, test-ubuntu-jemalloc-fortify, test-ubuntu-libc-malloc, test-ubuntu-no-malloc-usable-size, test-ubuntu-32bit, test-ubuntu-tls, test-ubuntu-tls-no-tls, test-ubuntu-io-threads, test-ubuntu-reclaim-cache, test-valgrind-test, test-valgrind-misc, test-valgrind-no-malloc-usable-size-test, test-valgrind-no-malloc-usable-size-misc, test-sanitizer-address, test-sanitizer-undefined, test-rpm-distros-jemalloc, test-rpm-distros-tls-module, test-rpm-distros-tls-module-no-tls, test-macos-latest, test-macos-latest-sentinel, test-macos-latest-cluster, build-macos, test-freebsd, test-alpine-jemalloc, test-alpine-libc-malloc, reply-schemas-validator]
     steps:
       - name: Collect job status
         run: |

src/aof.c

@@ -946,7 +946,7 @@ void stopAppendOnly(void) {
     server.aof_last_incr_size = 0;
     server.aof_last_incr_fsync_offset = 0;
     server.fsynced_reploff = -1;
-    atomicSet(server.fsynced_reploff_pending, 0);
+    atomic_store_explicit(&server.fsynced_reploff_pending, 0, memory_order_relaxed);
     killAppendOnlyChild();
     sdsfree(server.aof_buf);
     server.aof_buf = sdsempty();
@@ -985,11 +985,11 @@ int startAppendOnly(void) {
     }
     server.aof_last_fsync = server.mstime;
     /* If AOF fsync error in bio job, we just ignore it and log the event. */
-    int aof_bio_fsync_status;
-    atomicGet(server.aof_bio_fsync_status, aof_bio_fsync_status);
+    int aof_bio_fsync_status = atomic_load_explicit(&server.aof_bio_fsync_status, memory_order_relaxed);
     if (aof_bio_fsync_status == C_ERR) {
-        serverLog(LL_WARNING, "AOF reopen, just ignore the AOF fsync error in bio job");
-        atomicSet(server.aof_bio_fsync_status, C_OK);
+        serverLog(LL_WARNING,
+            "AOF reopen, just ignore the AOF fsync error in bio job");
+        atomic_store_explicit(&server.aof_bio_fsync_status, C_OK, memory_order_relaxed);
     }
 
     /* If AOF was in error state, we just ignore it and log the event. */
@@ -1074,7 +1074,7 @@ void flushAppendOnlyFile(int force) {
              * (because there's no reason, from the AOF POV, to call fsync) and then WAITAOF may wait on
              * the higher offset (which contains data that was only propagated to replicas, and not to AOF) */
             if (!sync_in_progress && server.aof_fsync != AOF_FSYNC_NO)
-                atomicSet(server.fsynced_reploff_pending, server.master_repl_offset);
+                atomic_store_explicit(&server.fsynced_reploff_pending, server.master_repl_offset, memory_order_relaxed);
             return;
         }
     }
@@ -1244,8 +1244,9 @@ void flushAppendOnlyFile(int force) {
         latencyAddSampleIfNeeded("aof-fsync-always", latency);
         server.aof_last_incr_fsync_offset = server.aof_last_incr_size;
         server.aof_last_fsync = server.mstime;
-        atomicSet(server.fsynced_reploff_pending, server.master_repl_offset);
-    } else if (server.aof_fsync == AOF_FSYNC_EVERYSEC && server.mstime - server.aof_last_fsync >= 1000) {
+        atomic_store_explicit(&server.fsynced_reploff_pending, server.master_repl_offset, memory_order_relaxed);
+    } else if (server.aof_fsync == AOF_FSYNC_EVERYSEC &&
+               server.mstime - server.aof_last_fsync >= 1000) {
         if (!sync_in_progress) {
             aof_background_fsync(server.aof_fd);
             server.aof_last_incr_fsync_offset = server.aof_last_incr_size;
@@ -2409,7 +2410,7 @@ int rewriteAppendOnlyFileBackground(void) {
 
         /* Set the initial repl_offset, which will be applied to fsynced_reploff
          * when AOFRW finishes (after possibly being updated by a bio thread) */
-        atomicSet(server.fsynced_reploff_pending, server.master_repl_offset);
+        atomic_store_explicit(&server.fsynced_reploff_pending, server.master_repl_offset, memory_order_relaxed);
         server.fsynced_reploff = 0;
     }
 
@@ -2647,8 +2648,7 @@ void backgroundRewriteDoneHandler(int exitcode, int bysignal) {
             /* Update the fsynced replication offset that just now become valid.
              * This could either be the one we took in startAppendOnly, or a
              * newer one set by the bio thread. */
-            long long fsynced_reploff_pending;
-            atomicGet(server.fsynced_reploff_pending, fsynced_reploff_pending);
+            long long fsynced_reploff_pending = atomic_load_explicit(&server.fsynced_reploff_pending, memory_order_relaxed);
             server.fsynced_reploff = fsynced_reploff_pending;
         }
 

src/bio.c

@@ -62,6 +62,7 @@
 
 #include "server.h"
 #include "bio.h"
+#include <stdatomic.h>
 
 static char *bio_worker_title[] = {
     "bio_close_file",
@@ -256,17 +257,19 @@ void *bioProcessBackgroundJobs(void *arg) {
             /* The fd may be closed by main thread and reused for another
              * socket, pipe, or file. We just ignore these errno because
              * aof fsync did not really fail. */
-            if (valkey_fsync(job->fd_args.fd) == -1 && errno != EBADF && errno != EINVAL) {
-                int last_status;
-                atomicGet(server.aof_bio_fsync_status, last_status);
-                atomicSet(server.aof_bio_fsync_status, C_ERR);
-                atomicSet(server.aof_bio_fsync_errno, errno);
+            if (valkey_fsync(job->fd_args.fd) == -1 &&
+                errno != EBADF && errno != EINVAL)
+            {
+                int last_status = atomic_load_explicit(&server.aof_bio_fsync_status, memory_order_relaxed);
+
+                atomic_store_explicit(&server.aof_bio_fsync_errno, errno, memory_order_relaxed);
+                atomic_store_explicit(&server.aof_bio_fsync_status, C_ERR, memory_order_release);
                 if (last_status == C_OK) {
                     serverLog(LL_WARNING, "Fail to fsync the AOF file: %s", strerror(errno));
                 }
             } else {
-                atomicSet(server.aof_bio_fsync_status, C_OK);
-                atomicSet(server.fsynced_reploff_pending, job->fd_args.offset);
+                atomic_store_explicit(&server.aof_bio_fsync_status, C_OK, memory_order_relaxed);
+                atomic_store_explicit(&server.fsynced_reploff_pending, job->fd_args.offset, memory_order_relaxed);
             }
 
             if (job->fd_args.need_reclaim_cache) {

src/cluster.c

@@ -1048,10 +1048,12 @@ getNodeByQuery(client *c, struct serverCommand *cmd, robj **argv, int argc, int
                  * can safely serve the request, otherwise we return a TRYAGAIN
                  * error). To do so we set the importing/migrating state and
                  * increment a counter for every missing key. */
-                if (n == myself && getMigratingSlotDest(slot) != NULL) {
-                    migrating_slot = 1;
-                } else if (getImportingSlotSource(slot) != NULL) {
-                    importing_slot = 1;
+                if (clusterNodeIsMaster(myself) || c->flags & CLIENT_READONLY) {
+                    if (n == clusterNodeGetMaster(myself) && getMigratingSlotDest(slot) != NULL) {
+                        migrating_slot = 1;
+                    } else if (getImportingSlotSource(slot) != NULL) {
+                        importing_slot = 1;
+                    }
                 }
             } else {
                 /* If it is not the first key/channel, make sure it is exactly
@@ -1120,7 +1122,9 @@ getNodeByQuery(client *c, struct serverCommand *cmd, robj **argv, int argc, int
     /* MIGRATE always works in the context of the local node if the slot
      * is open (migrating or importing state). We need to be able to freely
      * move keys among instances in this case. */
-    if ((migrating_slot || importing_slot) && cmd->proc == migrateCommand) return myself;
+    if ((migrating_slot || importing_slot) && cmd->proc == migrateCommand && clusterNodeIsMaster(myself)) {
+        return myself;
+    }
 
     /* If we don't have all the keys and we are migrating the slot, send
      * an ASK redirection or TRYAGAIN. */

src/commands.def

@@ -975,7 +975,7 @@ struct COMMAND_STRUCT CLUSTER_Subcommands[] = {
 {MAKE_CMD("setslot","Binds a hash slot to a node.","O(1)","3.0.0",CMD_DOC_NONE,NULL,NULL,"cluster",COMMAND_GROUP_CLUSTER,CLUSTER_SETSLOT_History,1,CLUSTER_SETSLOT_Tips,0,clusterCommand,-4,CMD_NO_ASYNC_LOADING|CMD_ADMIN|CMD_STALE|CMD_MAY_REPLICATE,0,CLUSTER_SETSLOT_Keyspecs,0,NULL,3),.args=CLUSTER_SETSLOT_Args},
 {MAKE_CMD("shards","Returns the mapping of cluster slots to shards.","O(N) where N is the total number of cluster nodes","7.0.0",CMD_DOC_NONE,NULL,NULL,"cluster",COMMAND_GROUP_CLUSTER,CLUSTER_SHARDS_History,0,CLUSTER_SHARDS_Tips,1,clusterCommand,2,CMD_LOADING|CMD_STALE,0,CLUSTER_SHARDS_Keyspecs,0,NULL,0)},
 {MAKE_CMD("slaves","Lists the replica nodes of a master node.","O(N) where N is the number of replicas.","3.0.0",CMD_DOC_DEPRECATED,"`CLUSTER REPLICAS`","5.0.0","cluster",COMMAND_GROUP_CLUSTER,CLUSTER_SLAVES_History,0,CLUSTER_SLAVES_Tips,1,clusterCommand,3,CMD_ADMIN|CMD_STALE,0,CLUSTER_SLAVES_Keyspecs,0,NULL,1),.args=CLUSTER_SLAVES_Args},
-{MAKE_CMD("slots","Returns the mapping of cluster slots to nodes.","O(N) where N is the total number of Cluster nodes","3.0.0",CMD_DOC_DEPRECATED,"`CLUSTER SHARDS`","7.0.0","cluster",COMMAND_GROUP_CLUSTER,CLUSTER_SLOTS_History,2,CLUSTER_SLOTS_Tips,1,clusterCommand,2,CMD_LOADING|CMD_STALE,0,CLUSTER_SLOTS_Keyspecs,0,NULL,0)},
+{MAKE_CMD("slots","Returns the mapping of cluster slots to nodes.","O(N) where N is the total number of Cluster nodes","3.0.0",CMD_DOC_NONE,NULL,NULL,"cluster",COMMAND_GROUP_CLUSTER,CLUSTER_SLOTS_History,2,CLUSTER_SLOTS_Tips,1,clusterCommand,2,CMD_LOADING|CMD_STALE,0,CLUSTER_SLOTS_Keyspecs,0,NULL,0)},
 {0}
 };