Merge pull request #14730 from taosdata/feature/3.0_mhli

fix(sync): vnode snapshot sender

source/dnode/vnode/src/vnd/vnodeSync.c

@@ -450,42 +450,81 @@ static void vnodeSyncRollBackMsg(SSyncFSM *pFsm, const SRpcMsg *pMsg, SFsmCbMeta
          syncUtilState2String(cbMeta.state), pMsg->msgType, TMSG_INFO(pMsg->msgType));
 }
 
+#define USE_TSDB_SNAPSHOT
+
 static int32_t vnodeSnapshotStartRead(struct SSyncFSM *pFsm, void *pParam, void **ppReader) {
+#ifdef USE_TSDB_SNAPSHOT
   SVnode         *pVnode = pFsm->data;
   SSnapshotParam *pSnapshotParam = pParam;
   int32_t code = vnodeSnapReaderOpen(pVnode, pSnapshotParam->start, pSnapshotParam->end, (SVSnapReader **)ppReader);
   return code;
+#else
+  *ppReader = taosMemoryMalloc(32);
+  return 0;
+#endif
 }
 
 static int32_t vnodeSnapshotStopRead(struct SSyncFSM *pFsm, void *pReader) {
+#ifdef USE_TSDB_SNAPSHOT
   SVnode *pVnode = pFsm->data;
   int32_t code = vnodeSnapReaderClose(pReader);
   return code;
+#else
+  taosMemoryFree(pReader);
+  return 0;
+#endif
 }
 
 static int32_t vnodeSnapshotDoRead(struct SSyncFSM *pFsm, void *pReader, void **ppBuf, int32_t *len) {
+#ifdef USE_TSDB_SNAPSHOT
   SVnode *pVnode = pFsm->data;
   int32_t code = vnodeSnapRead(pReader, (uint8_t **)ppBuf, len);
   return code;
+#else
+  static int32_t times = 0;
+  if (times++ < 5) {
+    *len = 64;
+    *ppBuf = taosMemoryMalloc(*len);
+    snprintf(*ppBuf, *len, "snapshot block %d", times);
+  } else {
+    *len = 0;
+    *ppBuf = NULL;
+  }
+  return 0;
+#endif
 }
 
 static int32_t vnodeSnapshotStartWrite(struct SSyncFSM *pFsm, void *pParam, void **ppWriter) {
+#ifdef USE_TSDB_SNAPSHOT
   SVnode         *pVnode = pFsm->data;
   SSnapshotParam *pSnapshotParam = pParam;
   int32_t code = vnodeSnapWriterOpen(pVnode, pSnapshotParam->start, pSnapshotParam->end, (SVSnapWriter **)ppWriter);
   return code;
+#else
+  *ppWriter = taosMemoryMalloc(32);
+  return 0;
+#endif
 }
 
 static int32_t vnodeSnapshotStopWrite(struct SSyncFSM *pFsm, void *pWriter, bool isApply) {
+#ifdef USE_TSDB_SNAPSHOT
   SVnode *pVnode = pFsm->data;
   int32_t code = vnodeSnapWriterClose(pWriter, !isApply);
   return code;
+#else
+  taosMemoryFree(pWriter);
+  return 0;
+#endif
 }
 
 static int32_t vnodeSnapshotDoWrite(struct SSyncFSM *pFsm, void *pWriter, void *pBuf, int32_t len) {
+#ifdef USE_TSDB_SNAPSHOT
   SVnode *pVnode = pFsm->data;
   int32_t code = vnodeSnapWrite(pWriter, pBuf, len);
   return code;
+#else
+  return 0;
+#endif
 }
 
 static SSyncFSM *vnodeSyncMakeFsm(SVnode *pVnode) {
@@ -509,6 +548,7 @@ static SSyncFSM *vnodeSyncMakeFsm(SVnode *pVnode) {
 
 int32_t vnodeSyncOpen(SVnode *pVnode, char *path) {
   SSyncInfo syncInfo = {
+      //.snapshotStrategy = SYNC_STRATEGY_WAL_FIRST,
       .snapshotStrategy = SYNC_STRATEGY_NO_SNAPSHOT,
       .batchSize = 10,
       .vgId = pVnode->config.vgId,

source/libs/sync/src/syncAppendEntriesReply.c

@@ -235,10 +235,10 @@ int32_t syncNodeOnAppendEntriesReplySnapshot2Cb(SSyncNode* ths, SyncAppendEntrie
         // do nothing
 
       } else {
-        SSyncRaftEntry* pEntry;
-        int32_t         code = ths->pLogStore->syncLogGetEntry(ths->pLogStore, nextIndex, &pEntry);
-        ASSERT(code == 0);
-        syncNodeStartSnapshotOnce(ths, SYNC_INDEX_BEGIN, nextIndex, pEntry->term, pMsg);
+        SSnapshot oldSnapshot;
+        ths->pFsm->FpGetSnapshotInfo(ths->pFsm, &oldSnapshot);
+        SyncTerm newSnapshotTerm = oldSnapshot.lastApplyTerm;
+        syncNodeStartSnapshotOnce(ths, SYNC_INDEX_BEGIN, nextIndex, newSnapshotTerm, pMsg);
 
         // get sender
         SSyncSnapshotSender* pSender = syncNodeGetSnapshotSender(ths, &(pMsg->srcId));

tests/script/tsim/sync/vnodesnapshot-restart.sim

 system sh/stop_dnodes.sh
 
-system sh/cfg.sh -n dnode1 -c supportVnodes -v 0
+#system sh/cfg.sh -n dnode1 -c supportVnodes -v 0
 
 system sh/exec.sh -n dnode1 -s start
 system sh/exec.sh -n dnode2 -s start