feat: redistribute vgroup

source/dnode/mnode/impl/src/mndTrans.c

@@ -1078,6 +1078,8 @@ static int32_t mndTransExecuteRedoActionsSerial(SMnode *pMnode, STrans *pTrans)
   if (numOfActions == 0) return code;
   if (pTrans->redoActionPos >= numOfActions) return code;
 
+  int32_t retryTimes = 0;
+
   for (int32_t action = pTrans->redoActionPos; action < numOfActions; ++action) {
     STransAction *pAction = taosArrayGet(pTrans->redoActions, pTrans->redoActionPos);
 
@@ -1126,17 +1128,26 @@ static int32_t mndTransExecuteRedoActionsSerial(SMnode *pMnode, STrans *pTrans)
         pTrans->code = terrno;
         mError("trans:%d, %s:%d is executed and failed to sync to other mnodes since %s", pTrans->id,
                mndTransStr(pAction->stage), pAction->id, terrstr());
-        break;
+        if (retryTimes >= 3)
+          break;
+        else
+          continue;
       }
+      retryTimes = 0;
     } else if (code == TSDB_CODE_ACTION_IN_PROGRESS) {
       mDebug("trans:%d, %s:%d is in progress and wait it finish", pTrans->id, mndTransStr(pAction->stage), pAction->id);
       break;
     } else {
       terrno = code;
       pTrans->code = code;
-      mError("trans:%d, %s:%d failed to execute since %s", pTrans->id, mndTransStr(pAction->stage), pAction->id,
-             terrstr());
-      break;
+      pAction->epSet.inUse++;
+      if (pAction->epSet.inUse >= pAction->epSet.numOfEps) pAction->epSet.inUse = 0;
+      mError("trans:%d, %s:%d failed to execute since %s, inUse set to %d", pTrans->id, mndTransStr(pAction->stage),
+             pAction->id, terrstr(), pAction->epSet.inUse);
+      if (retryTimes >= 3)
+        break;
+      else
+        continue;
     }
   }
 

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

@@ -72,7 +72,19 @@ static int32_t vnodeProcessAlterReplicaReq(SVnode *pVnode, SRpcMsg *pMsg) {
     return -1;
   }
 
-  return syncPropose(pVnode->sync, &rpcMsg, false);
+  int32_t code = syncPropose(pVnode->sync, &rpcMsg, false);
+  if (code != 0) {
+    vDebug("vgId:%d, failed to propose reconfig msg since %s", TD_VID(pVnode), terrstr());
+    if (syncLeaderTransfer(pVnode->sync) != 0) {
+      vError("vgId:%d, failed to transfer leader since %s", TD_VID(pVnode), terrstr());
+    } else {
+      vDebug("vgId:%d, transfer leader success, propose reconfig config again", TD_VID(pVnode));
+      taosMsleep(10);
+      code = syncPropose(pVnode->sync, &rpcMsg, false);
+    }
+  }
+
+  return code;
 }
 
 void vnodeProposeMsg(SQueueInfo *pInfo, STaosQall *qall, int32_t numOfMsgs) {

tests/script/tsim/dnode/redistribute_vgroup_replica3_v1_follower.sim

@@ -4,6 +4,11 @@ system sh/deploy.sh -n dnode2 -i 2
 system sh/deploy.sh -n dnode3 -i 3
 system sh/deploy.sh -n dnode4 -i 4
 system sh/deploy.sh -n dnode5 -i 5
+system sh/cfg.sh -n dnode1 -c transPullupInterval -v 1
+system sh/cfg.sh -n dnode2 -c transPullupInterval -v 1
+system sh/cfg.sh -n dnode3 -c transPullupInterval -v 1
+system sh/cfg.sh -n dnode4 -c transPullupInterval -v 1
+system sh/cfg.sh -n dnode5 -c transPullupInterval -v 1
 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
@@ -128,9 +133,9 @@ if $rows != 1 then
 endi
 if $data(2)[4] == leader then
   $leaderExist = 1
-  $leaderVnode = 4
-  $follower1 = 2
-  $follower2 = 3
+  $leaderVnode = 2
+  $follower1 = 3
+  $follower2 = 4
 endi
 if $data(2)[6] == leader then
   $leaderExist = 1
@@ -140,9 +145,9 @@ if $data(2)[6] == leader then
 endi
 if $data(2)[8] == leader then
   $leaderExist = 1
-  $leaderVnode = 2
-  $follower1 = 3
-  $follower2 = 4
+  $leaderVnode = 4
+  $follower1 = 2
+  $follower2 = 3
 endi
 if  $leaderExist != 1 then
   goto step3
@@ -171,8 +176,6 @@ if $rows != 1 then
   return -1
 endi
 
-return
-
 print =============== step33: move follower1
 print redistribute vgroup 2 dnode $leaderVnode dnode $follower1 dnode 5 
 sql redistribute vgroup 2 dnode $leaderVnode dnode $follower1 dnode 5