Merge pull request #13715 from taosdata/fix/mnode

test: execute trans in follower

source/dnode/mnode/impl/inc/mndTrans.h

@@ -75,6 +75,7 @@ int32_t mndTransPrepare(SMnode *pMnode, STrans *pTrans);
 int32_t mndTransProcessRsp(SRpcMsg *pRsp);
 void    mndTransPullup(SMnode *pMnode);
 int32_t mndKillTrans(SMnode *pMnode, STrans *pTrans);
+void    mndTransExecute(SMnode *pMnode, STrans *pTrans);
 
 #ifdef __cplusplus
 }

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

@@ -61,6 +61,12 @@ void mndSyncCommitMsg(struct SSyncFSM *pFsm, const SRpcMsg *pMsg, SFsmCbMeta cbM
     }
     tsem_post(&pMgmt->syncSem);
   } else {
+    STrans *pTrans = mndAcquireTrans(pMnode, transId);
+    if (pTrans != NULL) {
+      mndTransExecute(pMnode, pTrans);
+      mndReleaseTrans(pMnode, pTrans);
+    }
+
     if (cbMeta.index - sdbGetApplyIndex(pMnode->pSdb) > 100) {
       SSnapshotMeta sMeta = {0};
       if (syncGetSnapshotMeta(pMnode->syncMgmt.sync, &sMeta) == 0) {

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

@@ -52,8 +52,8 @@ static bool    mndTransPerformCommitActionStage(SMnode *pMnode, STrans *pTrans);
 static bool    mndTransPerformCommitStage(SMnode *pMnode, STrans *pTrans);
 static bool    mndTransPerformRollbackStage(SMnode *pMnode, STrans *pTrans);
 static bool    mndTransPerfromFinishedStage(SMnode *pMnode, STrans *pTrans);
+static bool    mndCantExecuteTransAction(SMnode *pMnode) { return !pMnode->deploy && !mndIsMaster(pMnode); }
 
-static void    mndTransExecute(SMnode *pMnode, STrans *pTrans);
 static void    mndTransSendRpcRsp(SMnode *pMnode, STrans *pTrans);
 static int32_t mndProcessTransReq(SRpcMsg *pReq);
 static int32_t mndProcessKillTransReq(SRpcMsg *pReq);
@@ -517,12 +517,12 @@ static int32_t mndTransActionUpdate(SSdb *pSdb, STrans *pOld, STrans *pNew) {
 
   if (pOld->stage == TRN_STAGE_COMMIT) {
     pOld->stage = TRN_STAGE_COMMIT_ACTION;
-    mTrace("trans:%d, stage from commit to commitAction", pNew->id);
+    mTrace("trans:%d, stage from commit to commitAction since perform update action", pNew->id);
   }
 
   if (pOld->stage == TRN_STAGE_ROLLBACK) {
     pOld->stage = TRN_STAGE_FINISHED;
-    mTrace("trans:%d, stage from rollback to finished", pNew->id);
+    mTrace("trans:%d, stage from rollback to finished since perform update action", pNew->id);
   }
   return 0;
 }
@@ -914,7 +914,7 @@ static int32_t mndTransWriteSingleLog(SMnode *pMnode, STrans *pTrans, STransActi
 
 static int32_t mndTransSendSingleMsg(SMnode *pMnode, STrans *pTrans, STransAction *pAction) {
   if (pAction->msgSent) return 0;
-  if (!pMnode->deploy && !mndIsMaster(pMnode)) return -1;
+  if (mndCantExecuteTransAction(pMnode)) return -1;
 
   int64_t signature = pTrans->id;
   signature = (signature << 32);
@@ -1114,9 +1114,9 @@ static int32_t mndTransExecuteRedoActionsSerial(SMnode *pMnode, STrans *pTrans)
       pTrans->lastEpset = pAction->epSet;
     }
 
-    if (code == 0) {
-      if (!pMnode->deploy && !mndIsMaster(pMnode)) break;
+    if (mndCantExecuteTransAction(pMnode)) break;
 
+    if (code == 0) {
       pTrans->code = 0;
       pTrans->redoActionPos++;
       mDebug("trans:%d, %s:%d is executed and need sync to other mnodes", pTrans->id, mndTransStr(pAction->stage),
@@ -1160,6 +1160,8 @@ static bool mndTransPerformRedoActionStage(SMnode *pMnode, STrans *pTrans) {
     code = mndTransExecuteRedoActions(pMnode, pTrans);
   }
 
+  if (mndCantExecuteTransAction(pMnode)) return false;
+
   if (code == 0) {
     pTrans->code = 0;
     pTrans->stage = TRN_STAGE_COMMIT;
@@ -1185,6 +1187,8 @@ static bool mndTransPerformRedoActionStage(SMnode *pMnode, STrans *pTrans) {
 }
 
 static bool mndTransPerformCommitStage(SMnode *pMnode, STrans *pTrans) {
+  if (mndCantExecuteTransAction(pMnode)) return false;
+
   bool    continueExec = true;
   int32_t code = mndTransCommit(pMnode, pTrans);
 
@@ -1233,6 +1237,8 @@ static bool mndTransPerformUndoActionStage(SMnode *pMnode, STrans *pTrans) {
   bool    continueExec = true;
   int32_t code = mndTransExecuteUndoActions(pMnode, pTrans);
 
+  if (mndCantExecuteTransAction(pMnode)) return false;
+
   if (code == 0) {
     pTrans->stage = TRN_STAGE_ROLLBACK;
     mDebug("trans:%d, stage from undoAction to rollback", pTrans->id);
@@ -1250,6 +1256,8 @@ static bool mndTransPerformUndoActionStage(SMnode *pMnode, STrans *pTrans) {
 }
 
 static bool mndTransPerformRollbackStage(SMnode *pMnode, STrans *pTrans) {
+  if (mndCantExecuteTransAction(pMnode)) return false;
+
   bool    continueExec = true;
   int32_t code = mndTransRollback(pMnode, pTrans);
 
@@ -1284,10 +1292,11 @@ static bool mndTransPerfromFinishedStage(SMnode *pMnode, STrans *pTrans) {
   return continueExec;
 }
 
-static void mndTransExecute(SMnode *pMnode, STrans *pTrans) {
+void mndTransExecute(SMnode *pMnode, STrans *pTrans) {
   bool continueExec = true;
 
   while (continueExec) {
+    mDebug("trans:%d, continue to execute, stage:%s", pTrans->id, mndTransStr(pTrans->stage));
     pTrans->lastExecTime = taosGetTimestampMs();
     switch (pTrans->stage) {
       case TRN_STAGE_PREPARE:

tests/script/jenkins/basic.txt

@@ -57,7 +57,7 @@
 
 # ---- mnode
 ./test.sh -f tsim/mnode/basic1.sim
-#./test.sh -f tsim/mnode/basic2.sim
+./test.sh -f tsim/mnode/basic2.sim
 ./test.sh -f tsim/mnode/basic3.sim
 ./test.sh -f tsim/mnode/basic4.sim
 

tests/script/tsim/mnode/basic2.sim

@@ -92,6 +92,8 @@ sql show mnodes
 if $rows != 2 then 
   return -1
 endi
+print ===> $data00 $data01 $data02 $data03 $data04 $data05
+print ===> $data10 $data11 $data12 $data13 $data14 $data15
 
 sql show users
 if $rows != 2 then 
@@ -111,6 +113,8 @@ step3:
 		return -1
 	endi
 sql show dnodes -x step3
+print ===> $data00 $data01 $data02 $data03 $data04 $data05
+print ===> $data10 $data11 $data12 $data13 $data14 $data15
 if $data(1)[4] != ready then
   goto step3
 endi

tests/script/tsim/mnode/basic3.sim

@@ -39,11 +39,9 @@ endi
 print =============== step2: create mnode 2
 sql create mnode on dnode 2
 sql create mnode on dnode 3
-return
-system sh/exec.sh -n dnode1 -s stop -x SIGKILL
 sql_error create mnode on dnode 4
 
-
+$leaderExist = 0
 $x = 0
 step2: 
 	$x = $x + 1
@@ -52,13 +50,20 @@ step2:
 		return -1
 	endi
 sql show mnodes -x step2
-if $data(1)[2] != leader then
-  goto step2
+
+print ===> $data00 $data01 $data02 $data03 $data04 $data05
+print ===> $data10 $data11 $data12 $data13 $data14 $data15
+print ===> $data20 $data21 $data22 $data23 $data24 $data25
+if $data(1)[2] == leader then
+  $leaderExist = 1
 endi
-if $data(2)[2] != follower then
-  goto step2
+if $data(2)[2] == leader then
+  $leaderExist = 1
+endi
+if $data(3)[2] == leader then
+  $leaderExist = 1
 endi
-if $data(3)[2] != follower then
+if  $leaderExist != 1 then
   goto step2
 endi
 
@@ -70,10 +75,10 @@ if $rows != 2 then
 endi
 
 # wait mnode2 mnode3 recv data finish
-sleep 10000
+sleep 1000
 
 print =============== step4: stop dnode1
-system sh/exec.sh -n dnode1 -s stop
+system sh/exec.sh -n dnode1 -s stop -x SIGKILL
 
 $x = 0
 step4: 
@@ -92,13 +97,22 @@ if $rows != 2 then
   return -1
 endi
 
-sleep 1000
-sql show dnodes
+$x = 0
+step41: 
+	$x = $x + 1
+	sleep 1000
+	if $x == 10 then
+		return -1
+	endi
+sql show dnodes -x step41
+print ===> $data00 $data01 $data02 $data03 $data04 $data05
+print ===> $data10 $data11 $data12 $data13 $data14 $data15
+print ===> $data20 $data21 $data22 $data23 $data24 $data25
 if $data(2)[4] != ready then
-  return -1
+  goto step41
 endi
 if $data(3)[4] != ready then
-  return -1
+  goto step41
 endi
 
 print =============== step5: stop dnode1
@@ -117,15 +131,29 @@ print $data(1)[0] $data(1)[1] $data(1)[2]
 print $data(2)[0] $data(2)[1] $data(2)[2] 
 print $data(3)[0] $data(3)[1] $data(3)[2] 
 
-if $data(2)[2] != offline then
-  goto step5
-endi
-
 sql show users
 if $rows != 2 then 
   return -1
 endi
 
+$x = 0
+step51: 
+	$x = $x + 1
+	sleep 1000
+	if $x == 10 then
+		return -1
+	endi
+sql show dnodes -x step51
+print ===> $data00 $data01 $data02 $data03 $data04 $data05
+print ===> $data10 $data11 $data12 $data13 $data14 $data15
+print ===> $data20 $data21 $data22 $data23 $data24 $data25
+if $data(1)[4] != ready then
+  goto step51
+endi
+if $data(3)[4] != ready then
+  goto step51
+endi
+
 print =============== step6: stop dnode1
 system sh/exec.sh -n dnode2 -s start
 system sh/exec.sh -n dnode3 -s stop
@@ -147,6 +175,24 @@ if $rows != 2 then
   return -1
 endi
 
+$x = 0
+step61: 
+	$x = $x + 1
+	sleep 1000
+	if $x == 10 then
+		return -1
+	endi
+sql show dnodes -x step61
+print ===> $data00 $data01 $data02 $data03 $data04 $data05
+print ===> $data10 $data11 $data12 $data13 $data14 $data15
+print ===> $data20 $data21 $data22 $data23 $data24 $data25
+if $data(1)[4] != ready then
+  goto step61
+endi
+if $data(2)[4] != ready then
+  goto step61
+endi
+
 system sh/exec.sh -n dnode1 -s stop
 system sh/exec.sh -n dnode2 -s stop
 system sh/exec.sh -n dnode3 -s stop