From 948d01a3ace4b06880a2c5bf711607763e7fdecf Mon Sep 17 00:00:00 2001
From: Shengliang Guan <slguan@taosdata.com>
Date: Thu, 28 Jul 2022 11:17:53 +0800
Subject: [PATCH 1/3] enh: let alter all dnodes work

---
 source/dnode/mnode/impl/src/mndDnode.c | 44 ++++++++++++++++----------
 1 file changed, 28 insertions(+), 16 deletions(-)

diff --git a/source/dnode/mnode/impl/src/mndDnode.c b/source/dnode/mnode/impl/src/mndDnode.c
index d84455ac94..3f90f087fd 100644
--- a/source/dnode/mnode/impl/src/mndDnode.c
+++ b/source/dnode/mnode/impl/src/mndDnode.c
@@ -805,14 +805,6 @@ static int32_t mndProcessConfigDnodeReq(SRpcMsg *pReq) {
     return -1;
   }
 
-  SDnodeObj *pDnode = mndAcquireDnode(pMnode, cfgReq.dnodeId);
-  if (pDnode == NULL) {
-    mError("dnode:%d, failed to config since %s ", cfgReq.dnodeId, terrstr());
-    return -1;
-  }
-  SEpSet epSet = mndGetDnodeEpset(pDnode);
-  mndReleaseDnode(pMnode, pDnode);
-
   SDCfgDnodeReq dcfgReq = {0};
   if (strcasecmp(cfgReq.config, "resetlog") == 0) {
     strcpy(dcfgReq.config, "resetlog");
@@ -860,16 +852,36 @@ static int32_t mndProcessConfigDnodeReq(SRpcMsg *pReq) {
     }
   }
 
-  int32_t bufLen = tSerializeSDCfgDnodeReq(NULL, 0, &dcfgReq);
-  void   *pBuf = rpcMallocCont(bufLen);
+  int32_t code = -1;
+  SSdb *pSdb = pMnode->pSdb;
+  void *pIter = NULL;
+  while (1) {
+    SDnodeObj *pDnode = NULL;
+    pIter = sdbFetch(pSdb, SDB_DNODE, pIter, (void **)&pDnode);
+    if (pIter == NULL) break;
 
-  if (pBuf == NULL) return -1;
-  tSerializeSDCfgDnodeReq(pBuf, bufLen, &dcfgReq);
+    if (pDnode->id == cfgReq.dnodeId || cfgReq.dnodeId == -1 || cfgReq.dnodeId == 0) {
+      SEpSet  epSet = mndGetDnodeEpset(pDnode);
+      int32_t bufLen = tSerializeSDCfgDnodeReq(NULL, 0, &dcfgReq);
+      void   *pBuf = rpcMallocCont(bufLen);
 
-  mInfo("dnode:%d, send config req to dnode, app:%p config:%s value:%s", cfgReq.dnodeId, pReq->info.ahandle,
-        dcfgReq.config, dcfgReq.value);
-  SRpcMsg rpcMsg = {.msgType = TDMT_DND_CONFIG_DNODE, .pCont = pBuf, .contLen = bufLen};
-  return tmsgSendReq(&epSet, &rpcMsg);
+      if (pBuf != NULL) {
+        tSerializeSDCfgDnodeReq(pBuf, bufLen, &dcfgReq);
+        mInfo("dnode:%d, send config req to dnode, app:%p config:%s value:%s", cfgReq.dnodeId, pReq->info.ahandle,
+              dcfgReq.config, dcfgReq.value);
+        SRpcMsg rpcMsg = {.msgType = TDMT_DND_CONFIG_DNODE, .pCont = pBuf, .contLen = bufLen};
+        tmsgSendReq(&epSet, &rpcMsg);
+        code = 0;
+      }
+    }
+
+    sdbRelease(pSdb, pDnode);
+  }
+  
+  if (code == -1) {
+    terrno = TSDB_CODE_MND_DNODE_NOT_EXIST;
+  }
+  return code;
 }
 
 static int32_t mndProcessConfigDnodeRsp(SRpcMsg *pRsp) {

From 9d72eda2c5183552dc7f9b8e95c000cd89da7bba Mon Sep 17 00:00:00 2001
From: Shengliang Guan <slguan@taosdata.com>
Date: Thu, 28 Jul 2022 11:18:05 +0800
Subject: [PATCH 2/3] fix: deadlock while rollback

---
 source/dnode/mnode/impl/src/mndTrans.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/source/dnode/mnode/impl/src/mndTrans.c b/source/dnode/mnode/impl/src/mndTrans.c
index 2d7117c11f..b2a0e6aac8 100644
--- a/source/dnode/mnode/impl/src/mndTrans.c
+++ b/source/dnode/mnode/impl/src/mndTrans.c
@@ -1287,6 +1287,7 @@ static bool mndTransPerformRedoActionStage(SMnode *pMnode, STrans *pTrans) {
     mDebug("trans:%d, stage keep on redoAction since %s", pTrans->id, tstrerror(code));
     continueExec = false;
   } else {
+    pTrans->failedTimes++;
     pTrans->code = terrno;
     if (pTrans->policy == TRN_POLICY_ROLLBACK) {
       if (pTrans->lastAction != 0) {
@@ -1306,7 +1307,6 @@ static bool mndTransPerformRedoActionStage(SMnode *pMnode, STrans *pTrans) {
       mError("trans:%d, stage from redoAction to rollback since %s", pTrans->id, terrstr());
       continueExec = true;
     } else {
-      pTrans->failedTimes++;
       mError("trans:%d, stage keep on redoAction since %s, failedTimes:%d", pTrans->id, terrstr(), pTrans->failedTimes);
       continueExec = false;
     }

From 5185fa90c52ccfc61a58b0c52203d0dd3ae65c26 Mon Sep 17 00:00:00 2001
From: Shengliang Guan <slguan@taosdata.com>
Date: Thu, 28 Jul 2022 11:48:14 +0800
Subject: [PATCH 3/3] test: add reproduce codes

---
 tests/script/tsim/mnode/basic4.sim | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)

diff --git a/tests/script/tsim/mnode/basic4.sim b/tests/script/tsim/mnode/basic4.sim
index 6b4c8c6b7c..0ffcdd8c00 100644
--- a/tests/script/tsim/mnode/basic4.sim
+++ b/tests/script/tsim/mnode/basic4.sim
@@ -45,11 +45,8 @@ endi
 if $data(2)[4] != ready then
   goto step2
 endi
-if $data(3)[4] != ready then
-  goto step2
-endi
 
-system sh/exec.sh -n dnode3 -s stop -x SIGKILL
+system sh/exec.sh -n dnode3 -s stop
 sql_error create mnode on dnode 3
 
 print =============== step3: show mnodes