fix rpc connection fail issue

source/libs/scheduler/inc/schedulerInt.h

@@ -365,7 +365,7 @@ int32_t schExecJob(void *pTrans, SArray *pNodeList, SQueryPlan *pDag, int64_t *p
 int32_t schAsyncExecJob(void *pTrans, SArray *pNodeList, SQueryPlan *pDag, int64_t *pJob, const char *sql, int64_t startTs, SSchResInfo *pRes);
 int32_t schFetchRows(SSchJob *pJob);
 int32_t schAsyncFetchRows(SSchJob *pJob);
-int32_t schUpdateTaskHandle(SSchJob *pJob, SSchTask *pTask, int32_t msgType, void *handle, int32_t execIdx);
+int32_t schUpdateTaskHandle(SSchJob *pJob, SSchTask *pTask, bool dropExecNode, void *handle, int32_t execIdx);
 int32_t schProcessOnTaskStatusRsp(SQueryNodeEpId* pEpId, SArray* pStatusList);
 
 

source/libs/scheduler/src/schJob.c

@@ -396,8 +396,8 @@ int32_t schUpdateTaskExecNode(SSchTask *pTask, void *handle, int32_t execIdx) {
   return TSDB_CODE_SUCCESS;
 }
 
-int32_t schUpdateTaskHandle(SSchJob *pJob, SSchTask *pTask, int32_t msgType, void *handle, int32_t execIdx) {
-  if (msgType == TDMT_SCH_LINK_BROKEN) {
+int32_t schUpdateTaskHandle(SSchJob *pJob, SSchTask *pTask, bool dropExecNode, void *handle, int32_t execIdx) {
+  if (dropExecNode) {
     SCH_RET(schDropTaskExecNode(pJob, pTask, handle, execIdx));
   }
 

source/libs/scheduler/src/schRemote.c

@@ -372,7 +372,8 @@ int32_t schHandleCallback(void *param, const SDataBuf *pMsg, int32_t msgType, in
     goto _return;
   }
 
-  SCH_ERR_JRET(schUpdateTaskHandle(pJob, pTask, msgType, pMsg->handle, pParam->execIdx));
+  bool dropExecNode = (msgType == TDMT_SCH_LINK_BROKEN || rspCode == TSDB_CODE_RPC_NETWORK_UNAVAIL);
+  SCH_ERR_JRET(schUpdateTaskHandle(pJob, pTask, dropExecNode, pMsg->handle, pParam->execIdx));
   
   SCH_ERR_JRET(schHandleResponseMsg(pJob, pTask, msgType, pMsg->pData, pMsg->len, rspCode));