Merge remote-tracking branch 'origin/develop' into hotfix/crash

src/client/src/tscServer.c

@@ -438,8 +438,9 @@ void tscKillSTableQuery(SSqlObj *pSql) {
      * here, we cannot set the command = TSDB_SQL_KILL_QUERY. Otherwise, it may cause
      * sub-queries not correctly released and master sql object of super table query reaches an abnormal state.
      */
-    pSql->pSubs[i]->res.code = TSDB_CODE_TSC_QUERY_CANCELLED;
+    rpcCancelRequest(pSub->pRpcCtx);
-    rpcCancelRequest(pSql->pSubs[i]->pRpcCtx);
+    pSub->res.code = TSDB_CODE_TSC_QUERY_CANCELLED;
+    tscQueueAsyncRes(pSub);
   }
 
   /*

src/client/src/tscSql.c

@@ -617,19 +617,18 @@ void taos_stop_query(TAOS_RES *res) {
   if (pSql->signature != pSql) return;
   tscDebug("%p start to cancel query", res);
 
-  pSql->res.code = TSDB_CODE_TSC_QUERY_CANCELLED;
 
   SQueryInfo *pQueryInfo = tscGetQueryInfoDetail(pCmd, pCmd->clauseIndex);
   if (tscIsTwoStageSTableQuery(pQueryInfo, 0)) {
     tscKillSTableQuery(pSql);
-    return;
   }
 
-  if (pSql->cmd.command >= TSDB_SQL_LOCAL) {
+  if (pSql->cmd.command < TSDB_SQL_LOCAL) {
-    return;
+    rpcCancelRequest(pSql->pRpcCtx);
   }
+  pSql->res.code = TSDB_CODE_TSC_QUERY_CANCELLED;
+  tscQueueAsyncRes(pSql);
 
-  rpcCancelRequest(pSql->pRpcCtx);
   tscDebug("%p query is cancelled", res);
 }
 

src/kit/taosdemo/taosdemo.c

@@ -870,6 +870,11 @@ void *readTable(void *sarg) {
   int64_t sTime = rinfo->start_time;
   char *tb_prefix = rinfo->tb_prefix;
   FILE *fp = fopen(rinfo->fp, "a");
+  if (NULL == fp) {
+    printf("fopen %s fail, reason:%s.\n", rinfo->fp, strerror(errno));
+    return NULL;
+  }
+  
   int num_of_DPT = rinfo->nrecords_per_table;
   int num_of_tables = rinfo->end_table_id - rinfo->start_table_id + 1;
   int totalData = num_of_DPT * num_of_tables;
@@ -925,6 +930,11 @@ void *readMetric(void *sarg) {
   TAOS *taos = rinfo->taos;
   char command[BUFFER_SIZE] = "\0";
   FILE *fp = fopen(rinfo->fp, "a");
+  if (NULL == fp) {
+    printf("fopen %s fail, reason:%s.\n", rinfo->fp, strerror(errno));
+    return NULL;
+  }
+  
   int num_of_DPT = rinfo->nrecords_per_table;
   int num_of_tables = rinfo->end_table_id - rinfo->start_table_id + 1;
   int totalData = num_of_DPT * num_of_tables;

src/kit/taosmigrate/taosmigrate.c

@@ -51,6 +51,7 @@ static error_t parse_opt(int key, char *arg, struct argp_state *state) {
       break;
     case 'f':
       arguments->fqdn = arg;
+      break;
     case 'g':
       arguments->dnodeGroups = arg;
       break;

src/kit/taosmigrate/taosmigrateMnodeWal.c

@@ -96,6 +96,7 @@ void walModWalFile(char* walfile) {
   if (wfd < 0) {
     printf("wal:%s, failed to open(%s)\n", newWalFile, strerror(errno));
     free(buffer);
+    close(rfd);
     return ;
   }
 
@@ -116,6 +117,11 @@ void walModWalFile(char* walfile) {
       break;
     } 
 
+    if (pHead->len >= 1024000 - sizeof(SWalHead)) {
+      printf("wal:%s, SWalHead.len(%d) overflow, skip the rest of file\n", walfile, pHead->len);
+      break;
+    } 
+
     ret = read(rfd, pHead->cont, pHead->len);
     if ( ret != pHead->len) {
       printf("wal:%s, failed to read body, skip, len:%d ret:%d\n", walfile, pHead->len, ret);

src/kit/taosmigrate/taosmigrateVnodeCfg.c

@@ -99,6 +99,8 @@ static int32_t readVnodeCfg(SVnodeObj *pVnode, char* cfgFile)
     goto PARSE_OVER;
   }
 
+  content[maxLen] = (char)0;
+
   root = cJSON_Parse(content);
   if (root == NULL) {
     printf("failed to json parse %s, invalid json format\n", cfgFile);

src/vnode/src/vnodeRead.c

@@ -135,7 +135,10 @@ static int32_t vnodeProcessQueryMsg(SVnodeObj *pVnode, SReadMsg *pReadMsg) {
     } else {
       assert(pQInfo == NULL);
     }
-
+    if (handle != NULL) {
+      dnodePutItemIntoReadQueue(pVnode, handle);
+      qReleaseQInfo(pVnode->qMgmt, (void**) &handle, false);
+    }
     vDebug("vgId:%d, QInfo:%p, dnode query msg disposed", vgId, pQInfo);
   } else {
     assert(pCont != NULL);
@@ -146,14 +149,10 @@ static int32_t vnodeProcessQueryMsg(SVnodeObj *pVnode, SReadMsg *pReadMsg) {
     } else {
       vDebug("vgId:%d, QInfo:%p, dnode query msg in progress", pVnode->vgId, *(void**) pCont);
       code = TSDB_CODE_VND_ACTION_IN_PROGRESS;
+      qTableQuery(*handle); // do execute query
     }
-  }
-
-  if (handle != NULL) {
-    qTableQuery(*handle); // do execute query
     qReleaseQInfo(pVnode->qMgmt, (void**) &handle, false);
   }
-
   return code;
 }
 

tests/script/unique/cluster/client1_0.sim

@@ -0,0 +1,16 @@
+sql connect
+
+$db = db1
+$stb = stb1
+print =============== client1_0:
+
+sql use $db
+
+$tblNum = 1000
+
+$i = 1
+while $i < $tblNum
+  $tb = tb . $i
+  sql create table $tb using $stb tags ($i, 'abcd')
+  $i = $i + 1
+endw 

tests/script/unique/cluster/cluster_main.sim

@@ -0,0 +1,494 @@
+system sh/stop_dnodes.sh
+system sh/deploy.sh -n dnode1 -i 1
+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/cfg.sh -n dnode1 -c numOfMnodes -v 3
+system sh/cfg.sh -n dnode2 -c numOfMnodes -v 3
+system sh/cfg.sh -n dnode3 -c numOfMnodes -v 3
+system sh/cfg.sh -n dnode4 -c numOfMnodes -v 3
+
+system sh/cfg.sh -n dnode1 -c walLevel -v 1
+system sh/cfg.sh -n dnode2 -c walLevel -v 1
+system sh/cfg.sh -n dnode3 -c walLevel -v 1
+system sh/cfg.sh -n dnode4 -c walLevel -v 1
+
+system sh/cfg.sh -n dnode1 -c balanceInterval -v 10
+system sh/cfg.sh -n dnode2 -c balanceInterval -v 10
+system sh/cfg.sh -n dnode3 -c balanceInterval -v 10
+system sh/cfg.sh -n dnode4 -c balanceInterval -v 10
+
+system sh/cfg.sh -n dnode1 -c numOfTotalVnodes -v 4
+system sh/cfg.sh -n dnode2 -c numOfTotalVnodes -v 4
+system sh/cfg.sh -n dnode3 -c numOfTotalVnodes -v 4
+system sh/cfg.sh -n dnode4 -c numOfTotalVnodes -v 4
+
+system sh/cfg.sh -n dnode1 -c alternativeRole -v 0
+system sh/cfg.sh -n dnode2 -c alternativeRole -v 0
+system sh/cfg.sh -n dnode3 -c alternativeRole -v 0
+system sh/cfg.sh -n dnode4 -c alternativeRole -v 0
+
+system sh/cfg.sh -n dnode1 -c maxtablesPerVnode -v 1000
+system sh/cfg.sh -n dnode2 -c maxtablesPerVnode -v 1000
+system sh/cfg.sh -n dnode3 -c maxtablesPerVnode -v 1000
+system sh/cfg.sh -n dnode4 -c maxtablesPerVnode -v 1000
+
+system sh/cfg.sh -n dnode1 -c arbitrator -v $arbitrator
+system sh/cfg.sh -n dnode2 -c arbitrator -v $arbitrator
+system sh/cfg.sh -n dnode3 -c arbitrator -v $arbitrator
+system sh/cfg.sh -n dnode4 -c arbitrator -v $arbitrator
+
+print ============== step0: start tarbitrator
+system sh/exec_tarbitrator.sh -s start
+
+print ============== step1: start dnode1/dnode2/dnode3
+system sh/exec.sh -n dnode1 -s start
+system sh/exec.sh -n dnode2 -s start
+system sh/exec.sh -n dnode3 -s start
+sleep 3000
+sql connect
+sql create dnode $hostname2
+sql create dnode $hostname3
+sleep 3000
+
+print ============== step2: create db1 with replica 3
+$db = db1
+print create database $db replica 3
+#sql create database $db replica 3 maxTables $totalTableNum
+sql create database $db replica 3
+sql use $db
+
+print ============== step3: create stable stb1
+$stb = stb1
+sql create table $stb (ts timestamp, c1 int, c2 int) tags(t1 int, t2 binary(8))
+
+print ============== step4: start 10 client1/ 10 client2/ 10 client3/ 10 client4/ 1 client5
+run_back unique/cluster/client1_0.sim
+#run_back unique/cluster/client1_1.sim
+#run_back unique/big_cluster/client1_2.sim
+#run_back unique/big_cluster/client1_3.sim
+#run_back unique/big_cluster/client1_4.sim
+#run_back unique/big_cluster/client1_5.sim
+#run_back unique/big_cluster/client1_6.sim
+#run_back unique/big_cluster/client1_7.sim
+#run_back unique/big_cluster/client1_8.sim
+#run_back unique/big_cluster/client1_9.sim
+
+
+print wait for a while to let clients start insert data
+sleep 5000    
+
+$loop_cnt = 0
+loop_cluster_do:
+print **** **** **** START loop cluster do **** **** **** ****
+print ============== step5: start dnode4 and add into cluster, then wait dnode4 ready
+system sh/exec.sh -n dnode4 -s start
+sql create dnode $hostname4
+
+wait_dnode4_ready_0:
+$cnt = $cnt + 1
+if $cnt == 10 then
+  return -1
+endi
+sql show dnodes
+if $rows != 4 then
+  sleep 2000
+  goto wait_dnode4_ready_0
+endi
+print $data0_1  $data1_1  $data2_1  $data3_1  $data4_1
+print $data0_2  $data1_2  $data2_2  $data3_2  $data4_2
+print $data0_3  $data1_3  $data2_3  $data3_3  $data4_3
+print $data0_4  $data1_4  $data2_4  $data3_4  $data4_4
+$dnode1Status = $data4_1  
+$dnode2Status = $data4_2
+$dnode3Status = $data4_3  
+#$dnode4Status = $data4_4 
+
+if $loop_cnt == 0 then
+  $dnode4Status = $data4_4 
+elif $loop_cnt == 1 then
+  $dnode4Status = $data4_6 
+elif $loop_cnt == 2 then
+  $dnode4Status = $data4_8 
+else then
+  print **** **** **** END loop cluster do 2**** **** **** ****
+  return
+endi
+
+if $dnode4Status != ready then
+  sleep 2000
+  goto wait_dnode4_ready_0
+endi
+
+
+print ============== step6: stop and drop dnode1, then remove data dir of dnode1 
+system sh/exec.sh -n dnode1 -s stop -x SIGINT
+
+$cnt = 0
+wait_dnode1_offline_0:
+$cnt = $cnt + 1
+if $cnt == 10 then
+  return -1
+endi
+sql show dnodes
+if $rows != 4 then
+  sleep 2000
+  goto wait_dnode1_offline_0
+endi
+print $data0_1  $data1_1  $data2_1  $data3_1  $data4_1
+print $data0_2  $data1_2  $data2_2  $data3_2  $data4_2
+print $data0_3  $data1_3  $data2_3  $data3_3  $data4_3
+print $data0_4  $data1_4  $data2_4  $data3_4  $data4_4
+ 
+$dnode2Status = $data4_2
+$dnode3Status = $data4_3  
+$dnode4Status = $data4_4 
+
+if $loop_cnt == 0 then
+  $dnode1Status = $data4_1 
+elif $loop_cnt == 1 then
+  $dnode1Status = $data4_5 
+elif $loop_cnt == 2 then
+  $dnode1Status = $data4_7 
+elif $loop_cnt == 3 then
+  $dnode1Status = $data4_9 
+else then
+  print **** **** **** END loop cluster do 1**** **** **** ****
+  return
+endi
+
+if $dnode1Status != offline then
+  sleep 2000
+  goto wait_dnode1_offline_0
+endi
+
+sql drop dnode $hostname1
+system rm -rf ../../../sim/dnode1
+
+
+print ============== step7: stop dnode2, because mnodes < 50%, so clusert don't provide services
+system sh/exec.sh -n dnode2 -s stop -x SIGINT
+
+sql show dnodes -x wait_dnode2_offline_0
+if $rows != 3 then
+  sleep 2000
+  goto wait_dnode2_offline_0
+endi
+wait_dnode2_offline_0:
+
+#$cnt = 0
+#wait_dnode2_offline_0:
+#$cnt = $cnt + 1
+#if $cnt == 10 then
+#  return -1
+#endi
+#sql show dnodes -x wait_dnode2_offline_0
+#if $rows != 3 then
+#  sleep 2000
+#  goto wait_dnode2_offline_0
+#endi
+#print $data0_1  $data1_1  $data2_1  $data3_1  $data4_1
+#print $data0_2  $data1_2  $data2_2  $data3_2  $data4_2
+#print $data0_3  $data1_3  $data2_3  $data3_3  $data4_3
+#print $data0_4  $data1_4  $data2_4  $data3_4  $data4_4
+#$dnode1Status = $data4_1  
+#$dnode2Status = $data4_2
+#$dnode3Status = $data4_3  
+#$dnode4Status = $data4_4 
+#
+#if $dnode2Status != offline then
+#  sleep 2000
+#  goto wait_dnode1_offline_0
+#endi
+
+print ============== step8: restart dnode2, then wait sync end
+system sh/exec.sh -n dnode2 -s start
+
+$cnt = 0
+wait_dnode2_ready_0:
+$cnt = $cnt + 1
+if $cnt == 10 then
+  return -1
+endi
+sql show dnodes
+if $rows != 3 then
+  sleep 2000
+  goto wait_dnode2_ready_0
+endi
+print $data0_1  $data1_1  $data2_1  $data3_1  $data4_1
+print $data0_2  $data1_2  $data2_2  $data3_2  $data4_2
+print $data0_3  $data1_3  $data2_3  $data3_3  $data4_3
+print $data0_4  $data1_4  $data2_4  $data3_4  $data4_4
+$dnode1Status = $data4_1  
+$dnode2Status = $data4_2
+$dnode3Status = $data4_3  
+$dnode4Status = $data4_4 
+
+if $dnode2Status != ready then
+  sleep 2000
+  goto wait_dnode2_ready_0
+endi
+
+
+print ============== step9: stop dnode3, then wait sync end
+system sh/exec.sh -n dnode3 -s stop -x SIGINT
+
+$cnt = 0
+wait_dnode3_offline_0:
+$cnt = $cnt + 1
+if $cnt == 10 then
+  return -1
+endi
+sql show dnodes
+if $rows != 3 then
+  sleep 2000
+  goto wait_dnode3_offline_0
+endi
+print $data0_1  $data1_1  $data2_1  $data3_1  $data4_1
+print $data0_2  $data1_2  $data2_2  $data3_2  $data4_2
+print $data0_3  $data1_3  $data2_3  $data3_3  $data4_3
+print $data0_4  $data1_4  $data2_4  $data3_4  $data4_4
+$dnode1Status = $data4_1  
+$dnode2Status = $data4_2
+$dnode3Status = $data4_3  
+$dnode4Status = $data4_4 
+
+if $dnode3Status != offline then
+  sleep 2000
+  goto wait_dnode3_offline_0
+endi
+
+print ============== step10: restart dnode3, then wait sync end
+system sh/exec.sh -n dnode3 -s start
+
+$cnt = 0
+wait_dnode3_ready_0:
+$cnt = $cnt + 1
+if $cnt == 10 then
+  return -1
+endi
+sql show dnodes
+if $rows != 3 then
+  sleep 2000
+  goto wait_dnode3_ready_0
+endi
+print $data0_1  $data1_1  $data2_1  $data3_1  $data4_1
+print $data0_2  $data1_2  $data2_2  $data3_2  $data4_2
+print $data0_3  $data1_3  $data2_3  $data3_3  $data4_3
+print $data0_4  $data1_4  $data2_4  $data3_4  $data4_4
+$dnode1Status = $data4_1  
+$dnode2Status = $data4_2
+$dnode3Status = $data4_3  
+$dnode4Status = $data4_4 
+
+if $dnode3Status != ready then
+  sleep 2000
+  goto wait_dnode3_ready_0
+endi
+
+print ============== step11: stop dnode4, then wait sync end
+system sh/exec.sh -n dnode4 -s stop -x SIGINT
+
+$cnt = 0
+wait_dnode4_offline_0:
+$cnt = $cnt + 1
+if $cnt == 10 then
+  return -1
+endi
+sql show dnodes
+if $rows != 3 then
+  sleep 2000
+  goto wait_dnode4_offline_0
+endi
+print $data0_1  $data1_1  $data2_1  $data3_1  $data4_1
+print $data0_2  $data1_2  $data2_2  $data3_2  $data4_2
+print $data0_3  $data1_3  $data2_3  $data3_3  $data4_3
+print $data0_4  $data1_4  $data2_4  $data3_4  $data4_4
+$dnode1Status = $data4_1  
+$dnode2Status = $data4_2
+$dnode3Status = $data4_3  
+#$dnode4Status = $data4_4 
+
+if $loop_cnt == 0 then
+  $dnode4Status = $data4_4 
+elif $loop_cnt == 1 then
+  $dnode4Status = $data4_6 
+elif $loop_cnt == 2 then
+  $dnode4Status = $data4_8 
+else then
+  print **** **** **** END loop cluster do 2**** **** **** ****
+  return
+endi
+
+if $dnode4Status != offline then
+  sleep 2000
+  goto wait_dnode4_offline_0
+endi
+
+print ============== step12: restart dnode4, then wait sync end
+system sh/exec.sh -n dnode4 -s start
+
+$cnt = 0
+wait_dnode4_ready_0:
+$cnt = $cnt + 1
+if $cnt == 10 then
+  return -1
+endi
+sql show dnodes
+if $rows != 3 then
+  sleep 2000
+  goto wait_dnode4_ready_0
+endi
+print $data0_1  $data1_1  $data2_1  $data3_1  $data4_1
+print $data0_2  $data1_2  $data2_2  $data3_2  $data4_2
+print $data0_3  $data1_3  $data2_3  $data3_3  $data4_3
+print $data0_4  $data1_4  $data2_4  $data3_4  $data4_4
+$dnode1Status = $data4_1  
+$dnode2Status = $data4_2
+$dnode3Status = $data4_3  
+#$dnode4Status = $data4_4 
+
+if $loop_cnt == 0 then
+  $dnode4Status = $data4_4 
+elif $loop_cnt == 1 then
+  $dnode4Status = $data4_6 
+elif $loop_cnt == 2 then
+  $dnode4Status = $data4_8 
+else then
+  print **** **** **** END loop cluster do 2**** **** **** ****
+  return
+endi
+
+if $dnode4Status != ready then
+  sleep 2000
+  goto wait_dnode4_ready_0
+endi
+
+print ============== step13: alter replica 2
+sql alter database $db replica 2
+sql show database
+print $data0_1  $data1_1  $data2_1  $data3_1  $data4_1
+
+if $data0_5 != 2 then
+  print rplica is not modify to 2, error!!!!!!
+  return
+endi
+
+print ============== step14: stop and drop dnode4, then remove data dir of dnode4 
+system sh/exec.sh -n dnode4 -s stop -x SIGINT
+
+$cnt = 0
+wait_dnode4_offline_1:
+$cnt = $cnt + 1
+if $cnt == 10 then
+  return -1
+endi
+sql show dnodes
+if $rows != 3 then
+  sleep 2000
+  goto wait_dnode4_offline_1
+endi
+print $data0_1  $data1_1  $data2_1  $data3_1  $data4_1
+print $data0_2  $data1_2  $data2_2  $data3_2  $data4_2
+print $data0_3  $data1_3  $data2_3  $data3_3  $data4_3
+print $data0_4  $data1_4  $data2_4  $data3_4  $data4_4
+ 
+$dnode2Status = $data4_2
+$dnode3Status = $data4_3  
+#$dnode4Status = $data4_4 
+
+if $loop_cnt == 0 then
+  $dnode4Status = $data4_4 
+elif $loop_cnt == 1 then
+  $dnode4Status = $data4_6 
+elif $loop_cnt == 2 then
+  $dnode4Status = $data4_8 
+else then
+  print **** **** **** END loop cluster do 2**** **** **** ****
+  return
+endi
+
+if $dnode4Status != offline then
+  sleep 2000
+  goto wait_dnode4_offline_1
+endi
+
+sql drop dnode $hostname4
+system rm -rf ../../../sim/dnode4
+
+
+print ============== step15: alter replica 1
+sql alter database $db replica 1
+sql show database
+print $data0_1  $data1_1  $data2_1  $data3_1  $data4_1
+
+if $data0_5 != 1 then
+  print rplica is not modify to 1, error!!!!!!
+  return
+endi
+
+
+print ============== step16: alter replica 2
+sql alter database $db replica 1
+sql show database
+print $data0_1  $data1_1  $data2_1  $data3_1  $data4_1
+
+if $data0_5 != 2 then
+  print rplica is not modify to 2, error!!!!!!
+  return
+endi
+
+print ============== step17: start dnode1 and add into cluster, then wait dnode1 ready
+system sh/exec.sh -n dnode1 -s start
+sql create dnode $hostname1
+
+wait_dnode1_ready_0:
+$cnt = $cnt + 1
+if $cnt == 10 then
+  return -1
+endi
+sql show dnodes
+if $rows != 3 then
+  sleep 2000
+  goto wait_dnode1_ready_0
+endi
+print $data0_1  $data1_1  $data2_1  $data3_1  $data4_1
+print $data0_2  $data1_2  $data2_2  $data3_2  $data4_2
+print $data0_3  $data1_3  $data2_3  $data3_3  $data4_3
+print $data0_4  $data1_4  $data2_4  $data3_4  $data4_4
+#$dnode1Status = $data4_1  
+$dnode2Status = $data4_2
+$dnode3Status = $data4_3  
+$dnode4Status = $data4_4 
+
+if $loop_cnt == 0 then
+  $dnode1Status = $data4_1 
+elif $loop_cnt == 1 then
+  $dnode1Status = $data4_5 
+elif $loop_cnt == 2 then
+  $dnode1Status = $data4_7 
+elif $loop_cnt == 3 then
+  $dnode1Status = $data4_9 
+else then
+  print **** **** **** END loop cluster do 3**** **** **** ****
+  return
+endi
+
+if $dnode1Status != ready then
+  sleep 2000
+  goto wait_dnode1_ready_0
+endi
+
+print ============== step18: alter replica 3
+sql alter database $db replica 3
+sql show database
+print $data0_1  $data1_1  $data2_1  $data3_1  $data4_1
+
+if $data0_5 != 3 then
+  print rplica is not modify to 3, error!!!!!!
+  return
+endi
+
+$loop_cnt = $loop_cnt + 1
+goto loop_cluster_do