Merge pull request #3439 from taosdata/feature/crash

TD-1363  Ensure the sync of mnode not interrupted

src/sync/src/syncMain.c

@@ -797,9 +797,11 @@ static void syncRecoverFromMaster(SSyncPeer *pPeer) {
   }
 
   taosTmrStopA(&pPeer->timer);
-  if (tsSyncNum >= tsMaxSyncNum) {
+
+  // Ensure the sync of mnode not interrupted
+  if (pNode->vgId != 1 && tsSyncNum >= tsMaxSyncNum) {
     sInfo("%s, %d syncs are in process, try later", pPeer->id, tsSyncNum);
-    taosTmrReset(syncTryRecoverFromMaster, 500 + (pNode->vgId*10)%200, pPeer, syncTmrCtrl, &pPeer->timer);
+    taosTmrReset(syncTryRecoverFromMaster, 500 + (pNode->vgId * 10) % 200, pPeer, syncTmrCtrl, &pPeer->timer);
     return;
   }
 

tests/script/jenkins/basic.txt

@@ -253,6 +253,7 @@ cd ../../../debug; make
 ./test.sh -f unique/cluster/balance2.sim
 ./test.sh -f unique/cluster/balance3.sim
 ./test.sh -f unique/cluster/cache.sim
+./test.sh -f unique/cluster/vgroup100.sim
 
 ./test.sh -f unique/column/replica3.sim
 

tests/script/unique/cluster/vgroup100.sim

+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/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 dnode1 -c maxTables -v 4
+system sh/cfg.sh -n dnode2 -c maxTables -v 4
+system sh/cfg.sh -n dnode3 -c maxTables -v 4
+
+system sh/cfg.sh -n dnode1 -c mnodeEqualVnodeNum -v 0
+system sh/cfg.sh -n dnode2 -c mnodeEqualVnodeNum -v 0
+system sh/cfg.sh -n dnode3 -c mnodeEqualVnodeNum -v 0
+
+print ============================== step1
+system sh/exec.sh -n dnode1 -s start
+sleep 2000
+sql connect
+
+print ============================== step2
+print ========= start dnode2
+sql create dnode $hostname2
+system sh/exec.sh -n dnode2 -s start
+sql create dnode $hostname3
+system sh/exec.sh -n dnode3 -s start
+
+sleep 5000
+sql show mnodes
+$dnode1Role = $data2_1
+$dnode2Role = $data2_2
+$dnode3Role = $data2_3
+
+print $dnode1Role
+print $dnode2Role 
+print $dnode3Role 
+
+print ============================== step3
+$count = 2
+while $count < 102
+  $db = d . $count
+  $tb = $db . .t
+  sql create database $db replica 3 cache 1 blocks 3
+  sql create table $tb (ts timestamp, i int)
+  sql insert into $tb values(now, 1)
+  $count = $count + 1
+  print insert into $tb values(now, 1) ==> finished
+endw
+
+print ============================== step4
+
+$count = 2
+while $count < 102
+  $db = d . $count
+  $tb = $db . .t
+  sql select * from $tb
+  if $rows != 1 then
+    print select * from $tb
+    return -1
+  endi
+  $count = $count + 1
+  print select * from $tb ==> rows: $rows 
+endw
+
+print ============================== step5
+system sh/exec.sh -n dnode1 -s stop  -x SIGINT
+system sh/exec.sh -n dnode2 -s stop  -x SIGINT
+system sh/exec.sh -n dnode3 -s stop  -x SIGINT
+sleep 5000
+
+print ============================== step6
+system sh/exec.sh -n dnode1 -s start
+system sh/exec.sh -n dnode2 -s start
+system sh/exec.sh -n dnode3 -s start
+sleep 3000
+
+print ============================== step7
+
+$x = 0
+show7: 
+	$x = $x + 1
+	sleep 2000
+	if $x == 50 then
+		return -1
+	endi
+	
+sql show mnodes -x show7
+$dnode1Role = $data2_1
+$dnode2Role = $data2_2
+$dnode3Role = $data2_3
+if $dnode1Role != master then
+	goto show7
+endi
+if $dnode2Role != slave then
+	goto show7
+endi
+if $dnode2Role != slave then
+	goto show7
+endi
+
+print ============================== step8
+$x = 0
+show8: 
+	$x = $x + 1
+	sleep 2000
+	if $x == 20 then
+		return -1
+	endi
+
+$count = 2
+while $count < 102
+  $db = d . $count
+  $tb = $db . .t
+  sql select * from $tb
+  if $rows != 1 then
+    print select * from $tb
+    goto show8
+  endi
+  $count = $count + 1
+  print select * from $tb ==> rows: $rows 
+endw
+
+system sh/exec.sh -n dnode1 -s stop  -x SIGINT
+system sh/exec.sh -n dnode2 -s stop  -x SIGINT
+system sh/exec.sh -n dnode3 -s stop  -x SIGINT
\ No newline at end of file