refresh cluster state in MtmReplicationStartupHook

knizhnik · knizhnik · commit 0812669a2b29 · 2016-08-23T18:13:34.000+03:00
diff --git a/contrib/mmts/multimaster.c b/contrib/mmts/multimaster.c
@@ -2492,9 +2492,9 @@ MtmReplicationMode MtmGetReplicationMode(int nodeId, sig_atomic_t volatile* shut
 		MtmSleep(STATUS_POLL_DELAY);
 	}
 	if (recovery) { 
-		MTM_LOG1("%d: Restart replication for node %d after end of recovery", MyProcPid, nodeId);
+		MTM_LOG1("%d: Restart replication frim node %d after end of recovery", MyProcPid, nodeId);
 	} else { 
-		MTM_LOG1("%d: Continue replication slot for node %d", MyProcPid, nodeId);
+		MTM_LOG1("%d: Continue replication from node %d", MyProcPid, nodeId);
 	}
 	/* After recovery completion we need to drop all other slots to avoid receive of redundant data */
 	return recovery ? REPLMODE_RECOVERED : REPLMODE_NORMAL;
@@ -2596,7 +2596,10 @@ MtmReplicationStartupHook(struct PGLogicalStartupHookArgs* args)
 			MtmCheckQuorum();
 		} else {
 			MtmUnlock();
-			elog(ERROR, "Disabled node %d tries to reconnect without recovery", MtmReplicationNodeId); 
+			MtmRefreshClusterStatus(true);
+			if (BIT_CHECK(Mtm->disabledNodeMask, MtmReplicationNodeId-1)) {
+				elog(ERROR, "Disabled node %d tries to reconnect without recovery", MtmReplicationNodeId); 
+			}
 		}
 	} else {
 		MTM_LOG1("Node %d start logical replication to node %d in normal mode", MtmNodeId, MtmReplicationNodeId); 
diff --git a/contrib/mmts/pglogical_receiver.c b/contrib/mmts/pglogical_receiver.c
@@ -611,7 +611,7 @@ pglogical_receiver_main(Datum main_arg)
 			{
 				ereport(LOG, (errmsg("%s: COPY Stream has abruptly ended...",
 									 worker_proc)));
-				break;
+				goto OnError;
 			}
 
 			/* Failure when reading copy stream, leave */
diff --git a/contrib/mmts/t/001_basic_recovery.pl b/contrib/mmts/t/001_basic_recovery.pl
@@ -89,7 +89,7 @@
 diag("polling node 2");
 for (my $poller = 0; $poller < 3; $poller++) {
 	my $pollee = 2;
-	ok($cluster->poll($poller, 'postgres', $pollee, 10, 1), "node $pollee is online according to node $poller");
+	ok($cluster->poll($poller, 'postgres', $pollee, 1000, 1), "node $pollee is online according to node $poller");
 }
 
 diag("getting cluster state");

Original file line number	Diff line number	Diff line change
`@@ -611,7 +611,7 @@ pglogical_receiver_main(Datum main_arg)`
`611`	`611`	`{`
`612`	`612`	`ereport(LOG, (errmsg("%s: COPY Stream has abruptly ended...",`
`613`	`613`	`worker_proc)));`
`614`		`- break;`
	`614`	`+ goto OnError;`
`615`	`615`	`}`
`616`	`616`
`617`	`617`	`/* Failure when reading copy stream, leave */`
Original file line number	Diff line number	Diff line change
`@@ -89,7 +89,7 @@`
`89`	`89`	`diag("polling node 2");`
`90`	`90`	`for (my $poller = 0; $poller < 3; $poller++) {`
`91`	`91`	`my $pollee = 2;`
`92`		`- ok($cluster->poll($poller, 'postgres', $pollee, 10, 1), "node $pollee is online according to node $poller");`
	`92`	`+ ok($cluster->poll($poller, 'postgres', $pollee, 1000, 1), "node $pollee is online according to node $poller");`
`93`	`93`	`}`
`94`	`94`
`95`	`95`	`diag("getting cluster state");`