Details
-
Bug
-
Status: Closed
-
Major
-
Resolution: Fixed
-
0.31
-
None
Description
The log for node leaving the group only issued when role is changed from any value other then UNREACHABLE to UNREACHABLE .
Test does not wait or verify that remote node CO role attribute is set to REPLICA state. Thus, if node2 is stopped immediately after start, the remote node CO might not be able to detect the role change as it will be always in UNREACHABLE role.
Test needs to verify that remote node CO role is REPLICA before stopping the node.
Here is an example of test failure
Argument(s) are different! Wanted: eventLogger.message( LogSubject does not match. Expected: [grp(/group)] , actual : [vh(/node2)/ms(BDBConfigurationStore)] , Expected message does not match. Expected: HA-1006 : Left : Node : 'node2' (localhost:10001), actual: CFG-1003 : ClosedExpected hierarchy does not match. Expected: qpid.message.highavailability.left, actual: qpid.message.configstore.close ); -> at org.apache.qpid.server.virtualhostnode.berkeleydb.BDBHAVirtualHostNodeOperationalLoggingTest.testRemoteNodeDetached(BDBHAVirtualHostNodeOperationalLoggingTest.java:272) Actual invocation has different arguments: eventLogger.message( org.apache.qpid.server.logging.subjects.MessageStoreLogSubject@794579f3, CFG-1003 : Closed ); -> at org.apache.qpid.server.virtualhostnode.AbstractVirtualHostNode.closeConfigurationStore(AbstractVirtualHostNode.java:268)