Scenario:
- Configure two EAP servers in replicated HA topology
- On backup set max-saved-replicated-journal-size=2, allow-failback=true and restart-backup=true
- Do sequence: kill live, start live, kill live, start live...
Expectation: When number of saved replicated journals exceeds 2, the backup should not be restarted but stopped.
Actual state: max-saved-replicated-journal-size is ignored and backup is not stopped. Number of journals grows ad infinitum.
I think that root cause is in SharedNothingLiveActivation. In the condition [1] there should be OR between !isRestartBackup and check if the number of journals was exceeded.
[1]
//if we have to many backups kept or are not configured to restart just stop, otherwise restart as a backup if (!replicatedPolicy.getReplicaPolicy().isRestartBackup() && activeMQServer.countNumberOfCopiedJournals() >= replicatedPolicy.getReplicaPolicy().getMaxSavedReplicatedJournalsSize() && replicatedPolicy.getReplicaPolicy().getMaxSavedReplicatedJournalsSize() >= 0) { activeMQServer.stop(true); ActiveMQServerLogger.LOGGER.stopReplicatedBackupAfterFailback(); } else { activeMQServer.stop(true); ActiveMQServerLogger.LOGGER.restartingReplicatedBackupAfterFailback(); activeMQServer.setHAPolicy(replicatedPolicy.getReplicaPolicy()); activeMQServer.start(); }
- clones
-
JBEAP-5261 (7.1.0) max-saved-replicated-journal-size is ignored
- Verified
- is incorporated by
-
JBEAP-4679 (7.0.z) Upgrade Artemis from 1.1.0.SP17 to 1.1.0.SP18
- Verified
- relates to
-
JBEAP-5164 [GSS](7.0.z) Server should always remove old files beyond getMaxSavedReplicatedJournalsSize
- Verified
- is caused by
-
ARTEMIS-612 Loading...