pause running jobs on failed node if pauseTriggeredJobOnFailedNode flag set #176

[pavelkokush]

No description provided.

[jontejj]

Should it catch exceptions and continue to try pausing the next trigger?

[pavelkokush]

then it will be inconsistent state: some triggers paused, some not. Also then we will have same job running at same time on diff nodes. So better to rollback full transaction, as it now?

[jontejj]

Aha! So the transaction is not inside this method. Good!

…

On Wed, Aug 9, 2017, 11:18 pkokush ***@***.***> wrote: ***@***.**** commented on this pull request. ------------------------------ In quartz-core/src/main/java/org/quartz/impl/jdbcjobstore/JobStoreSupport.java <#178 (comment)> : > @@ -3575,6 +3587,17 @@ protected void clusterRecover(Connection conn, List<SchedulerStateRecord> failed .updateTriggerStatesForJobFromOtherState( conn, jKey, STATE_PAUSED, STATE_PAUSED_BLOCKED); + + if (pauseTriggeredJobOnNodeFail) { + if (ftRec.getJobKey() != null) { + log.warn(String.format("job %s paused, because it was running when node %s failed. Make sure that job is NOT running, then unpause it", ftRec.getJobKey(), ftRec.getFireInstanceId())); + + List<OperableTrigger> triggers = getTriggersForJob(conn, ftRec.getJobKey()); + for (OperableTrigger trigger : triggers) { + pauseTrigger(conn, trigger.getKey()); then it will be inconsistent state: some triggers paused, some not. Also then we will have same job running at same time on diff nodes. So better to rollback full transaction, as it now? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#178 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABDq9roDp4pdu7xeFvNG_0TSfeCAnwIQks5sWWtfgaJpZM4OvldA> .

[zemian]

Hi, as I stated in your original issue #176, that this is currently working as expected that you might need additional config change. I will close this PR as invalid for now.