[jbossts-issues] [JBoss JIRA] Commented: (JBTM-468) make RecoveryManager restartable

[Andrew Dinn (JIRA)]

    [ https://jira.jboss.org/jira/browse/JBTM-468?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12448889#action_12448889 ] 

Andrew Dinn commented on JBTM-468:
----------------------------------

It is perhaps worth noting that terminate method comes in two flavours, one of which takes a flag specifying whether or not to wait for any in-progress scan to terminate and the other of which takes no arguments and always waits. In the case where the transaction manager is being stopped in order to allow for the possibility of a restart  the wait is most definitely _required_.

> make RecoveryManager restartable
> --------------------------------
>
>                 Key: JBTM-468
>                 URL: https://jira.jboss.org/jira/browse/JBTM-468
>             Project: JBoss Transaction Manager
>          Issue Type: Feature Request
>      Security Level: Public(Everyone can see) 
>          Components: Application Server Integration, Recovery
>    Affects Versions: 4.5
>            Reporter: Jonathan Halliday
>            Assignee: Mark Little
>             Fix For: 4.6
>
>
> Stopping and restarting the transaction manager inside JBossAS causes problems, because the recovery system does not like to be restarted once it is stopped:
> java.lang.IllegalThreadStateException
> 	at java.lang.Thread.start(Thread.java:571)
> 	at com.arjuna.ats.internal.arjuna.recovery.RecoveryManagerImple.start(RecoveryManagerImple.java:236)
> 	at com.arjuna.ats.arjuna.recovery.RecoveryManager.startRecoveryManagerThread(RecoveryManager.java:226)
> 	at com.arjuna.ats.jbossatx.jts.TransactionManagerService.start(TransactionManagerService.java:285)
> Options would seem to be disposing the static singleton _recoveryManager in RecoveryManager and recreating it on the next start, effectively causing a new RecoveryManagerImple to be instantiated, or keep the RecoveryManagerImple instance and rework its impl so that the  PeriodicRecovery _periodicRecovery is disposed and recreated on restart.
> Either one has slightly odd semantics as the PeriodicRecovery is a Thread (which can't be restarted so has to be replaced) and also the object that has the module list and the server socket, so disposal and/or recreation of this state will be needed on restart. Maybe better to rename stop() to dispose() or something similarly final, to indicate the semantics are 'throw this away and start again' rather than giving the impression that stop/start == suspend/resume which is misleading.

-- 
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: https://jira.jboss.org/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira