[jboss-jira] [JBoss JIRA] (WFLY-125) Attribute passivate-events-on-replicate of cluster-passivation-store can't be change via drm or cli

[Richard Achmatowicz (JIRA)]

    [ https://issues.jboss.org/browse/WFLY-125?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12804476#comment-12804476 ] 

Richard Achmatowicz commented on WFLY-125:
------------------------------------------

OK, in having a closer look at this issue on Brian's prompting, we need to confirm whether or not this is a bug. After having a second look at Paul's original code, i'm not so sure that it is.

Passivation stores for SFSBs are created using a BackingCacheEntryStoreSource service to create the stores. The stores are created upon SFSB deployment and per-SFSB type.

Currently, when you change the value of any of the PassivationStore attributes, the handler will look up the service for the BackingCacheEntryStoreSource and will update the configuration of that source with the new attribute. This change will be reflected in all subsequent creations of passivation stores (BackingCacheEntryStore) for SFSBs. This is why the attributes were flagged as RESTART_NONE and no reload is required when the attributes are changed. The change will not affect the passivation status of SFSBs deployed before the change.

So, if you are expecting this flag to change the passivation status for SFSBs which have already been deployed, it will not do so - only for deployments which occur after the change. Given that this attribute applies to the passivation store itself, this seems like the reasonable behaviour.     

Ondrej, I was looking at your test cases in ClusterPassivationTestCase:
testPassivationOverNodesPassivated
testPassivationOverNodesNotPassivated
testMoreDataPassivation

In the first two test cases, (i) the SFSBs are deployed, (ii) the attributes are changed (iii) the servers are reloaded (iv) the test is run.
In the third test case, (i) the SFSBs are deployed, (ii) the attributes are changed, (iii) the test is run.

Which test case was being used to identify this bug? This last test case changes the attributes after deployment, and so the passivation behaviour should not change, given what was mentioned previously. 











                
> Attribute passivate-events-on-replicate of cluster-passivation-store can't be change via drm or cli
> ---------------------------------------------------------------------------------------------------
>
>                 Key: WFLY-125
>                 URL: https://issues.jboss.org/browse/WFLY-125
>             Project: WildFly
>          Issue Type: Bug
>          Components: EJB
>            Reporter: Ondřej Chaloupka
>            Assignee: Richard Achmatowicz
>              Labels: eap6_need_triage
>             Fix For: 8.0.0.Beta1
>
>
> Setting attribute passivate-events-on-replicate via drm or cli does not take affect until operation :reload is used.
> In case that the attribute ( <cluster-passivation-store name="infinispan" cache-container="ejb" passivate-events-on-replicate="false" />) is set directly in standalone*.xml configuration file the supposed functionality is working.
> The test for this could be checked on:
> https://github.com/ochaloup/jboss-as/blob/JBQA-5855-test-migration-passivation-in-cluster/testsuite/integration/clust/src/test/java/org/jboss/as/test/clustering/cluster/ejb3/stateful/passivation/ClusterPassivationTestCase.java

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira