[jboss-jira] [JBoss JIRA] (WFLY-11978) EAR with 2+ JPA does not shutdown cleanly

[Scott Marlow (Jira)]

    [ https://issues.jboss.org/browse/WFLY-11978?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13726617#comment-13726617 ] 

Scott Marlow commented on WFLY-11978:
-------------------------------------

[~dlmiles] does this recreate if the application has not actually serviced any user requests?

I tried to recreate with a very simple jar that contains:
{code}
<?xml version="1.0" encoding="UTF-8"?>
<persistence xmlns="http://java.sun.com/xml/ns/persistence" version="1.0">
  <persistence-unit name="TEST_PU">
    <jta-data-source>java:jboss/datasources/ExampleDS</jta-data-source>
    <shared-cache-mode>ENABLE_SELECTIVE</shared-cache-mode>
    <properties>
    <property name="hibernate.enhancer.enableDirtyTracking" value="true"/>
    <property name="hibernate.generate_statistics" value="true"/>
    <property name="hibernate.cache.use_second_level_cache" value="true"/>
    <property name="hibernate.cache.use_query_cache" value="true"/>
    </properties>
  </persistence-unit>

 <persistence-unit name="second_PU">
    <jta-data-source>java:jboss/datasources/ExampleDS</jta-data-source>
    <properties>
    <property name="hibernate.enhancer.enableDirtyTracking" value="false"/>
    <property name="hibernate.generate_statistics" value="false"/>
    <property name="hibernate.cache.use_second_level_cache" value="true"/>
    <property name="hibernate.cache.use_query_cache" value="true"/>
    </properties>
  </persistence-unit>
</persistence>
{code}

I did a jstack thread dump and don't see the same issue after undeployment (using the latest [WildFly master branch code|https://github.com/wildfly/wildfly]).  

Do you have a simple test ear that recreates this that you could attach?  Best if it uses java:jboss/datasources/ExampleDS, as I show in my ^ persistence.xml.

Also, from a quick look at the Infinispan [https://github.com/infinispan/infinispan/blob/master/core/src/main/java/org/infinispan/factories/AbstractComponentRegistry.java#L356], I don't really understand where the code is that does the notify() to wake this thread up.  I wonder if its a race where the state is ComponentStatus.STOPPING but changes by time wait() is called. 


> EAR with 2+ JPA does not shutdown cleanly
> -----------------------------------------
>
>                 Key: WFLY-11978
>                 URL: https://issues.jboss.org/browse/WFLY-11978
>             Project: WildFly
>          Issue Type: Bug
>          Components: JPA / Hibernate
>    Affects Versions: 16.0.0.Final
>         Environment: Windows JDK8 WF16
>            Reporter: Darryl Miles
>            Assignee: Scott Marlow
>            Priority: Major
>         Attachments: td8948.txt
>
>
> EAR with 2+ JPA does not shutdown cleanly.
> I see in the logs each JPA project have an entry:
> WFLYJPA0011: Stopping Persistence Unit (phase 2 of 2) Service 'blah....jpa.project1'
> WFLYJPA0011: Stopping Persistence Unit (phase 2 of 2) Service 'blah....jpa.project2'
> then a few lines later:
> WFLYJPA0011: Stopping Persistence Unit (phase 1 of 2) Service 'blah....jpa.project2'
> I never see the "phase 1 of 2" entry for "jpa.project1" in the log.
> The container will wait 300 seconds and timeout.
> The management console during the time shows
>   Operation: undeploy
>   Execution Status: awaiting-stablility.
> The container is killed and the configuration.xml still contains the EAR deployment info, as undeploy did not complete.



--
This message was sent by Atlassian Jira
(v7.12.1#712002)