[
https://issues.apache.org/jira/browse/FELIX-6243?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17061728#comment-17061728
]
Karl Pauls commented on FELIX-6243:
-----------------------------------
Hm, I'm not sure I understand the problem yet. The resolver threads should
timeout after a minute which should make this similar to the
RESOLVER_PARALLELISM=1 case (after a minute of inactivity). I guess the
question is what is holding on to the felix object in the first place...
> Link to root Felix object leaks to GC root
> ------------------------------------------
>
> Key: FELIX-6243
> URL: https://issues.apache.org/jira/browse/FELIX-6243
> Project: Felix
> Issue Type: Bug
> Affects Versions: framework-6.0.3
> Reporter: Andrey Davydov
> Priority: Major
> Attachments: felix.png
>
>
> Tested on java8.
> Felix object creates StatefulResolver and then StatefulResolver get link to
> Felix.
> StatefulResolver creates Executor in getExecutor() using nested class as
> ThreadFactory (when FelixConstants.RESOLVER_PARALLELISM greater then 1)
>
> So every thread in Executor has link to Executor
> Executor has link to ThreadFactory
> ThreadFactory has link to StatefulResolver, because it is nested
> StatefulResolver has link to Felix
>
> And it is the way from GC root to Felix and GC can't collect Felix after
> Felix.stop() and all external refs to Felix cut.
>
>
>
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
