[
https://issues.apache.org/jira/browse/ARIES-1732?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16112670#comment-16112670
]
Paolo Antinori commented on ARIES-1732:
---------------------------------------
Okay, thanks for pointing that out.
I have now removed the modification to `ManagedObjectManager` since I think I
have added there initially, due to me testing the wrong way.
Now it alters only the behavior of `CmPropertyPlaceholder.updated` and adds a
tid bit of logging in `ManagedObjectManager`.
https://github.com/apache/aries/compare/trunk...paoloantinori:cm_race_condition?expand=1
With just this, I'm able to avoid the issue with the reproducer in
https://issues.jboss.org/browse/ENTESB-7141
> blueprint-cm - race condition
> -----------------------------
>
> Key: ARIES-1732
> URL: https://issues.apache.org/jira/browse/ARIES-1732
> Project: Aries
> Issue Type: Bug
> Components: Blueprint
> Reporter: Paolo Antinori
>
> There is a possible race condition in
> `org.apache.aries.blueprint.compendium.cm.CmPropertyPlaceholder` and
> `org.apache.aries.blueprint.compendium.cm.ManagedObjectManager`
> There are situations where `ConfigAdmin` service might not be available and
> returns `null` when queried for the initial `config`.
> In other cases instead, `CmPropertyPlaceholder` receives updates from
> `ConfigAdmin` thread even before it has completed it's initialization.
> The code doesn't discriminate if `config` is `null` or not, and treats any
> value as valid, marking `CmPropertyPlaceholder` instance as initialized.
> This leads to spotty errors.
> The suggested fix, that simply prevents that `update(null)` invocation mark
> the `CmPropertyPlaceholder` instance as initialized, addresses the issue.
> Code here:
> https://github.com/apache/aries/compare/trunk...paoloantinori:ENTESB-7141?expand=1
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
