[
http://jira.jboss.com/jira/browse/JBAS-4314?page=all ]
Paul Ferraro resolved JBAS-4314.
--------------------------------
Resolution: Out of Date
This is no longer an issue. Local and remote replicant maps now use concurrent classes,
and no longer synchronize in a problematic fashion.
A deadlock condition in DistributedReplicantManagerImpl
-------------------------------------------------------
Key: JBAS-4314
URL:
http://jira.jboss.com/jira/browse/JBAS-4314
Project: JBoss Application Server
Issue Type: Bug
Security Level: Public(Everyone can see)
Components: Clustering
Affects Versions: JBossAS-4.0.2 Final
Environment: Linux 2.6.x, Sun-JVM 1.5.0_11-b03, 1.6.0-b105, JBoss 4.0.2.final.
Reporter: Bernd Köcke
Assigned To: Paul Ferraro
Attachments: DistributedReplicantManagerImpl.java.patch
While working on JBAS-4313 I found a deadlock condition in
DistributedReplicantManagerImpl. All methods are using the lock order localReplicants
-> replicants. But method purgeDeadMembers uses replicants -> localReplicants. To
solve the problem I moved the notifyKeyListeners out of the
synchronized(replicants)-block.
Newer versions of JBoss are using maps from the concurrent-package and not the normal
java.util.HashMap. Maybe this solves the problem, too.
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
http://jira.jboss.com/jira/secure/Administrators.jspa
-
For more information on JIRA, see:
http://www.atlassian.com/software/jira