[jboss-jira] [JBoss JIRA] Commented: (JBAS-7096) SLSB with dependency on HASingletonDeployer / Barrier results in error message on second node on startup

[Scott Marlow (JIRA)]

    [ https://jira.jboss.org/browse/JBAS-7096?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12556890#action_12556890 ] 

Scott Marlow commented on JBAS-7096:
------------------------------------

I might of recreated this issue in the Infinispan integration with AS 6 effort.  It could also be that I'm seeing something different but just in case, I'll add some notes here.

1.  I enabled trace logging in jboss-logging.xml
   <logger category="org.jboss.ha">
     <level name="TRACE" />
   </logger>

2.  I added more debug logging to the HASingletonController and family (I'm on a private branch, so these changes won't show up for a bit).  

3.  The debug logging is showing that the singleton controller is make a selection using DistributedReplicantManager.

4.  The DistributedReplicantManager.isMasterReplica("HASingletonDeployer") is returning true because no replicants are found for service "HASingletonDeployer".

That is the evidence at hand, not sure of anything else.  If others are hitting this issue, trying enabling trace logging for "org.jboss.ha" and look for the following server.log output.  If you see the following, we are seeing the same problem:

2010-10-13 22:18:52,354 TRACE [org.jboss.ha.core.framework.server.DistributedReplicantManagerImpl.DefaultPartition] (AsynchKeyChangeHandler Thread) no replicants, key=HASingletonDeployer, isMasterReplica=true

I'm not sure if the fix yet but wanted to share my findings in case we can learn from them.

One thing that I'm curious about is if we could switch to using the ha partition notion of being master instead.  That would be the first (or oldest surviving) node in the cluster.

> SLSB with dependency on HASingletonDeployer / Barrier results in error message on second node on startup
> --------------------------------------------------------------------------------------------------------
>
>                 Key: JBAS-7096
>                 URL: https://jira.jboss.org/browse/JBAS-7096
>             Project: JBoss Application Server
>          Issue Type: Bug
>      Security Level: Public(Everyone can see) 
>          Components: Clustering
>    Affects Versions: JBossAS-5.1.0.GA
>         Environment: RHEL 4 Update 7 (and OSX 10.5.7), Sun JRE 1.6.0_12, JBoss 5.1.0 - "all" configuration
>            Reporter: Richard Taylor
>            Assignee: Scott Marlow
>            Priority: Minor
>             Fix For: TBD
>
>
> Basic cluster, two JBoss 5.1.0 nodes, "all" configuration.   First node starts fine.  Second node starts, however at the end of starting, any SLSBs that depend on the barrier result in an error being logged at startup (on the second node). 
> Example:
> DEPLOYMENTS MISSING DEPENDENCIES:
>   Deployment "jboss.j2ee:ear=app-ear.ear,jar=app-ejb.jar,name=BoardPollerMDB,service=EJB3" is missing the following dependencies:
>     Dependency "<UNKNOWN jboss.j2ee:ear=app-ear.ear,jar=app-ejb.jar,name=BoardPollerMDB,service=EJB3>" (should be in state "Described", bu
> t is actually in state "** UNRESOLVED Demands 'jboss.ha:service=HASingletonDeployer,type=Barrier' **")
> DEPLOYMENTS IN ERROR:
>   Deployment "<UNKNOWN jboss.j2ee:ear=app-ear.ear,jar=app-ejb.jar,name=BoardPollerMDB,service=EJB3>" is in error due to the following reas
> on(s): ** UNRESOLVED Demands 'jboss.ha:service=HASingletonDeployer,type=Barrier' **
> The SLSBs on node 2 do properly start in a failover scenario when node 1 goes down.  So functionally things seem OK.  However the error message is a little concerning and it leaves us wondering if there is more to the problem.

-- 
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: https://jira.jboss.org/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira