[jboss-jira] [JBoss JIRA] (AS7-5794) JMX over remoting pollutes query results with ModelController objects

Tue Feb 26 09:42:56 EST 2013

    [ https://issues.jboss.org/browse/AS7-5794?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12756779#comment-12756779 ] 

Tristan Tarrant commented on AS7-5794:
--------------------------------------

This still happens 7.2 (actually I tried with EAP 6.1.DR4). Here is a log snippet which shows what happens:

2013-02-26 15:41:01,202 TRACE [ResourceDiscoveryComponent.invoker.daemon-13] (org.infinispan.rhq.CacheDiscovery)- Querying *:type=Cache,component=Cache,manager="DefaultCacheManager",* returned beans: [jboss.jsr77:j2eeType=J2EEDomain,name=jboss.jsr77, org.infinispan:type=Cache,name="___defaultcache(local)",manager="DefaultCacheManager",component=Cache, jboss.jsr77:j2eeType=J2EEServer,name=default, jboss.jsr77:j2eeType=JVM,name=default,J2EEServer=default]
2013-02-26 15:41:01,202 WARN  [ResourceDiscoveryComponent.invoker.daemon-13] (org.infinispan.rhq.CacheDiscovery)- MBeanServer returned spurious object jboss.jsr77:j2eeType=J2EEDomain,name=jboss.jsr77

> JMX over remoting pollutes query results with ModelController objects
> ---------------------------------------------------------------------
>
>                 Key: AS7-5794
>                 URL: https://issues.jboss.org/browse/AS7-5794
>             Project: Application Server 7
>          Issue Type: Bug
>          Components: Domain Management, JMX
>    Affects Versions: 7.1.2.Final (EAP)
>            Reporter: Tristan Tarrant
>            Assignee: Kabir Khan
>             Fix For: 7.3.0.Alpha1
>
>
> When issuing MBean queries over JMX remoting, the AS is returning a list of org.jboss.as.controller.ModelController objects in addition to those matching the query.
> This is a transcript of a chat we had about it on IRC:
> Oct 15 14:51:46 <ttarrant>      darranl, if I issue a jmx query over jmx-remoting I get many more objects than expected
> Oct 15 14:51:55 <ttarrant>      darranl, i.e. ones that do not match the query
> Oct 15 14:52:09 <ttarrant>      darranl, this is with 7.1.x
> Oct 15 14:52:57 <ttarrant>      darranl, this works if I use the standard JMX over RMI
> Oct 15 14:53:08 -->     fnasser (~fnasser at CPE602ad07ab726-CM602ad07ab723.cpe.net.cable.rogers.com) has joined #jboss-as7
> Oct 15 14:53:09 <--     fnasser has quit (Changing host)
> Oct 15 14:53:09 -->     fnasser (~fnasser at redhat/jboss/fnasser) has joined #jboss-as7
> Oct 15 14:53:09 ---     ChanServ gives voice to fnasser
> Oct 15 14:53:16 <ttarrant>      darranl, do you just pass the query the the mbeanserver ?
> Oct 15 14:53:39 <darranl>       ttarrant: what kind of objects? within AS7 I think there are two things that could affec
> t this 1 - The Remoting JMX protocol, 2 - The domain management representation over JMX
> Oct 15 14:53:54 <darranl>       For #1 yes we just pass it to the MBEanServer and return whatever it returns
> Oct 15 14:54:04 <darranl>       if it was a Remoting JMX bug maybe we are messing up the query
> Oct 15 14:54:16 <ttarrant>      darranl, the query is as follows: *:type=CacheManager,component=Interpreter,name=*
> Oct 15 14:54:16 <darranl>       But not sure if #2 could be the reason more is getting added
> Oct 15 14:54:32 <ttarrant>      darranl, wait a sec
> Oct 15 14:54:57 <darranl>       ttarrant: I would suggest getting a Jira raised and assigned to me, I can verify if it i
> s a remoting jmx issue or pass over if it is a domain management integration issue
> Oct 15 14:55:20 <darranl>       regardless of where it is happenign it sounds like you may have discovered a bug 
> Oct 15 14:56:23 <ttarrant>      darranl, let me get the objects it's returning
> Oct 15 14:56:39 <darranl>       ttarrant: when you enable the RMI approach you bypass both Remoting JMX AND the domain m
> anagement integration 
> Oct 15 14:56:42 <darranl>       ok
> Oct 15 15:26:08 <ttarrant>      darranl, ok, my query actually returns the object I queried for and a bunch of org.jboss.as.controller.ModelController (one for each subsystem)
> Oct 15 15:51:24 <darranl>       ttarrant: ok that does then sound like it is the domain management integration that is 'poluting' the query results 
> Oct 15 15:51:53 <darranl>       ttarrant: going RMI was just bypassing that as well 
> Oct 15 15:52:05 <ttarrant>      darranl, shall I open a Jira ?
> Oct 15 15:52:34 <ttarrant>      darranl, I work around the issue by manually filtering the returned objects based on class name
> Oct 15 15:53:23 <darranl>       ttarrant:  it would probably be one for kkhan to look into, think he is just back today so may be worth the Jira so he can have a look once he has caught back up ;-)

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira