When I created an EAP 5.1 cluster comprising two nodes, n01 and n02, on a single host, when I deployed a bundle (having a helloworld.war), if I tried to undeploy the war file on the file system using rm -fr of the deployment in the live system, the helloworld application still ran from memory, and its helloworld.jsp page continued to return results. JBoss does not try to undeploy the application. After I patched RHQ I was able to successfully undeploy via rm -fr. 2011-10-10 10:07:56,813 ERROR [org.jboss.remoting.transport.socket.ServerThread] (WorkerThread#1[127.0.0.1:39420]) WorkerThread#1[127.0.0.1:39420] failed java.io.NotSerializableException: org.jboss.system.server.profile.repository.NoopProfile Theory is that the calls into the JBoss deployer put JBoss into a bad state, making it fail to undeploy artifacts thereafter.
Created attachment 532137 [details] Fix jboss log spew and undeployment issues
commit e0cd4a2e9a6f6e78da88b23803cd367102aba710 Author: Robert Buck <rbuck> Date: 2011-11-07 16:14:45 -0500 [BZ 751873] Prevent JBoss log spew and undeploy issues; w/o this fix after a bundle is deployed to an EAP cluster and a war is updated, it is no longer possible to undeploy via rm -fr on the command line. This is on branch: bug/751873
setting the assignment field to rbuck since he worked on the fix
*** Bug 773663 has been marked as a duplicate of this bug. ***
Unassigning from rbuck. Its not immediately clear to me whether this ever got merged into master. Targeting it for JON3.1.0 for investigation and retesting/merging as appropriate.
this was already committed to master: commit 8d5f78b194c2e957f2d5778e3a4ead1fdd98713c Author: Robert Buck <rbuck> 2011-11-07 16:14:45 Committer: Robert Buck <rbuck> 2011-11-08 07:03:01 Parent: de2e667b239e40f261e9de116c5149889f217e64 (More wildcard updates propagated upwards from the agent api.) [BZ 751873] Prevent JBoss log spew and undeploy issues; w/o this fix after a bundle is deployed to an EAP cluster and a war is updated, it is no longer possible to undeploy via rm -fr on the command line.
reopened. When jboss eap5.1 is started in clustered mode with 2 nodes, after bundle deployment server is being un-deployed, bin directory is being removed.
(In reply to comment #7) > reopened. > > When jboss eap5.1 is started in clustered mode with 2 nodes, after bundle > deployment server is being un-deployed, bin directory is being removed. Armine, do you have any details on this what you mean by this? From the sound of it, perhaps you deployed a bundle to JBOSS_HOME or some other directory and the manageRootDirectory was not set to false in your recipe?
Hi Larry, The manageRootDirectory is missing in the recipe but i am not sure if I've deployed the bundle to JBOSS_HOME so for being sure please let me re-test the bug and come back to you. Armine H.
hi Larry, sorry, my answer took a while. The but is not visible anymore, initial case is verified -- yes, the case with bin removal must have been something with deploying into jboss home directory - so marking bug as verified. Thanks, Armine H