Error Adding Queue via jms if admin killed and re-started

Answer 1

that did the trick.
          Charles Paclat wrote:
          > I think that this is a safe bet. Let mw know how it works out when
          > you change it to use the adminstration server.
          >
          > "Larry Presswood" <[email protected]
          > <mailto:[email protected]>> wrote in message
          > news:[email protected]...
          > Ok upon more digging it uses context to the managed server which
          > hosts the jms server not the
          > admin server. Could that be the issue?
          >
          >
          > Larry Presswood wrote:
          >
          >>
          >>
          >> Charles Paclat wrote:
          >>
          >>> Larry,
          >>>
          >>> So the deal is that any mbean creations against that
          >>> AdminMBeanHome will fail if the adminstration server is not
          >>> present. This is whey steps A and B are failing. Is the
          >>> difference between C and D that the other managed servers are
          >>> not being restarted? Why is a managed server and not the
          >>> adminstration server responsible for creating the queues and
          >>> topics? It seems as though based on your description you have a
          >>> specific reason for doing this.
          >>
          >>
          >> The difference between C and D is that in C only one managed
          >> server is being res-started vs D the whole env admin server and
          >> all managed servers are being re-started.
          >>
          >> The managed server has a stateless ejb which
          >>
          >> A. creates context to Admin Server
          >> B. issues mbean calls. so it does use admin server to make the
          >> calls the bean is simply
          >> deployed in managed server for consitency.
          >> I am currently looking at the stateless bean which talks to the
          >> admin server mbeans to see if I can
          >> see any issues. I had first thought it was a cached context but
          >> it re-creates context each call however it does not issue a close
          >> after every call. I am going to try that although i wouldnt think
          >> that would matter
          >>
          >>>
          >>> Is there anything in the server logs associated with the
          >>> rediscovery and reconnect to the adminstration server when you
          >>> are running scenario C? I am going to cross post this message
          >>> to the JMS newsgroups as I think those folks may be more
          >>> familiar with the specifics of how the lifecycle of JMS is
          >>> interacting with JMX.
          >>
          >>
          >> We do see re-discovery log entries being made as I have said all
          >> managed servers have MSI = true and replication = true and are
          >> running on the same physical box duno if thats an issue??
          >>
          >>
          >>
          >>>
          >>> Charles
          >>>
          >>>
          >>>
          >>> "Larry Presswood" <[email protected]
          >>> <mailto:[email protected]>> wrote in message
          >>> news:[email protected]...
          >>> What I can say right now is here are a couple of different
          >>> sequence of events which have essentially
          >>> the same result in that the queues are not created in the
          >>> config.xml will have to check to see if the mbeans have been
          >>> created havent dont that yet.
          >>>
          >>> Scenario A.
          >>>
          >>> Weblogic Admin Server is brought down with a kill -9 (This
          >>> was done by accident)
          >>> We have three managed servers and one of our servers is
          >>> responsible for adding queues etc.
          >>> It is still up however any add operations fail ie nothing to
          >>> config.xml nothing shows up in console and i cant do jndi
          >>> lookup.
          >>>
          >>>
          >>> Scenario B.
          >>>
          >>> Weblogic Admin server is brought down with a graceful shutdown
          >>> Same result as A.
          >>>
          >>> Scenario C
          >>>
          >>> Weblogic Admin Server is brought down
          >>> Managed server which does mbean operations is brought down
          >>> Start Admin Server
          >>> Start Managed server
          >>>
          >>> Same results as A and B.
          >>>
          >>> Scenario D
          >>>
          >>> All servers brought down and back up
          >>> It works correctly.
          >>>
          >>>
          >>>
          >>>
          >>> Charles Paclat wrote:
          >>>
          >>>>Hi Larry,
          >>>>
          >>>>Based on my knowledge of how all of this is put together and see that might
          >>>>be problems in this area. To help me figure out if this is a bug in 7.0 it
          >>>>would be helpful if you could more fully describe the sequence of starting
          >>>>and stopping the admin and managed servers.
          >>>>
          >>>>Somethings that you can check for are if the new queue is being added to the
          >>>>"config.xml" file and if the various beans are being approriately created
          >>>>despite the fact that the queue is not actually coming into existence. Does
          >>>>the Queue/Topic MBean exist on the adminstration server? Is the "Config"
          >>>>version of that bean coming into existence. The Config MBean has the
          >>>>ObjectName as the one you originally created and but with the string
          >>>>"Config" appended to the object type and the additional field
          >>>>Location=<servername>.
          >>>>
          >>>>You can use the weblogic.Admin tool to check this. Alternatively you can
          >>>>deploy the attached webapplication on each of servers and then accessing the
          >>>>ulr http://<hostandport>/mbeans/mbeans.
          >>>>
          >>>>In the meantime I will do some investigation of my own.
          >>>>
          >>>> Thanks,
          >>>>
          >>>> Charles
          >>>>
          >>>>
          >>>>
          >>>>"Larry Presswood" <[email protected]> wrote in message
          >>>>news:[email protected]...
          >>>>
          >>>>
          >>>>>I am looking at our code but figured I would put a post out to see if
          >>>>>anyone else has
          >>>>>seen this.
          >>>>>
          >>>>>We are using weblogic 7.0 sp2
          >>>>>
          >>>>>We have code which uses mbeans to add a Queue/Topic. The issue is if the
          >>>>>admin server
          >>>>>is killed and then re-started (We do have discover managed servers =
          >>>>>true) then our code does not work. I as of yet cant tell if we are
          >>>>>eating an exception or if one is not being thrown.
          >>>>>
          >>>>>However its weird since even after starting the managed server it still
          >>>>>doesnt work.
          >>>>>
          >>>>>We do have MSI = true and MIS replications = true I dont know if this is
          >>>>>related or
          >>>>>not but could it be a known WLS bug??
          >>>>>
          >>>>>Anyhow I am digging into our code. Please let me know if anyone has
          >>>>>experienced the
          >>>>>same issue.
          >>>>>
          >>>>>
          >>>>>
          >>>>
          >>>>
          >>>>
          >>>>
          >>>
          >>
          >
          [att1.html]