| Author |
Message
|
| flaufer |
 Posted: Tue Sep 19, 2006 12:58 am Post subject: repository frequently switched from FR to nonFR. |
 |
|
Acolyte
Joined: 08 Dec 2004
Posts: 59
|
Folks,
some strange error message I'm seeing here in my logs of a PR queue manager in Cluster PROD. Parallel to that I'm receiving 2082 when trying to access alias queues pointing to cluster queues or 2085 when trying to access cluster queues (APPS accessing via FR queue manager). This is MQ5.3CSD07 on AIX5.2. Both qmgrs hosted on one single box.
Unfortunaly I've lost the ERROR LOGS from the FR but I remember the only thing I can see was that some colleague tried to issue a "REFRESH CLUSTER" command around the time we had the trouble.
Logs from PR:
09/18/06 12:39:51
AMQ9431: Remote queue manager no longer hosts a repository for cluster
EXPLANATION:
The local queue manager has received a message from remote queue manager QMID FRANKFURT_2004-06-07_20.21.57 indicating that it no longer hosts a repository for cluster PROD. CLUSSDR channel TO.FRANKFURT.PROD is altered so that it can no longer be used to access queue manager FRANKFURT_2004-06-07_20.21.57 within cluster PROD. If the local queue manager does not host a repository for cluster PROD the relevant subscriptions and publications are remade if possible.
ACTION:
None.
09/18/06 12:39:57
AMQ9457: Repository available, cluster PROD, channel TO.FRANKFURT.PROD, sender FRANKFURT_2004-06-07_20.21.57.
EXPLANATION:
The repository queue manager received a command from another queue manager, whose identifier is FRANKFURT_2004-06-07_20.21.57, reporting that it is again a repository for cluster PROD. The cluster-sender channel TO.FRANKFURT.PROD is changed so that it can be used to access the other queue manager in relation to the the cluster.
ACTION:
None.
This happened 4 times yesterday afternoon. Always with 8 seconds inbetween. We shutdown the queue manager when it happened for the first time but never actually did anything to fix it... and after these some seconds, everything was working fine as before.
I'm a bit puzzled here... also because I did not see anything special going on. No apparent HW error on the box, no FDC files written (the last was from two hours before we restarted the queue manager; it was waying something about possible lack of disk space, but disks have been free always).
Hope that somebody can point me in the right direction here... ;-/
Cheers,
Felix |
|
| Back to top |
|
 |
| jefflowrey |
| Posted: Tue Sep 19, 2006 2:19 am Post subject: |
|
|
Grand Poobah
Joined: 16 Oct 2002
Posts: 19981
|
So, as of right now, your cluster is broken.
Make sure that the cluster channel to and from at least one FR is working on your PR. That is, make sure the CLUSSDR to the FR is running and the CLUSRCVR from the FR is running. If they aren't, start them and see what happens.
Issue a refresh cluster on the PR and see what happens.
_________________
I am *not* the model of the modern major general. |
|
| Back to top |
|
|
| fjb_saper |
| Posted: Tue Sep 19, 2006 2:50 am Post subject: |
|
|
Grand High Poobah
Joined: 18 Nov 2003
Posts: 20756
Location: LI,NY
|
My GUESS would be that they had a problem with the FR named FRANKFURT. Your PR is pointing to that FR. They probably made the FR a PR, issued a refresh cluster repos(yes) and made it a FR again.
As your defined clussdr channel points to FRANKFURT you were not able to resolve any qc outside of your PR's repository for the brief time that FRANKFURT was a PR.
As I said, this is just a guess.
Enjoy 
_________________
MQ & Broker admin |
|
| Back to top |
|
|
| rdubey |
| Posted: Fri Dec 08, 2006 2:10 pm Post subject: |
|
|
Novice
Joined: 11 Jul 2006
Posts: 17
Location: USA
|
but if we see a refresh cluster command in mq logs of a partiall repository queue manager then that means that someone has definitely issued it . is that correct ?
mq never issues this command automatically ? right ?
the reason I am asking is that we had a major issue and nobody accepted of having issued this command so I am trying to research . |
|
| Back to top |
|
|
| fjb_saper |
| Posted: Fri Dec 08, 2006 2:26 pm Post subject: |
|
|
Grand High Poobah
Joined: 18 Nov 2003
Posts: 20756
Location: LI,NY
|
Forget the finger pointing.
If you really need to, secure the qmgr.
For starters use some security exit ...
see BLOCKIP2 or Roger Delacroix's security exit (M Capitalware)...
For seconds kill the command server ...
For thirds mandate ssh on the qmgr box
etc....
Or refuse anybody direct access but allow them use of a monitoring tool like Tivoli or QPasa....
As to fingerpointing,... you know it could just be someone using MS0B or pcf messages?...
Enjoy
_________________
MQ & Broker admin |
|
| Back to top |
|
|
| rdubey |
| Posted: Fri Dec 08, 2006 4:20 pm Post subject: |
|
|
Novice
Joined: 11 Jul 2006
Posts: 17
Location: USA
|
| this is not for finger pointing though but I am suspectiong what happened was something like what you mentioned in earlier post that FR was made a PR then refresh cluster was run since i am seeing messages like " no cluster respoitory for qmanger " many times preceding the refresh cluster command. |
|
| Back to top |
|
|
|
|
