| |
|
RSS Feed - WebSphere MQ Support
|
RSS Feed - Message Broker Support
|
  |
|
| MQ Z/OS Page Set Allocation |
« View previous topic :: View next topic » |
| Author |
Message
|
| BMQUser |
 Posted: Wed Feb 14, 2007 6:31 am Post subject: MQ Z/OS Page Set Allocation |
 |
|
Newbie
Joined: 15 Dec 2006
Posts: 3
|
We are using IBM Websphere MQ V5.3 on Z/OS and have hit a problem where volume testing results in page set #3 running out of space. I allocated a new page set #3 using CSQ530.SCSQPROC(CSQ4PAGE) JCL and changed the size of this dataset. The JCL ran without error.
Now when I start MQ (e.g. CSQ1 START QMGR) there is the following error:
CSQJ113E +CSQ1 RBA 000000000000 NOT IN ANY ACTIVE OR 796
ARCHIVE LOG DATA SET, CONNECTION-ID=CSQ1 THREAD-XREF=003.RCRSC 02
Looking at the documentation it seems the LOG cannot find the RBA required.
Do I need to delete the log files since I deleted / redefined one of the page data sets?
Thanks in advance |
|
| Back to top |
|
 |
| ChrisW |
| Posted: Wed Feb 14, 2007 6:58 am Post subject: |
|
|
Voyager
Joined: 20 May 2001
Posts: 78
Location: UK
|
MQ needs the RBA's in the page sets to be the same as that in the log, and startup will fail if there is any inconsistency. You need to copy the contents of the old page set to the new one - see z/OS Systems Administration Guide.
I hope you have the old one still! Otherwise you will have to look at doing a cold start etc. Again the info is in the SAG. |
|
| Back to top |
|
|
| bruce2359 |
| Posted: Wed Feb 14, 2007 7:02 am Post subject: |
|
|
Guest
|
Since this appears to be a test queue manager, recreate the entire mq instance: rerun csq4page for all pagesets AND run csq4bsds to recreate bsds and register new logs.
The message should be a warning to you in a production mq to make frequent pageset backups, AND maintain all your logs (archive and active) so that if a pageset becomes damaged, they will be available at restart. Read the System Admin Guide for z/os, especially the section on Recovery and Restart. Briefly summarized: no logs no mq. |
|
| Back to top |
|
|
| BMQUser |
| Posted: Wed Feb 14, 2007 7:50 am Post subject: |
|
|
Newbie
Joined: 15 Dec 2006
Posts: 3
|
 Thanks to all! I reallocated all pageset, BSDS, and log files and now everything is cool! |
|
| Back to top |
|
|
| Mr Butcher |
| Posted: Wed Feb 14, 2007 10:46 pm Post subject: |
|
|
Padawan
Joined: 23 May 2005
Posts: 1716
|
A suggestion.
because this was a test queuemanager you rebuild it from scratch like bruce2359 suggested (which worked).
but what if you have the same problem in production? will you also dump everything and restart from scratch? if so, okay. if not, then you should perform the actions you planed for your production environment in test environment, just to make sure your procedures are correct and working.
so i would perform the procedure ChrisW suggested to enlarge a Page Dataset. As he already mentioned, the procedure to do so is written in the System Administration Guide for z/OS.
This will help you to get prepared for a pageset enlargement in production.
_________________
Regards, Butcher |
|
| Back to top |
|
|
| bruce2359 |
| Posted: Thu Feb 15, 2007 7:11 am Post subject: |
|
|
Guest
|
From the original post, BMQuser created a new pageset 3, but did not copy the contents of the old/full pageset 3 to the new pageset 3.
Then BMQuser re-started the qmgr. On restart, mq compares the high-used rbas of all pagesets with high-used rba in bsds. If they are not equal, qmgr demands that all logs needed to restore pagesets to their consistent state be available. That's what the CSQJ113E message means.
It's working as designed.
I was suggesting that if this happens in production, pageset backups and logs are the only way to recover. In this case, BMQuser demonstrated what happens when a pageset gets destroyed (deleted and redefined) AND all logs are not available - abend. |
|
| Back to top |
|
|
| Mr Butcher |
| Posted: Thu Feb 15, 2007 9:07 am Post subject: |
|
|
Padawan
Joined: 23 May 2005
Posts: 1716
|
all i wanted is to recommend that he should practice the enlage-pageset procedure if he wants to use it in production one day. why do you feel offended?
so ... stay cool man! 
_________________
Regards, Butcher |
|
| Back to top |
|
|
| bruce2359 |
| Posted: Thu Feb 15, 2007 9:38 am Post subject: |
|
|
Guest
|
 Not offended in any way.
I am a true-believer that test, q/a and production, should be managed exactly the same way. In this instance, BMQuser seemed to be playing/testing - without benefit of system admin guide or msgs manual.
The qmgrs need (requirement, demand) for bsds, pagesets and logs for maintaining consistency is well documented in the system admin guide. Working with (managing, caring for) them is equally well documented.
My point to BMQuser was that damaged pagesets will be recovered from logs at restart, but only if you have logs. No logs; no qmgr. The abend message BMQuser received said exactly what the problem was: rba mismatch on new pageset, but no log(s) available to recover.
If this was production and BMQuser had no logs, the remaining options (conditional restart, for example) are few and definitely not pretty.
Again, not offended. |
|
| Back to top |
|
|
|
|
|
| |
|
Page 1 of 1 |
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
|
|
|
