| Author |
Message
|
| CShep |
 Posted: Sat Nov 08, 2008 6:00 pm Post subject: 6.0.2.5 on Solaris SPARC and dspmqver |
 |
|
Novice
Joined: 20 Oct 2004
Posts: 22
Location: London, England
|
Hi,
I've just applied fixpack 6.0.2.5 to a pair of T2000 Solaris 10 boxes, which were previously at 6.0.2.3. I followed the instrucions (removed old gsk, installed 'all') and received successful messages throughout on both servers. However, when I run dspmqver it returns 6.0.2.3
Two questions, if anybody can help,
1. do they need bouncing? I've only done a few Solaris upgrades previously and never had to before (I'm 99% sure)?
2. is there anything stupid I've missed?
Many thanks in advance,
CShep
_________________
A friend in need is a pain in the @ss |
|
| Back to top |
|
 |
| PeterPotkay |
| Posted: Sun Nov 09, 2008 5:45 am Post subject: |
|
|
Poobah
Joined: 15 May 2001
Posts: 7722
|
you do not need to bounce a UNIX server to upgrade MQ. What happens if you display all the MQ packages on the system? Are you sure the 6025 ones successfully installed?
_________________
Peter Potkay
Keep Calm and MQ On |
|
| Back to top |
|
|
| CShep |
| Posted: Sun Nov 09, 2008 7:35 am Post subject: |
|
|
Novice
Joined: 20 Oct 2004
Posts: 22
Location: London, England
|
Peter,
Thanks for the reply.
I had a think about it (it was late when I was doing this), and ran through the process again this morning. The gsk* packages installed ok, but the MQ update failed due to "Detected System V IPC Resources that are in use".
I had this previously on v5.3 and ran this command to clear it up,-
ipcs -a | grep mqm | awk '{printf( "-%s %s ", $1, $2 )}' | xargs ipcrm
It doesn't seem to be working though - I still get the IPC error above.
Any ideas?
many thanks for your help.
_________________
A friend in need is a pain in the @ss |
|
| Back to top |
|
|
| fjb_saper |
| Posted: Sun Nov 09, 2008 2:59 pm Post subject: |
|
|
Grand High Poobah
Joined: 18 Nov 2003
Posts: 20756
Location: LI,NY
|
Make sure all qmgrs on the box are stopped. Run /opt/mqm/bin/amqiclen -? with the relevant flags to clear up ipc memory. Remember you should always clear the ipc memory with the same user that is holding it.
Have fun 
_________________
MQ & Broker admin |
|
| Back to top |
|
|
| Jinny |
| Posted: Sun Nov 09, 2008 6:43 pm Post subject: |
|
|
Novice
Joined: 30 Jul 2008
Posts: 16
|
|
| Back to top |
|
|
| CShep |
| Posted: Mon Nov 10, 2008 3:42 am Post subject: |
|
|
Novice
Joined: 20 Oct 2004
Posts: 22
Location: London, England
|
@Jinny - many thanks. /var/mqm/qmgrs/ contains the @SYSTEM entry and a symbolic link to the SAN storage for the qmgr. I have performed an upgrade on 5.3 before with symbolic links in place and it's worked - so I'm fairly comfortable ruling that out. But, I removed the symbolic links and ran the process anyway - still fails 
@fjb_saper. Sorry, in my haste to post the problem (and resolve) I had ommitted saying I'd run the amqiclen already. I wasn't aware about the requirement to run as the user holidng the resources - previously I have run as root. I have run "/opt/mqm/bin/amqiclen -x -i" as mqm and root. When I run "/opt/mqm/bin/amqiclen -c -i" I receive the output shown below,-
SEM: 12 0 0 23549 /var/mqm/mqs.ini
SEM: 12 1 0 0 /var/mqm/mqs.ini
The queue manager is running again now - the app team are in the middle of UAT. I hope to get an agreed downtime later to retry, and I wonder if anybody has any more suggestions (please)?
Thanks to all so far,
_________________
A friend in need is a pain in the @ss |
|
| Back to top |
|
|
| mqjeff |
| Posted: Mon Nov 10, 2008 4:33 am Post subject: |
|
|
Grand Master
Joined: 25 Jun 2008
Posts: 17447
|
| The queue manager needs to be completely stopped when you upgrade, and you also need to stop all applications that might be connected to the queue manager... This includes any monitoring applications, listeners, channel agents, etc. etc. etc.... It seems like you missed one somewhere. |
|
| Back to top |
|
|
| mvic |
| Posted: Mon Nov 10, 2008 12:07 pm Post subject: |
|
|
Jedi
Joined: 09 Mar 2004
Posts: 2080
|
| fjb_saper wrote: |
| Make sure all qmgrs on the box are stopped. Run /opt/mqm/bin/amqiclen -? with the relevant flags to clear up ipc memory. Remember you should always clear the ipc memory with the same user that is holding it. |
If IPC shared mem, semaphores etc. are in use and this prevents
an install from proceeding or a queue manager from starting, then
the right thing to do is end the applications using the IPC resources,
not to remove the resources directly. The installer and/or strmqm
processing will take care of removing those that MQ needs to be
removed.
It's possible to cause MQ data corruption by removing semaphores
since these are used to serialise access to MQ data in shared memory
blocks and files.
Therefore please avoid using amqiclen and ipcrm against MQ's IPC
shared resources unless directed by IBM support. |
|
| Back to top |
|
|
| CShep |
| Posted: Wed Nov 12, 2008 12:32 pm Post subject: |
|
|
Novice
Joined: 20 Oct 2004
Posts: 22
Location: London, England
|
I've got some downtime (finally) and have retried. Managed to free all IPC Resources, but still get the same result ...
I have disabled monitoring (qflex) so to the best of my knowledge there is nothing using any MQ resources. Below is a screen dump from my session,-
root@ff1d3942app# pkgadd -d ./mqm-U815659.img
The following packages are available:
1 gsk7bas Certificate and SSL Base Runtime (gsk7bas)
(sparc) 7.0.4.17
2 gsk7bas64 Certificate and SSL Base Runtime (gsk7bas64)
(sparcv9) 7.0.4.17
3 mqm-06-00-02-05 WebSphere MQ for Sun Solaris - U815659
(sparcv9) 6.0.2.5
Select package(s) you wish to process (or 'all' to process
all packages). (default: all) [?,??,q]: all
Processing package instance <mqm-06-00-02-05> from </tmp/mqm/mqm-U815659.img>
WebSphere MQ for Sun Solaris - U815659(sparcv9) 6.0.2.5
Licensed Materials - Property of IBM
5724-H72
(C) Copyright IBM Corporation 1994, 2005 All rights reserved.
US Government Users Restricted Rights - Use, duplication or disclosure
restricted by GSA ADP Schedule Contract with IBM Corp.
## Executing checkinstall script.
ERROR: Detected System V IPC Resources that are in use.
Stop all WebSphere MQ processes before installing this
package.
checkinstall script suspends
Installation of <mqm-06-00-02-05> was suspended (administration).
No changes were made to the system.
There are 2 more packages to be installed.
Do you want to continue with package installation? [y,n,?] n
2 packages were not processed!
root@ff1d3942app# ipcs -a
IPC status from <running system> as of Wed Nov 12 21:28:00 MET 2008
T ID KEY MODE OWNER GROUP CREATOR CGROUP CBYTES QNUM QBYTES LSPID LRPID STIME RTIME CTIME
Message Queues:
T ID KEY MODE OWNER GROUP CREATOR CGROUP NATTCH SEGSZ CPID LPID ATIME DTIME CTIME
Shared Memory:
T ID KEY MODE OWNER GROUP CREATOR CGROUP NSEMS OTIME CTIME
Semaphores:
s 18 0x78760a --ra-ra-ra- root other root other 2 21:22:58 21:22:58
root@ff1d3942app#
If I wasn't already bald I'd be pulling my hair out!!
Is it time to speak to IBM?
_________________
A friend in need is a pain in the @ss |
|
| Back to top |
|
|
| mvic |
| Posted: Wed Nov 12, 2008 1:19 pm Post subject: |
|
|
Jedi
Joined: 09 Mar 2004
Posts: 2080
|
What are the outputs from
| Code: |
# become root
/opt/mqm/bin/amqiclen -cv
echo $?
/opt/mqm/bin/amqiclen -civ
echo $?
ps -ef | grep mq
ipcs -a |
immediately after such an observed failure? |
|
| Back to top |
|
|
| CShep |
| Posted: Wed Nov 12, 2008 1:28 pm Post subject: |
|
|
Novice
Joined: 20 Oct 2004
Posts: 22
Location: London, England
|
Hi mvic,
Output as directed,-
SEM: 18 0 0 6766 /var/mqm/mqs.ini
SEM: 18 1 0 0 /var/mqm/mqs.ini
SHM: 134217795 -1 -1 s/T24FM01T/@qmpersist/shmem/shm.0 23:11:39 10/11/2008
SHM: 134217815 -1 -1 s/T24FM01T/@qmpersist/shmem/shm.1 23:11:39 10/11/2008
SHM: 134217818 -1 -1 s/T24FM01T/@qmpersist/shmem/shm.2 23:11:39 10/11/2008
0
SEM: 18 0 0 6992 /var/mqm/mqs.ini
SEM: 18 1 0 0 /var/mqm/mqs.ini
0
root 6994 6991 0 22:27:09 pts/3 0:00 grep mq
root 6991 6664 0 22:27:09 pts/3 0:00 /bin/sh ./mqout.sh
IPC status from <running system> as of Wed Nov 12 22:27:08 MET 2008
T ID KEY MODE OWNER GROUP CREATOR CGROUP CBYTES QNUM QBYTES LSPID LRPID STIME RTIME CTIME
Message Queues:
T ID KEY MODE OWNER GROUP CREATOR CGROUP NATTCH SEGSZ CPID LPID ATIME DTIME CTIME
Shared Memory:
T ID KEY MODE OWNER GROUP CREATOR CGROUP NSEMS OTIME CTIME
Semaphores:
s 18 0x78760a --ra-ra-ra- root other root other 2 22:27:08 21:22:58
Thanks,
_________________
A friend in need is a pain in the @ss |
|
| Back to top |
|
|
| mvic |
| Posted: Wed Nov 12, 2008 2:10 pm Post subject: |
|
|
Jedi
Joined: 09 Mar 2004
Posts: 2080
|
Thank you. I'm a little out of my comfort zone on this topic. But...
On a Solaris machine here I see the sem id labelled /var/mqm/mqs.ini
is --ra-ra-ra- mqm mqm mqm mqm. Yours seems to be root/other
which is a difference that looks significant. But I don't know how significant.
I also see from the outputs that the $? is 0 in both runs of amqiclen.
My expectation is that the install should proceed successfully in such a
case.
If you should try the install again and is STILL fails, please collect the
output from those commands immediately following the failure. If $?
is 0 from those amqiclen commands then I can't understand why the
install fails.
Are applications, listeners etc. trying to start (even fleetingly) leading
to semaphores being in use at the time of the install... just a guess.
I just imagine it's possible for cron or something to be attempting to
start things that try to connect to a qmgr.
And yes I think it makes sense to open a PMR at the same time, if you
can find some time to do so. |
|
| Back to top |
|
|
| CShep |
| Posted: Wed Nov 12, 2008 2:44 pm Post subject: |
|
|
Novice
Joined: 20 Oct 2004
Posts: 22
Location: London, England
|
Thanks - I have lost sight of my comfort zone from here  . I'll check the difference in the sem id values with the SA's tomorrow. I've rerun as requested, with the same result...
SEM: 18 0 0 6993 /var/mqm/mqs.ini
SEM: 18 1 0 0 /var/mqm/mqs.ini
SHM: 134217795 -1 -1 s/T24FM01T/@qmpersist/shmem/shm.0 23:11:39 10/11/2008
SHM: 134217815 -1 -1 s/T24FM01T/@qmpersist/shmem/shm.1 23:11:39 10/11/2008
SHM: 134217818 -1 -1 s/T24FM01T/@qmpersist/shmem/shm.2 23:11:39 10/11/2008
0
SEM: 18 0 0 8477 /var/mqm/mqs.ini
SEM: 18 1 0 0 /var/mqm/mqs.ini
0
root 8479 8476 0 23:16:24 pts/3 0:00 grep mq
root 8476 6664 0 23:16:24 pts/3 0:00 /bin/sh ./mqout.sh
IPC status from <running system> as of Wed Nov 12 23:16:24 MET 2008
T ID KEY MODE OWNER GROUP CREATOR CGROUP CBYTES QNUM QBYTES LSPID LRPID STIME RTIME CTIME
Message Queues:
T ID KEY MODE OWNER GROUP CREATOR CGROUP NATTCH SEGSZ CPID LPID ATIME DTIME CTIME
Shared Memory:
T ID KEY MODE OWNER GROUP CREATOR CGROUP NSEMS OTIME CTIME
Semaphores:
s 18 0x78760a --ra-ra-ra- root other root other 2 23:16:24 21:22:58
I did notice this time (and upon checking the earlier run also) that I didn't capture the standard error output, which seems very shabby of me. My apologies. Here it is,
No such file(/var/mqm/qmgrs/@SYSTEM/shmem/shm.0)
No such file(/var/mqm/qmgrs/@SYSTEM/@ipcc/shmem/shm.0)
No such file(/var/mqm/qmgrs/@SYSTEM/@qmpersist/shmem/shm.0)
No such file(/var/mqm/qmgrs/T24FM01T/shmem/shm.0)
No such file(/var/mqm/qmgrs/T24FM01T/@ipcc/shmem/shm.0)
No such file(/var/mqm/qmgrs/@SYSTEM/@ipcc/shmem/shm.0)
No such file(/var/mqm/qmgrs/T24FM01T/@ipcc/shmem/shm.0)
Hope this helps, if not please let me know and I'll raise the PMR.
Regards,
_________________
A friend in need is a pain in the @ss |
|
| Back to top |
|
|
| mvic |
| Posted: Wed Nov 12, 2008 2:58 pm Post subject: |
|
|
Jedi
Joined: 09 Mar 2004
Posts: 2080
|
If those outputs (specifically the 0 return codes from amqiclen)
were collected immediately after seeing
| Quote: |
ERROR: Detected System V IPC Resources that are in use.
Stop all WebSphere MQ processes before installing this
package. |
then I can't understand what's wrong, and yes if possible I'd say raise
the PMR.
Did you check whether any MQ apps etc. are trying to start - eg. from
cron jobs? |
|
| Back to top |
|
|
| CShep |
| Posted: Wed Nov 12, 2008 3:20 pm Post subject: |
|
|
Novice
Joined: 20 Oct 2004
Posts: 22
Location: London, England
|
Sorry, yes I checked for any cron entries but there was nothing I could see. MQ monitoring is disabled. It's a stand-alone server so no clustering service / HA to worry about. Our central monitoring team uses Patrol, but do not monitor MQ on this server. I'll raise a PMR in the morning.
Thanks mvic, and the others who shared their knowledge.
Regards,
_________________
A friend in need is a pain in the @ss |
|
| Back to top |
|
|
|
|
