| Author |
Message
|
| PeterPotkay |
 Posted: Wed Nov 12, 2008 5:20 pm Post subject: |
 |
|
Poobah
Joined: 15 May 2001
Posts: 7722
|
Did you run ps -ef | grep mq right before yopu started the upgrade? If yes, I assume it returned no processes, right?
_________________
Peter Potkay
Keep Calm and MQ On |
|
| Back to top |
|
 |
| gbaddeley |
| Posted: Wed Nov 12, 2008 6:12 pm Post subject: |
|
|
Jedi Knight
Joined: 25 Mar 2003
Posts: 2538
Location: Melbourne, Australia
|
| mqjeff wrote: |
| 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. |
And if inetd is being used as the listener for MQ, it needs to be disabled too during the upgrade (edit /etc/services and refresh inetd process), otherwise incoming clients and channels can start an amqcrsta and create shared mem & semaphores.
_________________
Glenn |
|
| Back to top |
|
|
| CShep |
| Posted: Thu Nov 13, 2008 1:09 am Post subject: |
|
|
Novice
Joined: 20 Oct 2004
Posts: 22
Location: London, England
|
@Peter - yes, I checked for mq processes and there were none.
@gbaddely - no, we don't use inetd. The listener is defined in the queue manager.
Thanks again.
_________________
A friend in need is a pain in the @ss |
|
| Back to top |
|
|
| CShep |
| Posted: Thu Nov 13, 2008 1:34 pm Post subject: |
|
|
Novice
Joined: 20 Oct 2004
Posts: 22
Location: London, England
|
Hi all (again),
I'm sure you had hoped that this would be the end of the matter but ....
I had been running the pkgadd from /tmp/mqm/ but was asked by the SA's to run from a non /tmp directory. I did this tonight from /var/tmp/ and the install started 
I'm not sure of the reasoning behind using /var as opposed to /tmp and will ask for this tomorrow (and post the reasoning here).
However, I get a failed message at conclusion,-
pkgadd: ERROR: postinstall script did not complete successfully
Installation of <mqm-06-00-02-05> partially failed.
Above this are a whole load of these kinds of messages,-
ERROR: attribute verification of </opt/mqm/ssl/securityguide.ref.htm> failed
pathname does not exist
As this is Solaris 10 I'm wondering if this is related to Solaris Zones?
I used "pkgadd -d" but I've looked on Hursley blog and it advises to use -G for the global zone
Any ideas?
_________________
A friend in need is a pain in the @ss |
|
| Back to top |
|
|
| fjb_saper |
| Posted: Thu Nov 13, 2008 1:40 pm Post subject: |
|
|
Grand High Poobah
Joined: 18 Nov 2003
Posts: 20756
Location: LI,NY
|
So what happens if you use pkgadd -G ? 
_________________
MQ & Broker admin |
|
| Back to top |
|
|
| CShep |
| Posted: Thu Nov 13, 2008 3:20 pm Post subject: |
|
|
Novice
Joined: 20 Oct 2004
Posts: 22
Location: London, England
|
Good question. It was looking very good until postinstall - results below,-
## Executing postinstall script.
removef: ERROR: bad entry read in contents file
pathname=/var/sadm/pkg/mqm-06-00-02-05/save/opt/mqm/java/sdk/jre/lib/sparc/libjvm.so
problem=no link source specified
pkgadd: ERROR: postinstall script did not complete successfully
Installation of <mqm-06-00-02-05> partially failed.
I can find some info on "problem=no link source specified" but not in this specific context.
_________________
A friend in need is a pain in the @ss |
|
| Back to top |
|
|
| gbaddeley |
| Posted: Thu Nov 13, 2008 8:14 pm Post subject: |
|
|
Jedi Knight
Joined: 25 Mar 2003
Posts: 2538
Location: Melbourne, Australia
|
This error is usually due to permissions on the install working directory. It appears pkgadd su's to mqm to run some scripts. The usual work around is to chmod 777 the install working directory.
_________________
Glenn |
|
| Back to top |
|
|
| CShep |
| Posted: Wed Nov 19, 2008 2:51 am Post subject: |
|
|
Novice
Joined: 20 Oct 2004
Posts: 22
Location: London, England
|
For anybody still interested, here is the workaround I adopted,-
- converted the datastream package to a filesystem package
- removed the old gsk* pkg's and installed the new gsk pkg's from the filesystem (this part of the process wasn't actualy causing any problems previously, so not entirely essential ie. could have been done from the datastream pkg)
- edited two files in the filesystem pkg,
../install/checkinstall was edited to not check '/opt/mqm/bin/amqiclen -c -i -v'
../pkgmap was edited with revised chksum and size of new checkinstall file
- confirm all IPC resources were clear, and ran the pkgadd process
I've now followed this process on 3 servers, and it's been successful every time.
Thanks, one and all, for your advice and help.
_________________
A friend in need is a pain in the @ss |
|
| Back to top |
|
|
|
|
