| Author |
Message
|
| bmccarty |
 Posted: Tue Jul 16, 2002 10:30 am Post subject: return code 2071 MQRC_STORAGE_NOT_AVAILABLE |
 |
|
Apprentice
Joined: 18 Dec 2001
Posts: 43
|
Hi,
One of our MQ Java applications experienced a problem where the return code 2071 was returned from two seperate applications opening a connection to two seperate queue managers at nearly the same time.
A trace was not active for either MQ or the Java API so all I have is the error the was returned to the application.
This was a WAS app running on Solaris putting to a queue on the same machine for transpot to another queue manager.
I checked to see if there was available disk space and their appears to be plenty. Also the kernel parameters should be more than adequate, but possibly not since it is in a cluster?
I can't recreate the problem to capture it in a trace, but I am a worried that it might happen again.
If anyone has an insight on why I might be getting a 2071, please let me know.
Thanks!
Brian |
|
| Back to top |
|
 |
| bduncan |
| Posted: Tue Jul 16, 2002 10:39 am Post subject: |
|
|
Padawan
Joined: 11 Apr 2001
Posts: 1554
Location: Silicon Valley
|
The error can occur when you are using syncpoint and the UOW exceeds the amount of available space. Since an auto-rollback will occur when the error occurs, presumably all the disk space will be returned immediately, and when you look at the disk after the fact, you'll see space available. Are your applications using syncpoint?
_________________
Brandon Duncan
IBM Certified MQSeries Specialist
MQSeries.net forum moderator |
|
| Back to top |
|
|
| mrlinux |
| Posted: Tue Jul 16, 2002 11:14 am Post subject: |
|
|
Grand Master
Joined: 14 Feb 2002
Posts: 1261
Location: Detroit,MI USA
|
The error you are getting is related to the max data segment size for a process. FOR HPUX there is a kernel parameter that needs to be increased, Iam assuming there should be something similiar for Solaris.
I have had trouble with this one. One of the issues is when your apps connect to the Qmgr they go through an agent process which will allocate
memory from the data segement . Well the agent process is multithreaded so the agent will allocates buffers for up to 20 process (defualt) before starting a new agent process. So if you are allocating
4mb Buffer for your apps and you have 20 of them it will require the kernel parameter to be set to > 80mb.
_________________
Jeff
IBM Certified Developer MQSeries
IBM Certified Specialist MQSeries
IBM Certified Solutions Expert MQSeries |
|
| Back to top |
|
|
| bmccarty |
| Posted: Tue Jul 16, 2002 1:54 pm Post subject: |
|
|
Apprentice
Joined: 18 Dec 2001
Posts: 43
|
Thanks guys for the help.
I am a little confused as to what parameter is actually the max data segement size as I can't find a parameter with that exact name. Do you mean one of the IPC Shared Memory parameters like SHHMAX, SHMMIN or SHMMNI?
I know MQ pretty well but I am not much of a Solaris admin.
Thanks again!
Brian |
|
| Back to top |
|
|
| mrlinux |
| Posted: Tue Jul 16, 2002 7:53 pm Post subject: |
|
|
Grand Master
Joined: 14 Feb 2002
Posts: 1261
Location: Detroit,MI USA
|
Well I know MQSeries and HPUX very well, Iam not to familiar with SUN Kernel parms, It is not one the shared memory segments.
When a program is run it is allocated a code segement (Where the code loads) and a Data Segment ( Where variables are stored) Stack Segment,
There is a limit that is placed on each process to keep it from eating all of the system memory.
Well the kernel parm for HPUX is maxdseg (I think), I dont have access to a HPUX box right now.
_________________
Jeff
IBM Certified Developer MQSeries
IBM Certified Specialist MQSeries
IBM Certified Solutions Expert MQSeries |
|
| Back to top |
|
|
| mrlinux |
| Posted: Wed Jul 17, 2002 9:24 am Post subject: |
|
|
Grand Master
Joined: 14 Feb 2002
Posts: 1261
Location: Detroit,MI USA
|
The Kernel parm on HPUX is maxdsiz.
_________________
Jeff
IBM Certified Developer MQSeries
IBM Certified Specialist MQSeries
IBM Certified Solutions Expert MQSeries |
|
| Back to top |
|
|
| bmccarty |
| Posted: Wed Jul 17, 2002 12:17 pm Post subject: |
|
|
Apprentice
Joined: 18 Dec 2001
Posts: 43
|
I'm still trying to figure out what the corresponding parameter is on Solaris because I think you may be on to something.
I can't find a maxdsiz or something similar. If anybody has a suggestion, please let me know.
Thanks,
Brian |
|
| Back to top |
|
|
| bmccarty |
| Posted: Mon Jul 22, 2002 7:26 am Post subject: |
|
|
Apprentice
Joined: 18 Dec 2001
Posts: 43
|
I opened a PMR with IBM and it looks like that there is a shared memory leak. I was using MQ 5.2 CSD04 and they built an eFix for me. I have not been able to install it to test it yet because we don't normally apply eFixes, but evidently the APAR is included in CSD05.
Hopefully this will fix our problem.
Thanks for the help.
Brian |
|
| Back to top |
|
|
|
|
