View previous topic :: View next topic
|
Author |
Message |
Paul Hacker
New User
Joined: 02 Oct 2008 Posts: 15 Location: Tyler, texas
|
|
|
|
Here is the problem:
Sometimes, rarely, we have a system lockup due to applications (on MVS, ZOS, CICS) going bonkers.
I'm being charged with, if this happens, to find the offending program and cancel it.
Now using MAINVIEW (BMC product) and the SDSF Log, is there a logical way to find such lockups?
Our systems programmer is away and this is one of the task laid on me!
Thanks for any pointers! |
|
Back to top |
|
|
dick scherrer
Moderator Emeritus
Joined: 23 Nov 2006 Posts: 19244 Location: Inside the Matrix
|
|
|
|
Hello,
How much of the system "locks up"? Do most jobs continue running? Does one particular cics have the problem? |
|
Back to top |
|
|
Robert Sample
Global Moderator
Joined: 06 Jun 2008 Posts: 8696 Location: Dubuque, Iowa, USA
|
|
|
|
I go into Mainview, issue the command JCPUR to look at address spaces by CPU utilization -- if any are going "bonkers" that's most likely where they will show up. |
|
Back to top |
|
|
Paul Hacker
New User
Joined: 02 Oct 2008 Posts: 15 Location: Tyler, texas
|
|
|
|
Dick,
The whole system bogs down and locks up! It's a very rare occurance, sometimes Printways causes it and other times the HBO software has a 'bug' and that causes it.
Robert,
Thanks, I'll try that.
Paul |
|
Back to top |
|
|
Paul Hacker
New User
Joined: 02 Oct 2008 Posts: 15 Location: Tyler, texas
|
|
|
|
How to find if a COBOL program is looping, thru Mainview
If you want to know if a program is looping (on mainframe) and if you have Mainview for MVS, then
1. Go to JCPUR which shows realtime job CPU utilization
2. Place the cursor on the job you are interested in and hit enter
3. It'll take you to EZMJOBR window
4. Under SYSPROG Service, you'll find "MVScope CPU Tracing". Place your cursor there and hit enter
5. Then place your cursor on "Begin CPU Trace" and hit enter. It'll start CPU tracing.
6. Once it finishes, see if it shows any one particular statement. If so, it could mean, the program was in a loop during the trace period.
7. Get the Offset info, go to the compile listing and search for the Offset under HEXLOC (you may not find the exact offset, but the one just before this value is good enough). Get the line number and get the statement that was in loop. |
|
Back to top |
|
|
dick scherrer
Moderator Emeritus
Joined: 23 Nov 2006 Posts: 19244 Location: Inside the Matrix
|
|
|
|
Hello,
Thanks for the info
Do keep in mind though - A single cobol application program, in the tightest loop, should not "shut down"/"lock up" a mainframe. . .
If it happens to be a cics program, it should timeout rather quickly. . . |
|
Back to top |
|
|
enrico-sorichetti
Superior Member
Joined: 14 Mar 2007 Posts: 10873 Location: italy
|
|
|
|
posting the same question on multiple forums is considered bad netiquette!
the people watching the mvs related forums are mostly the same
and is annoying to see the same question asked more than once
posting twice will not get You double answers in the same time or the same number of answers in half time |
|
Back to top |
|
|
Paul Hacker
New User
Joined: 02 Oct 2008 Posts: 15 Location: Tyler, texas
|
|
|
|
Sorry about that enrico. Didn't know many people view both forums.
Now the last time we had a lockup it was cause by an assembly languguage program called through a COBOL program. The vendors fixed that with a patch they sent us after the excitement!
Thanks for the help guys.
Paul |
|
Back to top |
|
|
|