IBM Mainframe Discussion List <[email protected]> wrote on 01/08/2009 09:58:16 AM:
> Update. We experienced another 'hang' lasting 2 minutes. LOGREC indicated > SFTABN (S0222) on the same low priority batch job before both hangs. The > programmer indicated the job was doing many inserts/updates to various DB2 > tables before it was canceled (this was on a test DB2 system). > > We opened an incident with IBM - they indicated a DB2 roll back could not > have caused the hang of the Console address space. > > We are still persuing with IBM. > > Are the any z/OS 1.9 uniprocessor configs out there? Have any default > settings changed in 1.9 that I should look at? I am reviewing a uniprocessor > white paper from Dec 2006. Any other ideas?? You might consider writing a simple program to detect a 'hang'. For example, loop issuing an STIMER WAIT for 15 seconds. Do a STCK before the STIMER. When you wake up from the STIMER, do another STCK. and compute the elaspsed time. If it is more that the interval you are interested in (say, something like a minute), issue an ABEND. Set a SLIP A=SVCD for that abend. Run this program at high priority. Use the TRACE ST command to set up a reasonably large system trace (it can be larger then 1MB per CPU as of z/OS 1.9). Under the assumption that your high priority program will be delayed when the "hang" occurs, this should produce a dump when the "hang" ends, and the system trace in the dump will show you something about what was going on at the tail end of the "hang". Jim Mulder z/OS System Test IBM Corp. Poughkeepsie, NY ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
