> SYSMDUMPs can be written to SYSOUT, and then, for the ones you want >to look at with IPCS, extracted to a sequential data set. pool. You can >use >a spool file management program (for example, SDSF or EJES) to do the >extraction.
While that is true, I do not recommend doing that. Most z/OS installations have ISV products that read complete jobs from JES2 spool and archive them somewhere else. The products I know only look at the number of lines in the job for their decision to get it from spool. They totally disregard how *long* these lines are. A sysmdump in JES spool doesn't have all that many lines, but those lines are not 132 byte long. They are 4160 byte long. We got badly bitten by a thoughtless sysmdump dd sysout=* in a (USS) application that kept abending with a sysmdump every time a new address space got started (and that was several per minute!) It lead to a 'spool shortage' in our archival product followed by a severe JES2 spool shortage. I had a hell of a time figuring out why the archival product was in a spool shortage and did not reject the extremely large jobs (byte-wise) in the first place. As a consequence, I removed all sysmdump dd sysout=* cards in all STCs. I told the owners of those STCs that they would have to allocate sysmdump data sets for their STCs and figure out how to save the one they had gotten before STC restart overwrote it. Admittedly the setup we had was dumb. The STC sysmdumps should never have gone to sysout=*, they should have gone to a class that won't get archived. I know that my current installation has a similarly dumb setup, this time for batch jobs. They all write those 3700 page sysudumps and in addition the ceedumps to sysout=* and have it archived for however long the jobs is required to be archived. Even with linesize 132 that is a lot of DASD and later tape space. Never mind that there is usually a TDUMP accompanying the ceedump. In my current installation I appear to be the only one who can read a SYSMDUMP using IPCS, and I had exactly one person ask me how I did it and if he can use IPCS, too. Regards, Barbara ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
