Gerhard Postpischil wrote: >The only error, but not RC, I remember vividly is the case where a PDS didn't >have enough directory blocks to add another member, did the add to the >directory anyway, causing the last name to be dropped from the directory. This >was particularly annoying when doing a sysgen, as the ruined data set (usually >LINKLIB or SVCLIB) became useless, and the sysgen had to be restarted from >scratch, with larger allocations.
Urrgggghhh! Lots of [exciting] pain! ;-) At least I'm not alone with that pain. [1] ;-D Backup + checkpointing plans won't help you here, because you BUILD things up from scratch... >Back then we had a practice of taking values from the storage estimates book, >and increasing them at least three-fold. And still run out of space? I usually used values of 10 times of everything (primary, secondary, dir space, etc). Other reason is that those estimates are sometimes based on a specific device and specific defaults, not very helpful, mind you... After all setup is completed, I just trimmed excess space just to keep my storage admin guy happy. ;-) Groete / Greetings Elardus Engelbrecht [1] - I was at one stage unfortunate not to able to claim overtime worked just because of that disaster where I had to redo everything after hours (no SMP/E), because the project was behind schedule and I have a *grumpy* boss. :-/ Fun, big fun... ;-) ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
