System Died - How To Handle Restarts?

Been an "exciting" day so far. You know you're in for it when you walk into the data center, see that the amber light is on *both* systems and no green light. All I know for sure is that the batteries in one UPS drained around 0330 and the other one after 0400 Saturday. Neither system has the QPWRRSTIPL system value set to '1' (Automatic IPL after power restored allowed). Both systems, by the way, are 530's running V5R4.

So, anyway, obviously I pushed the IPL button (gotta get those orders in and the product out, you know). SMAPP had been set (60 minutes), but the system came back in just a few minutes. One of the things that happened was that the jobs on the job scheduler, which should have been run over the weekend, started kicking off. Rather than just do a HLDJOBSCDE *ALL, I think I located the API to put the default scheduler list into a user space (QWCLSCDE) from which I can hold the ones that I want held (some are, more or less, permanently held so I would want to, also, produce a list of what was held). Further digging through our beloved archives and the Info Center led me to the QJOBSCD job, which, I think, is what processes the job scheduler. And that, at least on our system, is running in the QUSRWRK subsystem so I would need to check for an abnormal termination in our startup CL before allowing it to be started. (I'm seeking validation here, if you haven't discerned that already.)

I was, also, wondering what the group's thoughts are (pro and con) about setting the QPWRRSTIPL to '1'. My original thought was that I wanted to manually control when the system came back. This was based upon some history: All previous outages here occurred during regular business hours and, at least once, I manually did a PWRDWNSYS Restart(*No) and brought it back up later. At another place at which I worked we had frequent outages, restarts, followed by another outage so I didn't want to i5 getting jerked around. Also, in this case, I'm not sure when power was restored so auto-restarting with no one here (I had to manually kill a couple of those job scheduler-invoked jobs because of conditions/conflicts that I knew about) would have been just as fraught with peril as it was this morning. Thoughts? How do you handle this situation?

Thanks.


Jerry C. Adams
IBM System i Programmer/Analyst
B&W Wholesale
office: 615-995-7024
email: jerry@xxxxxxxxxxxxxxx