E9.2 iSeries Restart - Confirm Kernels Came Down Cleanly Before a Restart

[czarcasm]

Looking for some ideas.... we have had some intermittent issues in our automated restart process. Each night our services restart and this all happens via a scripted series of commands that run on a schedule. The issue we have found is that some of the various kernels have not ended properly and remain in a MTXW status after shutdown or they are simply orphaned in the subsystem. My simple solution would be to just write a small program to clear the log files in the JDE920 directory (which should be emptied at every restart) and if we have any straggling files, then we need to signal the support team that something did not work correctly and requires investigation. The challenge I am finding is in writing that "simple" program as the way our team has attempted to make this happen through dream writer but doing this appears to actually override the locks and purge the directory anyway (so this wont work). Any ideas or other folks that have found a way to signal on a botched restart for an iSeries environment?

 

[markdcci]

Czarcasm,
If the goal is to find out which jobs remain in a MTXW status, rather than simply try to clear logs and count on a lock to prevent it could you write a CL program to capture all jobs in a MTXW state in that subsystem after the attempted shutdown?

By the way, I would have thought there would be a lock on those log files if the jobs were still in a MTXW state, so I'm a little surprised there wasn't. But a CL program would be a more direct method.

Alterntatively, if the goal is to get your support team to look into the problem, it sounds like you already have orphaned items in the subsystem - is that not enough for them? You likely have other processes running that are causing the MTXW state. I would suggest they have someone monitor the restart one night in the hopes that the issue is a little more obvious. And/or look at the iSeries event log.

One question: We usually did a weekly restart of services on the various AS/400/iSeries/System i servers I have managed and as time progressed I began to think that was too frequent. Nightly seems very frequent. Did you have an issue that caused you to restart nightly?

 

[glen7x]

Do you have replication running? I had issues with replicated at a site. Do you also have kernel recycling around the times of restarts?