Upgraded a server from 6.0 SP5 to 6.5 SP4a using the overlay CDs last week. Seemed to work fine. Proliant server has 3.5 Gig RAM.

The server has abended each of the last two nights, although it's not writing it to the abend log for some reason. It is writing it to the CPQIML, so I can give the general abend errors. I have the Abend error handling set to zero so it wouldn't auto reboot, but I haven't seen an abend not update the log before... I have info from the Remote Manager System Health Page as well.

Remote Manager logs - short term memory allocator is out of memory. Attempts to get more memory failed. This occurs at the same time that Compaq's CPQIML shows the abends.


1st day's abend:

Fatal exception (Number 14, Cause Abend: Page Fault Running Process: Server 18 Code executing in module LIBC.NLM v9.0a at offset +C2136h)

2nd day's abend:

RCONAG6 main 1 Code executing in module LIBC.NLM v9.0a at offset +8D2D7h

Any ideas? My guess is these are secondary abends to the short term memory allocator error main error. Where should I look for what the problem is causing the memory allocator error? I have seg.nlm if any stats are needed from it.

The server has 3 TFS volumes. I have changed NSS to 5% because of that. We run ARCserve 11.1, so I adjusted the BTI.CFG cachesize setting to 350000 - NWMKDE.NLM is at the top of the list and shows this amount. An old Powerchute.NLM is actually taking up about 150 Meg, which is odd, but it is running on all of the other servers fine under 6.5 SP4a, so I don't believe it's an incompatibility with 6.5.

On all of the other servers at 6.5 SP4a, I am still getting logical space errors, but they only seem to show up during the first few minutes of the start of the ARCserve backup jobs in the middle of the night, then go away. So, I've considered them cosmetic at this point.

Any help would be appreciated. Thanks.