NW6.5 SP8
HP DL380 G4 - 3GB RAM
GW8 - MTA, POA, GWIA, WEBACC - NMessenger - ZFD4 - iprint -BKUPEXEC

Sorry for the long post but trying to cover all bases here.

We've just started getting errors from Backup Exec (v9.2 build 1401) when trying to backup eDirectory "Could not connect to xxxxx.Novell Directory:. The login credentials may be incorrect. All selections for this device will be skipped."

Looking into this we found a post about memory errors in java. We checked and couldn't see any errors but instead unload BE then ran BESTOP before reloading with BESTART.

Backup that night worked fine, but the same problem returned last night, so we decided to do a DSREPAIR this morning - unattended full repair, the repair aborted with these errors in the log:


ERROR: Could not track parent of subordinate count. Error: -150

ERROR: updating object 801C Error: -150

WARNING: The current dib set name is set to: NDT
NOTICE: Successfully set current dib set to: nds

Repair process aborted

and these cache memory allocator errors on the console screen:

28-04-2010 11:13:58 am: SERVER-5.70-0 [nmID=6001D]
Cache memory allocator out of available memory.

28-04-2010 11:13:58 am: SERVER-5.70-0 [nmID=2000A]
Short term memory allocator is out of memory.
3 attempts to get more memory failed.
request size in bytes 16777220 from Module DSREPAIR.NLM

So pretty certain the errors we're getting are due to memory and downloaded seg.nlm (segstats.txt attached). Seems to me that both server.nlm and ncpl.nlm are consuming a hell of a lot of RAM, and we also have nearly 400MB of fragmented RAM.

The server has been running since the 23rd Dec 2009 which was when we rebooted after upgrading to GW8 and the following patches (zip file names downloaded from Novell support), prior to this the server had been running fine (with GW7) for a couple of years and with GW6 for a couple prior to that:


Any ideas? Other than rebooting the server and waiting another 4 months?