BUGCHECK: Starting processes during server load crashes with the ff message...

[Naji Zakka]

We are using Tru64 UNIX V4.0F on an alpaserver DS20 with 2 Alpha 500 MHz CPU's. At certain times during the day when we try to start certain processes like "vrestore" or "starting a progress database server", the processes cannot start and we get the following BUGCHECK error message:

%DECthreads bugcheck (version V3.15-397), terminating execution.
% Reason: vpInit: (os/kern) resource shortage (6) nxm_task_init(2,0x14001a000,0
x11ffff9e0)
% Running on OSF1 V4.0 on AlphaServer DS20 500 MHz, 2048Mb; 2 CPUs

Any ideas on this? Is there a tunable parameter that needs to be changed?

Thank you

[Michael Schulte zur Sur]

Hi,

have a look at this:
http://www.ornl.gov/lists/mailing-lists/tru64-unix-managers/2002/10/msg00170.html

greetings,

Michael

[Ralf Puchner]

First, this bugcheck can only happen during process initialization. The most
likely cause is a problem allocating the 2-level scheduling "shared area"
through which the kernel and user-mode schedulers communicate.

So you could hit a vm_mapentries limit. You could hit a process data space limit
(either the user soft limit or the hard system-wide quota). You could run out of
swap space (at least if the system is set to eager swap). One or more of the
RADs (if machine supports this) could be unable to provide the necessary page(s).

Another possiblity is a cpu or memory problem. Have a look into the binary.errlog if machine checks occures.