db2dari processes remaning after db2stop/db2start

[Bruce Pullen]

db27.2 fp7 on AIX5.2

I just bounced the database (as a specific db2dari process had reached
user memory limit), and some (orpahned) db2dari processes are still
there.

db2fenc1 42682 1 108 Sep 28 - 14308:10 db2dari (81842)
....
db2fenc1 90546 1 108 Sep 28 - 431:39 db2dari (63356)

Does anyone know if this is normal and, more importantly, whether
they'll still be picked up by my sps?

Any help appreciated.
Thanks
Bruce

[Sean McKeough]

Given the cpu consumption I'd guess they're stuck in an infinite loop in
a stored procedure.

The processes themselves are orphaned...they won't interfere with new
calls (other than keeping libraries loaded, and slowing down your system).

Bruce Pullen wrote:

db27.2 fp7 on AIX5.2

I just bounced the database (as a specific db2dari process had reached
user memory limit), and some (orpahned) db2dari processes are still
there.

db2fenc1 42682 1 108 Sep 28 - 14308:10 db2dari (81842)
...
db2fenc1 90546 1 108 Sep 28 - 431:39 db2dari (63356)

Does anyone know if this is normal and, more importantly, whether
they'll still be picked up by my sps?

Any help appreciated.
Thanks
Bruce

[Bruce Pullen]

Thanks Sean.

Final question - sorry, but are you implying the high CPU usage is due
to an inifinite loop in code (i.e. a code bug)? The reason the DB was
bounced was because the application could no longer load libraries to
(2 specific) db2dari processes as the latter had just about reached
the memory ulimit.

[Sean McKeough]

We see runaway processes like this fairly often, and it 'usually' ends
up being a problem in the sp (couldn't be a bug, right) :-). The only
way to know for sure is to attach a debugger to the process and see what
code it's stuck in.

Come to think of it, another possibility would be recursive
traps...sometimes the stack can get hammered so badly by an sp, that our
trap handler goes off in space.
Bruce Pullen wrote:

Thanks Sean.

Final question - sorry, but are you implying the high CPU usage is due
to an inifinite loop in code (i.e. a code bug)? The reason the DB was
bounced was because the application could no longer load libraries to
(2 specific) db2dari processes as the latter had just about reached
the memory ulimit.