Re: Logical log backup with onbar acting strangely
Posted in 2005
Hi,
analyzing this may involve some larger amount
of data (log files, etc.) ... Do you have a Support
Contract ? Then I recommend to open a case
with IBM Informix Tech Support.
If you do not have a support contract, then I can
take a look at the ON-Bar activity log file,
especially at an excerpt around the time where you
noticed the burst of processes (some bit before
and after the actual burst). You can compress this
and send it directly to me (rather than the whole
of IIUG list). Not sure what I will be able to find out,
but that's what we can try ...
Regards,
Martin
--
Martin Fuerderer
IBM Informix Development Munich, Germany
Information Management
owner-informix-list@iiug.org wrote on 09.09.2005 15:11:29:
> Hello everyone.
> I'm having problem with backing up logical logs for the whole week.
> Platform is IDS 9.40 UC5W2, Solaris 9, Sun V480.
> Storage manager is Legato Networker.
>
> On Monday, I've had an Informix engine hang (possibly due to a bug) but
> couldn't bring it down, neither gracefully nor immediately, and the
oninit> process didn't want to be killed, either.
> So I rebooted the machine and started Informix again and no problem,
every
> piece of data was intact. Great.
>
> Unfortunately, strange things started to happen: very often I was kicked
out
> of the shell with "fork failed" message. This turned out to be because
of
> the number of processes per user limit was hit.
> After checking the process spawn rate it was obvious that logical log
backup
> was forking more than 8 thousand processes (namely, onbar -b -l).
> 15 to 20 seconds after this "multifork" thing the number of processes
start
> to drop and in few seconds they all die away. Logical log that fired
backing
> up gets backed up normally and everything seems fine until next logical
log
> fills up. Same thing with scheduled backups, logical log that gets
backed up
> in the end fires the problem again. And now the spicy detail: all of
this
> doesn't happen always, and I cannot recognize the rule. Lucky thing is
that
> Informix works normally, no problem with it, but still, it sucks to be
> denied access to the system even if it's working OK.
> Since this is a production machine I'm quite nervous about having
problems
> like this so any help would be appreciated.
> Thanks everyone!
>
> P.S. I've checked kernel parameters after reboot, no strange things
there.
sending to informix-list