Re: Strange System Behaviour

From: Andreas Schulze (b79xan_at_gmx.de)
Date: 05/27/04


Date: Thu, 27 May 2004 13:04:37 +0200


"ColinG" <colingresty@yahoo.co.uk> schrieb im Newsbeitrag
news:8fe85d6d.0405270151.3b07063e@posting.google.com...
> We have a system that is exhibiting strange behaviour.
> I don't know a lot about IBM systems, so appologies for the woolly
> identification.
> The machine is an IBM and has been variously described as an RS6000, P
> series and H80.
> It is running AIX 4.3, has the latest microcode updates, latest
> firmware updates.
> It is dual processor, has 4Gb memory, around 400Gb of disk attached to
> SSA Cards.
> Typical user count during the day is 750 telnet users running an ISAM
> based application.
>
> The problem is that three times in the past year the system has slowed
> to a point where all telnet sessions die, and even the console cannot
> be logged into.
> Interaction has stopped completely, yet the machine will respond to
> pings.
>
> No error logs of any kind are generated, and the system doesn't even
> log the fact that the power switch has been pressed (the only way
> out.)
>
> If anyone has seen this before, I would be extremely interested in
> hearing their experiences.
> Our support vendors have absolutely no idea what is going on since
> there is no diagnostic trace left.
>
> thanks
> Colin

Hi Collin,

That sort of strangulation might be caused by a filesystem becomming filled
to 100%. The usual suspects are /, /tmp, /var. Use cron and df to monitor
that regularly and write the output into a file (residing in a different
filesystem - not /tmp!). Shorten the interval of the syncd. Run skulker
(every night). Check number of licences and number of processes per user.
Btw. should you have activated system dumps and you did not get any you
might increase the size of the dump device and/or the dump copy device.

HTH,
Andreas



Relevant Pages

  • Re: Strange System Behaviour
    ... > We have a system that is exhibiting strange behaviour. ... > I don't know a lot about IBM systems, ... application is leaking the memory and have the vendor fix it. ... case I run Oracle on them and it appears to be the listener not ...
    (comp.unix.aix)
  • Re: Strange System Behaviour
    ... >> We have a system that is exhibiting strange behaviour. ... >> I don't know a lot about IBM systems, so appologies for the woolly ... SNIP ...
    (comp.unix.aix)