DNS server issue hangs entire site

[phxazcraig]

In article <8wnFm.694$5Q1.305@kovat.provo.novell.com>, De wrote:
> here are recent changes to the subject server, although all over a week
> ago:
>
> - Memory added to bring it to 8GB approx. a month ago

NW is a 32 bit OS, it will only be able to address 4GB of RAM, and it will
really only be able to use 3.3GB.

> - A failed RAID drive replaced approx. a month ago

Not related.

> - Norton Antivirus, latest version, made active approx. 8 days ago

Hmm - possible, though unlikely.

> - Memory tuning on the server, using Memcalc, approx. 8 days ago

You should no longer need to use this with SP7 or SP8. I'd be a bit
suspicious here, if you have hard-configured the amount of RAM available to
eDir.
>
> None of the servers have shown any problems in the 8 days prior to this,
> although the AIRP server has consistently shown the following error
> averaging twice/day, and the extra memory & tuning do not seem to have
> helped:
> "CPU Utilization-0 on server AIRP was in a SUSPECT state" showing some
> value between 94 & 100.

No big deal, but it would be nice to know what process/thread is at the top
of the list when this happens. (No big deal if the utilization drops back
down to normal, which should probably be under 10% most of the time).

>
> Looking for thoughts on how to diagnose this without much by way of real
> error messages, since it does not seem to have been a switch issue as I
> originally thought possible (since I see no sign of network dropping on
> any of the servers). And looking how to fix it and/or guard against it
> happening again.

Look in Monitor, Kernel, Busiest Threads. Look in Monitor at the memory,
and see if there are memory leaks. (Get SEG.NLM to help with that).
>
> Restarting the primary file/print server at the site did resolve the
> problem, but I really don't want it happening again & I don't know where
> to start. My only thought is that the server seems to be getting
> overwhelmed & perhaps something critical happened between the backup &
> NAV, even though it was fine the week before ... or perhaps something
> wrong with the memory tuning (?)

I'd rework the memory tuning, and let it go back to auto tuning itself.
Check for NAV updates. Check the console log as well, in case it showed
something worthwhile.


Craig Johnson
Novell Support Connection SysOp