[Mimedefang] memory leak? is it an IBM x-series?

[Ron Peterson]

On Fri, 13 Feb 2004, Ron Peterson wrote:
> On Fri, 13 Feb 2004 Anthony.Caetano at t-systems.co.za wrote:

> > We had similar problems with an IBM x345 server -- we tried many things 
> > eventually updating to 2.4.23-pre7 and up fixed it.
> > 
> > I updated the tg3 driver and stability improved a bit, I updated the ibm 
> > ServeRaid driver (ips module) level to 610 and stability improved a bit, 
> > disabled hyperthreading did many and various other things 
> > (bios/driver/patch), but for us 2.4.23-pre7 and up fixed it -- running 
> > 2.4.25-pre6 now without problems.

> Meanwhile, I've purchased a handful of Intel PRO/1000 MT adapters which
> I'm going to try.  If I have the same problem with a completely different
> adapter, that should rule that out.

Lost the machine again, but getting closer to the cause, I think.

I had reset confMAX_DAEMON_CHILDREN to 60.  I had a little over 70
sendmail processes running.  I killed all sendmail processes.  Didn't come
back right away.  I took down the virtual interface eth0:2 which the
primary MX record for our domain points to.  Shortly after, the machine
was just fine.  Brought the interface and sendmail back up, and everything
is hunky dory.

In the process of looking at the kernel configuration for the
IntelPRO/1000, I discovered the module includes a NAPI option.  So I dug
around to see what that was about.  Found the following very interesting
article.  Don't know if it directly applies to my situation, but the
possibility sure seems to be there.  I'm going to put an IntelPRO/1000 in
another piece of identical but less critical hardware today, to play with
it a bit.  Assuming that all goes well, I'll put one in my mail server
next.

http://www.usenix.org/publications/library/proceedings/als01/full_papers/jamal/jamal_html/napi2.html

_________________________
Ron Peterson
Network & Systems Manager
Mount Holyoke College