Autor: Trevor Sky Garside Data: Dla: exim-users CC: Trevor Sky Garside Temat: Re: [Exim] HELP! Exim processes are running away with 100+MB of
RAM, crashing my mail servers!
On Wed, 30 Jan 2002, Philip Hazel wrote:
> See concurrent thread "Possible DOS for Exim".
>
> Do you have a system filter that refers to $message_headers?
>
> Have you received a message with an exceedingly large number of headers?
Ahh, were it only that simple. I'm not using any filters whatsoever. I
had a look through my config file to double check that nothing was looking
for headers, and nothing is. :(
Is there something I can do specifically to find out why Exim would be
allocating so much RAM? I've been watching it for some time now on our
more powerful mail server (p-III 733 w/ 768MB RAM) and I see two things:
Sometimes, a process will jump up to, say, 200MB RAM, then go away, and
that'll be it.
Other times, I'll see one process hit about 100MB or so, then a bunch of
processes right after it are at the 100MB+ mark, within 1 or 2 PIDs of the
first one to get big.
Again, I can't seem to corrolate anything between message size or abnormal
headers with this happening. Eventually, the processes finish, and the box
goes back to normal. What's keeping it alive is when the system load
reaches 3, queue runners stop, delivery stops, and SMTP connections are
refused.
