I had the same problem: the fault was in the configuration file:
take care that the 'condition' statement is before the 'spam' statement
otherwise EVERY mail well be send to spamd and hat consumes LOTS of
memory and CPU
warn message = X-Spam-Score: $spam_score ($spam_bar)
condition = ${if <{$message_size}{500k}{1}{0}}
spam = exim:true
warn message = X-Spam-Report: $spam_report
condition = ${if >{$spam_score_int}{40}{1}{0}}
spam = exim:true
# Add X-Spam-Flag if spam is over system-wide threshold
warn message = X-Spam-Flag: YES
condition = ${if <{$message_size}{500k}{1}{0}}
spam = exim
# Reject spam messages with score over 5, using an extra condition.
deny message = SPAM This message scored $spam_score SPAM points.
condition = ${if >{$spam_score_int}{50}{1}{0}}
spam = exim:true
hope that helps
Wolfgang
>Greetings list,
>
>following system is behaving very strange:
>
>RH Enterprise Linux 4
>Exim 4.52
>SA 3.0.4
>
>After upgrading exim from 4.43 to 4.52 today, the memory usage
>from _SpamAssassin_ skyrocketed. And it skyrocketed that much
>and so fast that I had to do a hardware-reboot after like 1 hour.
>Mind you, that machine has 2 Gig physical and 4 Gig swap ram.
>
>I was barely able to get a ps axu to see that all of my 25 spamd
>instances were hogging around 500mb ram each. Sadly the kernel
>started to kill random applications to free memory then and I lost the
>machine.
>
>I have now reduced the spamd instances to 10 on all servers to
>reduce the chance of the system running out of memory, but that
>shouldn't be an issue normally since they are well suited for
>their job.
>
>With exim 4.43 this behaviour does not occur.
>
>Top from another mx with the same setup, but still running exim 4.43:
>
> PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
>19462 root 16 0 53072 46m 2940 S 0.0 2.3 0:02.26 spamd
>19463 root 16 0 48056 41m 2940 S 0.0 2.1 0:01.67 spamd
>19466 root 16 0 39520 33m 2940 S 0.0 1.6 0:01.95 spamd
>19471 root 17 0 34880 28m 2940 S 0.0 1.4 0:01.61 spamd
>19465 root 16 0 33424 27m 2944 S 0.0 1.4 0:01.27 spamd
>19469 root 17 0 33232 27m 2940 S 0.0 1.3 0:01.31 spamd
>19468 root 17 0 32540 26m 2932 S 0.0 1.3 0:00.78 spamd
>19470 root 16 0 32488 26m 2932 S 0.0 1.3 0:00.65 spamd
>19464 root 17 0 32476 26m 2940 S 0.0 1.3 0:01.06 spamd
>19467 root 16 0 32296 26m 2932 S 0.0 1.3 0:00.55 spamd
>19458 root 25 0 31152 25m 2856 S 0.0 1.2 0:00.48 spamd
Top from the today updated mx with exim 4.52 (freshly restarted
spamassassin and exim like 30 minutes ago):
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
3386 root 16 0 189m 184m 2944 S 0.0 9.1 0:07.24 spamd
3384 root 16 0 69480 63m 2944 S 0.0 3.1 0:03.69 spamd
3387 root 16 0 51984 46m 2944 S 0.0 2.3 0:02.89 spamd
3381 root 16 0 45404 39m 2944 S 0.0 2.0 0:03.83 spamd
3383 root 16 0 45300 39m 2948 S 0.0 2.0 0:06.71 spamd
3382 root 17 0 42120 36m 2944 S 0.0 1.8 0:03.06 spamd
3380 root 16 0 35788 30m 2948 S 0.0 1.5 0:02.61 spamd
3389 root 16 0 35700 30m 2944 S 0.0 1.5 0:01.81 spamd
3385 root 17 0 35464 30m 2940 S 0.0 1.5 0:02.19 spamd
3388 root 16 0 33660 28m 2944 S 0.0 1.4 0:01.99 spamd
3375 root 25 0 30332 25m 2860 S 0.0 1.2 0:00.46 spamd
As you can see, spamd is already starting to hog memory again.
There are no current tcp connections from exim to spamd at that
very moment, so it can't result from a hanging tcp-session.
I have no clue who to blame and where to actually start searching
for the problem. I know that SpamAssassin is the memory hog,
but obviously, for some strange reason, exim seems to trigger
that problem.
Does anyone have a hint where to start digging?
sash
--------------------------------------------------
Sascha Runschke
Netzwerk Administration
IT-Services
ABIT AG
Robert-Bosch-Str. 1
40668 Meerbusch
----------------------------------------------------------------------
Wolfgang Fuertbauer (e-Mail: Wolfgang.Fuertbauer@???)
EBEWE Pharma
Mondseestrasse 11
4866 Unterach, Austria
Tel +43 7665 8123 315
Fax +43 7665 8123 11
http://www.ebewe.com
