Eric "Shubes" wrote:
Warren (mailing lists) wrote:
Jake Vickers wrote:
Warren (mailing lists) wrote:
sa-learn is using so much memory and CPU time that it is killing my
server.  I have renamed it for now, but would like to know if there
is a way to limit the amount of memory or processes that it uses?
I had the same problem here. 67M a process eats up the ram quickly.
Go to your run file for spamd (/var/qmail/supervise/spamd/run) and add
the --maxchildren=x option to limit how many processes it will spawn:
exec /usr/bin/spamd --max-children=5 -x -u vpopmail -s stderr 2>&1

After trying what you described and going as low as 2, It seems that
spamd still spawns many sa-learn jobs regardless of that setting.  I
still have to rename sa-learn, at least until I get more RAM.

W

Warren,

How are you running sa-learn? Are you talking about the autolearn option?
(apparently)

Perhaps your bayes journal hasn't sync'd in a while (or ever). Try running
# sa-learn -u vpopmail --force-expire
# chown vpopmail:vchkpw /home/vpopmail/.spamassassin/bayes_toks
and see if that doesn't speed things up. What does this sa-learn report when
you run it?

You know I only half read your original email Warren. I thought you had meant that the spamd process was bogging your machine down as emails came in. How are you running sa-learn? Is it in the mailfilter script? Sorry about that!

Attachment: smime.p7s
Description: S/MIME Cryptographic Signature

Reply via email to