Alexandre Gauthier wrote: >> /usr/bin/perl5.8.6 -T -w /usr/bin/spamd > You run spamassassin using the perl script? Erm...(checks cron jobs and scripts)...no, not me. > Or it might be spamd itself, that's a possibility.. I had the perl > version of spamd (as compared to spamc the C daemon) bring down mail > servers inside client networks... I do in fact have spamc running for mail checking, I call spamc from my procmailrc: /usr/bin/spamc spamd being the daemon, and spamc the client for the daemon. I start spamd from launchd, using the plist here: http://adrinux.perlucida.com/node/25 And that also calls spamd directly. So I looked and it turns out that /usr/bin/spamd is actually a perl script, with shebang line "!/usr/bin/perl5.8.6 -T -w" which pretty much explains it. > I'd look at the spamc option if I were you, that's a stab in the dark, > but... I am running spamc :) A quick refresher (having just reminded myself): spamc (written in C) is the client for spamd (written in perl) a daemonized version of spamassassin (written in perl). Calling /usr/bin/spamassassin for every mail you process can certainly bring servers to their knees, calling spamc/spamd is the lower overhead option. Your command did allow me to differentiate between spamc/d and sa-learn though. And it seems to be sa-learn that was hogging the cpu, at a time when no new mail was coming in (which is why I didn't think it was spamc/d). I run sa-learn from cron periodically to learn those grey-area spam and ham that I've hand sorted. But for the ham I'm doing it in an inefficient manner, scanning over all my stored mail in order to pick up the now very rare 'spammy' ham. I probably need Mail.app and Thunderbird plugins to to sa-learn ham that I've hand sorted or something like that. -- adrinux (aka Adrian Simmons) <http://adrinux.perlucida.com> e-mail <mailto:adrinux at gmail.com> AOL/Yahoo IM: perlucida, Microsoft: adrian at perlucida.com