Don't even TRY to run SpamAssassin on such a small machine.
I ran a much earlier incarnation on a small machine and decided it was
just barely practical at the mail volumes of the time. It took about a
minute to scan each message. Moving to a larger machine helped. Moving
to a much larger machine really helped.
{O.O}
----- Original Message -----
From: "James Lavery" <[EMAIL PROTECTED]>
To: "'Daryl C. W. O'Shea'" <[EMAIL PROTECTED]>
Cc: <users@spamassassin.apache.org>
Sent: Tuesday, October 24, 2006 15:22
Subject: RE: Max-children setting not high enough causing spamassassin to hang?
Hi Daryl,
Aha - yes I think you've right. The Slug only has (I think) 32MB of memory,
and the symptoms are definitely of swap thrashing.
So I presume what I should do is reduce the max children, and put up with
reduced spamassassin performance when a lot of mails come in.
James
-----Original Message-----
From: Daryl C. W. O'Shea [mailto:[EMAIL PROTECTED]
Sent: 24 October 2006 07:10
To: James Lavery
Cc: users@spamassassin.apache.org
Subject: Re: Max-children setting not high enough causing spamassassin to
hang?
James Lavery wrote:
Hi all,
Version: 3.1.4
OS - Unslung 6.8 on a Linksys NSLU2 (Slug) Running spamd with
fetchmail and postfix
How much memory do one of those things have?
A couple of days ago the Slug hung with masses of disk activity; when
I looked at what was going on, it seemd that spamd and/or spamc were
chewing up resources.
Looking at the spamd log file (relevant part included below) indicates
that I hadn't got my max-child setting high enough to cope with a
flurry of emails coming in. Is this my problem, or is this a red
herring, and is it the warning logged by child process 17429 '
copy_config timeout, respawning child process..'? After this, it sems
that 17429 no longer talks to the parent process.
respawning child... ie. the child kills itself and the parent spawns a new
one. Dead children don't talk to their parents much. ;)
I'd be surprised if too low a max-children setting would be causing the
problem, so can someone shed light on what the problem was with this
child process? I had to restart spamd to get things going again.
It's the opposite. Too large of a setting causes the machine to use up
all it's physical memory and swap thrash. copy_config timeouts are
always indicative of this, or insane CPU load.
Daryl
