You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@spamassassin.apache.org by John Fleming <jo...@wa9als.com> on 2004/10/30 17:04:36 UTC
Load average probs
I've followed the "memory-hog" threads of late, but as a rather
inexperienced Linux person, some of it is over my head. I'm using SA v2.64
with network tests and bayes. Everything is fine except that several times
a day the server seems overwhelmed with load averages 4-12. Here is a minor
example hitting 4.29:
07:45:01 runq-sz plist-sz ldavg-1 ldavg-5 ldavg-15
07:55:01 0 60 0.07 0.09 0.03
08:05:01 2 58 0.00 0.00 0.00
08:15:01 2 58 0.00 0.00 0.00
08:25:01 2 58 0.00 0.00 0.00
08:35:01 2 63 0.02 0.01 0.00
08:45:01 2 58 0.00 0.00 0.00
08:55:06 0 75 4.29 2.48 1.02
09:05:01 0 58 0.00 0.42 0.63
Average: 2 59 0.12 0.08 0.04
When the high LA hits, available RAM is essentially nil and the swap space
is about 85% used as well. When I've seen it hit 12 or so, it seemed that
the HDD activity would never stop, and I've manually killed spamassassin and
any spamd's.
This is on a recent Dell 400SC with 512 MB RAM and IDE drives. I suppose
this might be a ridiculous question to some of you, but can you make an
"experienced guess" as to whether the high load averages are due to my using
network tests, due to insufficient RAM, or other?
It -is- related to SpamAssassin, not some other process running on the
machine. This is a small-scale family server receiving only about 250
emails a day. I'm using a simple postfix/procmail/SA setup. Procmail calls
spamc.
I guess I wouldn't have expected to see these load averages on such a
small-scale server.
Thanks any tips! - John
Re: Load average probs
Posted by Tom Collins <to...@tomlogic.com>.
On Oct 30, 2004, at 8:04 AM, John Fleming wrote:
> When the high LA hits, available RAM is essentially nil and the swap
> space
> is about 85% used as well. When I've seen it hit 12 or so, it seemed
> that
> the HDD activity would never stop, and I've manually killed
> spamassassin and
> any spamd's.
On my server, I ran into a scenario where a spamd process would get
stuck and start eating up memory. I manually expired my bayes
database, but it would still happen.
After trying a lot of things, I got rid of my old autowhitelist (single
whitelist for all accounts on server). It had been around for a long
time, and had grown to over 300MB. That solved the problem, so it must
have either been corrupted in some way, or spamd was trying to do
maintenance on it.
I don't know enough about how the whitelist file is maintained to know
which was the case, but if you might want to try temporarily removing
your whitelist to see if the problems go away. Just stop spamd, rename
the whitelist, and start it up again.
--
Tom Collins - tom@tomlogic.com
QmailAdmin: http://qmailadmin.sf.net/ Vpopmail: http://vpopmail.sf.net/
Info on the Sniffter hand-held Network Tester: http://sniffter.com/
--
Tom Collins - tom@tomlogic.com
QmailAdmin: http://qmailadmin.sf.net/ Vpopmail: http://vpopmail.sf.net/
Info on the Sniffter hand-held Network Tester: http://sniffter.com/
Re: Load average probs
Posted by Michael Parker <pa...@pobox.com>.
On Sat, Oct 30, 2004 at 10:04:36AM -0500, John Fleming wrote:
>
> This is on a recent Dell 400SC with 512 MB RAM and IDE drives. I suppose
> this might be a ridiculous question to some of you, but can you make an
> "experienced guess" as to whether the high load averages are due to my using
> network tests, due to insufficient RAM, or other?
>
If I had to guess, I'd guess bayes auto expire. Are you running site
wide bayes? You could try several things: 1) turn off auto expire and
run sa-learn --force-expire via cron at a predicatable hour (ie when
the server is basically idle), 2) instal 3.0.1 and see if that
helps or 3) since you have 3.0.1 installed now try Bayes in SQL which
has an expire that is 7-9 times faster than Bayes DBM.
Michael
Re: Load average probs
Posted by jdow <jd...@earthlink.net>.
On another paw I note that most family tools are not left running
24x7. If this is his case then a large portion of his 250 messages
may be coming in right after he boots. If he is setup to spawn
too many spamds then he could experience a memory crisis.
{^_-}
From: "JamesDR" <ro...@bellsouth.net>
> He did say what his mail average is:
> "This is a small-scale family server receiving only about 250
> emails a day."
> You may want to read the entire message next time.
> Thanks,
> JamesDR
>
> Loren Wilton wrote:
> >>When the high LA hits, available RAM is essentially nil and the swap
space
> >>is about 85% used as well. When I've seen it hit 12 or so, it seemed
that
> >>the HDD activity would never stop, and I've manually killed spamassassin
> >
> > and
> >
> >>any spamd's.
> >
> >
> > You don't say what your normal mail rate is, nor what rules you are
running.
> >
> > 2.64 didn't have some of the memory problems that were really bad on the
> > initial 3.0 release, and are now largely corrected. Generally all that
> > would cause the sort of thing you are seeing would be either running too
> > many spamd children for the available memory, or using some of the
really
> > big addon rule sets like BigEvil.
> >
> > How many spamd children do you have running? You should probably limit
to
> > 3-5 at most with 512M, at a guess. Are you using BigEvil? If so, get
rid
> > of it and use the surbl patch instead.
> >
> > You are clearly getting into swap, and this implies that you have
overflowed
> > memory.
> >
> > Loren
> >
> >
>
Re: Load average probs
Posted by JamesDR <ro...@bellsouth.net>.
He did say what his mail average is:
"This is a small-scale family server receiving only about 250
emails a day."
You may want to read the entire message next time.
Thanks,
JamesDR
Loren Wilton wrote:
>>When the high LA hits, available RAM is essentially nil and the swap space
>>is about 85% used as well. When I've seen it hit 12 or so, it seemed that
>>the HDD activity would never stop, and I've manually killed spamassassin
>
> and
>
>>any spamd's.
>
>
> You don't say what your normal mail rate is, nor what rules you are running.
>
> 2.64 didn't have some of the memory problems that were really bad on the
> initial 3.0 release, and are now largely corrected. Generally all that
> would cause the sort of thing you are seeing would be either running too
> many spamd children for the available memory, or using some of the really
> big addon rule sets like BigEvil.
>
> How many spamd children do you have running? You should probably limit to
> 3-5 at most with 512M, at a guess. Are you using BigEvil? If so, get rid
> of it and use the surbl patch instead.
>
> You are clearly getting into swap, and this implies that you have overflowed
> memory.
>
> Loren
>
>
Re: Load average probs
Posted by Loren Wilton <lw...@earthlink.net>.
> When the high LA hits, available RAM is essentially nil and the swap space
> is about 85% used as well. When I've seen it hit 12 or so, it seemed that
> the HDD activity would never stop, and I've manually killed spamassassin
and
> any spamd's.
You don't say what your normal mail rate is, nor what rules you are running.
2.64 didn't have some of the memory problems that were really bad on the
initial 3.0 release, and are now largely corrected. Generally all that
would cause the sort of thing you are seeing would be either running too
many spamd children for the available memory, or using some of the really
big addon rule sets like BigEvil.
How many spamd children do you have running? You should probably limit to
3-5 at most with 512M, at a guess. Are you using BigEvil? If so, get rid
of it and use the surbl patch instead.
You are clearly getting into swap, and this implies that you have overflowed
memory.
Loren