bayes problem locking problem ?
Brady A. Tucker
batucker at ICNET.NET
Wed May 26 23:12:37 IST 2004
I've been having similar problems on one of my 3 boxes, it's the 'busiest'
of the three.
I did move bayes it into tmpfs earlier today (with an hourly backup script,
and a re-populate .spamassassin on boot), significantly bettered my 'Spam
Checks completed' rates, and has helped keep my spool (which has been
running > 1000) down to incoming only.
Dangerous.. but it works, and I guess I can live with an hour of lost Bayes
info on failure/reboot (rare). Don't understand why this started for me all
the sudden either, about a week ago, even whacking the bayes db and
starting with the starter database previously supplied by some wonderful
person here didn't alleviate the problem.
I've been trying to find something that's change.. some DBI module or
something that's freaking out.... to no avail as of yet, no yums/up2dates
pre-problem, LOTS since... blah.
Somethings definitely up software wise, I tar'ed up 1200 messages moved
them to my tertiary server and it cleared through them in < 15 minutes
(Exact same config on tertiary as below except Fedora Core 1)
Configuration/Load :
-Average about 180,000 messages/day through this system.
-Acts mostly as a gateway - with about 10% of passed mail being delivered
locally hosted domains
-Sendmail + Redhat 9 + MS + Razor + Pyzor + DCC (daemon)
-incoming in tmpfs
-bayes in tmpfs (now)
-Barton 2500 + 1 Gig Corsair DDR400 Ram + Mylex Ultra 160 Raid 5 Array
-Kernel 2.4.20-31.9
Brady A. Tucker
Internet Complete! inc.
http://www.icnet.net
-----Original Message-----
From: MailScanner mailing list [mailto:MAILSCANNER at JISCMAIL.AC.UK]On
Behalf Of Robert Mena
Sent: Wednesday, May 26, 2004 9:07 AM
To: MAILSCANNER at JISCMAIL.AC.UK
Subject: Re: bayes problem locking problem ?
Hi MailScanners :) anyone with a solution for this ?
I've restarted spassassin but with no bayes and so far
it is running smooth (but letting all spam go
through)...
--- Eelco Wesemann <eelco at NETNATION.NL> wrote:
> Hi,
>
> No sollution, but a "same here". I've had similar
> behavour on my setup:
> sendmail+clamav+sa.
> When running MS and SA in debug mode, I noticed that
> around 50% to 80% of the
> processing time was spent on obtaining a lock on the
> bayes database.
>
> My setup:
> debian unstable, kernel 2.6.5
> sendmail-8.12.11
> MailScanner-4.30.3
> clamav-0.71
> spamassassin-2.63
>
> Max SpamAssassin Size = 30000
> Max Unscanned Bytes Per Scan = 100000000
> Max Unsafe Bytes Per Scan = 50000000
> Max Unscanned Messages Per Scan = 40
> Max Unsafe Messages Per Scan = 40
> Max Normal Queue Size = 2000
>
> Bayes magic:
> 0.000 0 2 0 non-token
> data: bayes db version
> 0.000 0 163094 0 non-token
> data: nspam
> 0.000 0 13224 0 non-token
> data: nham
> 0.000 0 435207 0 non-token
> data: ntokens
> 0.000 0 1084290758 0 non-token
> data: oldest atime
> 0.000 0 1085525323 0 non-token
> data: newest atime
> 0.000 0 1085525334 0 non-token
> data: last journal sync atime
> 0.000 0 1084784968 0 non-token
> data: last expiry atime
> 0.000 0 345600 0 non-token
> data: last expire atime delta
> 0.000 0 189064 0 non-token
> data: last expire reduction count
>
> Would putting the bayes dirs in tmpfs increase the
> speed of the locking of the
> bayes-databases by SA?
>
> Regards,
>
> Eelco
>
> Robert Mena wrote:
> > Hi everyone,
> >
> > I replaced my exim+rav setup sunday with a
> > exim+ms+clamav+sa. Everything was ok, the load
> was
> > higher but the flow of the messages seemed smooth.
> >
> > Today I've noticed that the exim.in queue grew up
> to
> > 8000 messages. In the two previous days it never
> > exceeded 500. The out queue was higher yesterday
> (up
> > to 10K messages) despite the smoother behaviour so
> It
> > does not seem an attack or increased volume of
> > incoming spam.
> >
> > After I've switched off the spamassassin (still
> > mailscanner + clamav) the exim.in queue went from
> 8K
> > to 0 in a matter of minutes.
> >
> > Any ideas ?
> >
> > My setup
> > - Linux Fedora Core 1 - kernel
> 2.4.22-1.2188.nptlsmp
> > - mailscanner-4.30.3-2
> > - spamassassin-2.63-1
> > - exim 4.32
> > - clamav-0.70
> >
> > My General Configuration setup looks like this:
> >
> > Max SpamAssassin Size = 30000
> > Max Unscanned Bytes Per Scan = 0
> > Max Unsafe Bytes Per Scan = 50000000
> > Max Unscanned Messages Per Scan = 0
> > Max Unsafe Messages Per Scan = 100
> > Max Normal Queue Size = 5000
> >
> > My sa-learn dump:
> > 0.000 0 2 0 non-token
> > data: bayes db version
> > 0.000 0 8856 0 non-token
> > data: nspam
> > 0.000 0 8072 0 non-token
> > data: nham
> > 0.000 0 357214 0 non-token
> > data: ntokens
> > 0.000 0 1000210340 0 non-token
> > data: oldest atime
> > 0.000 0 1085521354 0 non-token
> > data: newest atime
> > 0.000 0 1085521390 0 non-token
> > data: last journal sync atime
> > 0.000 0 1085409209 0 non-token
> > data: last expiry atime
> > 0.000 0 852711 0 non-token
> > data: last expire atime delta
> > 0.000 0 0 0 non-token
> > data: last expire reduction count
>
> -------------------------- MailScanner list
> ----------------------
> To leave, send leave mailscanner to
> jiscmail at jiscmail.ac.uk
> Before posting, please see the Most Asked Questions
> at
> http://www.mailscanner.biz/maq/ and the archives
> at
> http://www.jiscmail.ac.uk/lists/mailscanner.html
__________________________________
Do you Yahoo!?
Friends. Fun. Try the all-new Yahoo! Messenger.
http://messenger.yahoo.com/
-------------------------- MailScanner list ----------------------
To leave, send leave mailscanner to jiscmail at jiscmail.ac.uk
Before posting, please see the Most Asked Questions at
http://www.mailscanner.biz/maq/ and the archives at
http://www.jiscmail.ac.uk/lists/mailscanner.html
-------------------------- MailScanner list ----------------------
To leave, send leave mailscanner to jiscmail at jiscmail.ac.uk
Before posting, please see the Most Asked Questions at
http://www.mailscanner.biz/maq/ and the archives at
http://www.jiscmail.ac.uk/lists/mailscanner.html
More information about the MailScanner
mailing list