Mail server backlogging - seems like a Bayes problem

[Matt Kettler]

<x-flowed>
At 04:28 PM 10/21/2004, Julian Field wrote:
> >Any chance MS is timing SA out and killing it repeatedly? (this can happen
> >if SA does expiry and it takes too long for MailScanner's rather absurdly
> >short default timeout).
>
>What timeout would you consider less "absurd"?

Well, the default is 40 seconds. Way too short given SA's tendency to
wander off and do some DB maintenance (journal sync, expiry, etc) whenever
it feels the need.

As I said in the previous mail, I'd consider 60 seconds a bare minimum, but
even that will result in quite a few timeouts if the auto expire feature
isn't turned off. Perhaps 90 would be better.

I use 120 seconds AND I have auto expire off.
(/etc/mail/spamassassin/local.cf:bayes_auto_expire 0)

Even then once in a rare while SA does an expiry during a message scan
anyway (bug in 2.64?), and MailScanner kills it, leaving a
bayes_toks.expire* file behind.

Did it just a couple days ago on the 20th. In general it's fairly rare.
Still bayes expiry is very slow in SA, and it's a behavior SA sometimes
engages in during message scans.

I have not seen MS kill SA for any reason other than bayes expiry in at
least a year or so. (ever since I set my DNS timeouts in SA to something
shorter than MailScanner's SA timeout, that's not been a problem, it's
always bayes expiry).

------------------------ MailScanner list ------------------------
To unsubscribe, email jiscmail at jiscmail.ac.uk with the words:
'leave mailscanner' in the body of the email.
Before posting, read the MAQ (http://www.mailscanner.biz/maq/) and
the archives (http://www.jiscmail.ac.uk/lists/mailscanner.html).
</x-flowed>