SA lint in MailWatch - 158 seconds?!

[Glenn Steen]

On 28/09/06, Glenn Steen <glenn.steen at gmail.com> wrote:
> On 28/09/06, Martin Hepworth <martinh at solidstatelogic.com> wrote:
> > Brett Charbeneau wrote:
> > > Folks,
> > >
> > >     I'm trying to tune my Debian mail server running MailScanner 4.55.10
> > > - it's a P4 3.2 Ghz with 4GB of RAM. I have mounted MS's incoming
> > > directory as a tempfs and run a local BIND cache.
> > >     I'm experiencing some slow delivery and am using MailWatch as one
> > > tool to keep an eye on the process. In the "Tools/Links" portion of
> > > MailWatch there is a way to run SpamAssassin Lint with some nice timing
> > > details.
> > >     I'm getting 166 seconds or more on the
> > >
> > > dbg: config: score set 1 chosen.
> > >
> > >     line
> > >     Does anyone know where I would look to tighten this up? I'm not even
> > > sure what it refers to...
> > >     Many thanks!
> > >
> > Brett
> >
> > check you're not ALL the RBL's in SA.
> >
> > Pick a couple and give the rest zero scores in
> > /etc/mail/spamassassin/mailscanner.cf
> >
> Martin,
>
> AFAICS it only does a simple (well:-) resolver test before deciding
> whether DNS lookups are available or not. The perhaps tardy lookups
> would happen _after_ reporting which score set has been chosen... so
> that is probably not "it". As usual, I might be wrong though:-).
>
> As Ugo said, it might be something strange due to running it as the
> webserver user... And as Scott noted (implied), Brett isn¨t seeing
> either of DNS or Bayes.
> So... something is really fishy here.
>
> Brett, you don't have any of teh configuration files on NFS (some
> automounted stuff perhaps)? That could account for the ridiculous
> times you are seeing, perhaps...
>
(Showing my postfix roots again... when will I ever learn to think
before typing:-)
Another thought... In your regular logs/on the console, do you see any
HW errors? I'm thinking HDD failures (the usual SeekComplete
thingies)?

-- 
-- Glenn
email: glenn < dot > steen < at > gmail < dot > com
work: glenn < dot > steen < at > ap1 < dot > se