identical messages -- some get bayes score, some don't

[Glenn Steen]

2009/1/9 Cannon Watts <cwatts at elsberry.k12.mo.us>:
> On Fri, January 9, 2009 5:31 am, Kai Schaetzl wrote:
>> Cannon Watts wrote on Thu, 8 Jan 2009 21:22:59 -0600 (CST):
>>
>>> [2456] dbg: async: starting: URI-DNSBL,
>>> DNSBL:dob.sibl.support-intelligence.net:agentbenefitsteam.com
>>> (timeout 10.0s, min 2.0s)
>>>
>>> [2456] dbg: async: starting: URI-NS,
>>> NS:agentbenefitsteam.com
>>> (timeout 10.0s, min 2.0s)
>>>
>>> [2456] dbg: async: starting: DNSBL-A,
>>> dns:A:154.248.19.72.plus.bondedsender.org.
>>> (timeout 10.0s, min 2.0s)
>>>
>>> [2456] dbg: async: starting: DNSBL-TXT,
>>> dns:TXT:154.248.19.72.bl.spamcop.net.
>>> (timeout 10.0s, min 2.0s)
>>
>> there's a problem with your DNS or caching ns. Until you haven't solved
>> that better disable network tests. Even after you are ok again you may
>> want to disable some of these tests as they are not worth it.
>
> Probably getting beyond the scope of this list, but any tips on debugging
> this?  This particular box is running its own caching DNS that, prior to
> seeing that debugging info, I would have said works perfectly.  Dozens of
> clients on our local network use that DNS server without a problem, not
> to mention the fact that sendmail on this same machine has no DNS problems.
>
> How would I go about disabling 'some of these tests'?  set skip_rbl_checks
> in /etc/mamil/spamassassin/mailscanner.cf?
>
>
>>> And perhaps most importantly:
>>> [2456] dbg: locker: safe_lock: trying to get lock on
>>> /etc/MailScanner/bayes/bayes with 10 timeout
>>
>> check the permissions, look for existing lock files and remove them.
>> Apparently, this didn't happen for all messages. So, check messages one by
>> one and see if it then still happens. Maybe there's a performance problem?
>
> I don't understand how permissions could be an issue given the
> circumstances.  SpamAssassin is running as root, and all of these messages
> are in the same mailbox -- it's not as if they're owned by different users.
>
> I did run each message separately through spamassassin -D.  This time they
> all received Bayes scores, with 15 scoring BAYES_50 and 13 scoring BAYES_60.
> All of them generated the dns timeouts, but only 19 of the 28 generated the
> bayes timeout.
>
> I don't see any suspicious lock files, but then I'm not sure what I'm
> looking for.
>
Do you have a very large bayes_seen file? If so... remove it. Then
redo the tests and see if things aren't better.
Or do you have any expire files, from failed expiry runs?

> I suppose there could be a performance problem, but considering I just
> moved   this server from a 933 Mhz Pentium with less than a gig of ram
> (where it
> was working reasonably well) to a 2 GHz quad-core w/ 4 GB of RAM and 15k
> rpm disks (where I've never seen the system load go over 0.5), I tend to
> look elsewhere first.
>
Ok, so this situation popped up right after the move?

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