Morten Nilsen wrote:
> Philippe Laquet wrote:
> 
>>Morten Nilsen wrote:
>>
>>>http://84.234.141.4/amavis-problem/index2.html
>>
>>Yup, note that the process number 9300 shuts down at 00:59, the process
>>ID probably indicates an older amavisd process initiated some hours ago.
>>
>>Nov 14 00:59:45 ryo-ohki amavis[9300]: Net::Server: 2005/11/14-00:59:45
>>Server closing!
> 
> 
> that's likely the max_requests limit kicking in
> 
> 
>>As it dies, another child process is started :
>>
>>Nov 14 00:59:48 ryo-ohki amavis[28524]: Net::Server: Process Backgrounded
>>Nov 14 00:59:48 ryo-ohki amavis[28524]: Net::Server: 2005/11/14-00:59:48
>>Amavis (type Net::Server::PreForkSimple) starting! pid(28524)
>>Nov 14 00:59:48 ryo-ohki amavis[28524]: Net::Server: Binding to UNIX
>>socket file /var/spool/amavis/amavisd.sock using SOCK_STREAM
>>
>>And gives the error about the already bound port 10024 :
>>
>>Nov 14 00:59:48 ryo-ohki amavis[28524]: Net::Server: 2005/11/14-00:59:48
>>Can't connect to TCP port 10024 on 127.0.0.1 [Address already in use]\n
>> at line 88 in file /usr/lib/perl5/site_perl/5.8.5/Net/Server/Proto/TCP.pm
>>
> 
> 
> that port would be kept open by the hung process
> 
> 
>>>now, this is the point where mail stops getting delivered, but the 
>>>actual problem started some time earlier, right after 23:00 .. but what?
>>>
>>
>>The average load on your server may be due to CRON jobs, I don't know,
>>but there are "AV check" that takes a huge time even for small messages.
> 
> 
> no. the big green blob is a single amavisd process stuck at 99% cpu
> 
> 
>>What is your SpamAssassin setup, do you use Razor2 ?
> 
> 
> /var/spool/amavis/.spamassassin/local.cf:

The correct filename should be : /var/amavis/.spamassassin/user_prefs

What is the content of your /etc/spamassassin/local.cf ?

The local.cf is only used under /etc/spamassassin for global settings.
All other configuration files under home directories are named
"user_prefs" - As far as I know... ;)

> use_auto_whitelist 0
> use_bayes 1
> 
> score BAYES_60 1
> score BAYES_70 2
> score BAYES_80 3
> score BAYES_90 5
> score BAYES_99 10
> 
> -rw-------  1 amavis amavis     10539008 Nov 14 23:24 bayes_seen
> -rw-------  1 amavis amavis 310314565632 Nov 14 23:15 bayes_toks

Yup, really strange... I don't remember that kind of wrong filesize, but
that is interesting. Did you try to look on the main site (FAQ...) :

http://www.ijs.si/software/amavisd/

What is the version of amavisd you are using ? And under which TSL ?

> 
> what's the deal with that huge number?
> 
> du -h bayes_*
> 8.1M    bayes_seen
> 4.1M    bayes_toks
> 
> I have a shitload of bayes_toks.expireXXXXX files .. could they be related?
> 
> 
>>In amavisd.conf how many children process are set ?
> 
> 
> I used to have 2, increased it to 5 after I posted the logs


> 
>>And at least, take a look in your Postfix master.cf file, in the amavisd
>>section you probably have a limited number of chlidren process, does it
>>match your amavisd.conf setting ?
> 
> 
> no such thing in master.cf
> 
> Cheers,


Kind regards,

Philippe Laquet

        

        
                
___________________________________________________________________________ 
Appel audio GRATUIT partout dans le monde avec le nouveau Yahoo! Messenger 
Téléchargez cette version sur http://fr.messenger.yahoo.com
_______________________________________________
tsl-discuss mailing list
[email protected]
http://lists.trustix.org/mailman/listinfo/tsl-discuss

Reply via email to