Rocco wrote:

> Thanks Gary V for you answers,
> Could  you suggest me something about this problem..

>> I have the same performance problems..
>> 
>> I will prune the bayes_token (InnoDB) table with the following query:
>> 
>> delete from bayes_token where atime <=
>> UNIX_TIMESTAMP(DATE_ADD(CURDATE(), INTERVAL -30 DAY));
>>

I would never do this. I would let sa-learn and spamassassin
maintain the Bayes database.

>> but I get however a growing data lenght for that table, as 
>> well a growing index lenght.
>> 
>> At certain time, I get poor performance for amavisd-new  
>> (Indeed I get this kind of error from amavis-milter - I have 
>> sendmail-milter
>> configuration:
>> 
>> amavis-milter: thread_create() failed: 12, try again
>> 
>> )
>> 
>> so I have to switch from MySQL DB to the native (local) DB for SA.
>> 
>> Maybe is it worth to change to MyISAM table?
>>
> TIA,
> rocsca

In general I would use MySQL for Bayes (with InnoDB) and would
increase innodb_buffer_pool_size (and innodb_log_file_size if it lets
you) from the defaults. The defaults appear to be a performance
bottleneck.

I don't know anything about amavis-milter or why it gave you this error.

I would create and use the lastupdate field in the AWL table as I have
noted: http://www200.pair.com/mecham/spam/debian-spamassassin-sql.html

Actually, I personally don't use AWL but don't persuade others not to.

Gary V


-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
AMaViS-user mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/amavis-user
AMaViS-FAQ:http://www.amavis.org/amavis-faq.php3
AMaViS-HowTos:http://www.amavis.org/howto/

Reply via email to