debian 7.6 + ispconfig 3.0.5.4p5 + amavis dramatically slowdown

Discussion in 'Installation/Configuration' started by themark, Mar 12, 2015.

  1. themark

    themark Member

    Dear,
    on the system described on the subject, we have several problem with amavis.
    Basically, it's works pretty slow, in some cases. And we cannot figure out why this happen.
    We have another server (well, few different server) with different s.o. (basically centos) with the same configuration, almost the same mail traffic and same hardward spec that does not present that slowdown.

    As "slowdown" i mean, that in few cases happen that from the time that the mail come to the server, to the time that the system put her on the mailbox can spend more than 60-90 seconds. This is a big problem when happen pick of mail traffic. In fact, we have more that 100-200 mail on the queue when this happen.
    If we disable amavis (commenting content_filter = amavis:[127.0.0.1]:10024 and restarting postfix) all the queue, goes complete in few seconds.

    Mysql connection works, cpu load is small, there is no slow query on the mysql, bayes filter works, clamav works...so, my questione is: what could be the source of this slowdown?

    This problem drive my crazy...any hint?
    Thank you.
     
  2. themark

    themark Member

    Addendum, a log of the message (8273362CA0). Come to the server at the 10:08:21, and put to the mailbox at the 10:11:51

    """"""""""""""""""""""""""""""""""""""""""
    cat /var/log/mail.log |grep 8273362CA0
    Mar 12 10:08:21 node238 postfix/smtpd[32660]: 8273362CA0: client=unknown[xx.xxx.xx.xx], sasl_method=LOGIN, sasl_username=[email protected]
    Mar 12 10:08:26 node238 postfix/cleanup[763]: 8273362CA0: message-id=<DA763EB2BFF64738A10D3CC2DD0B3AF5@xxxxx>
    Mar 12 10:08:28 node238 postfix/qmgr[3536]: 8273362CA0: from=<[email protected]>, size=22447, nrcpt=2 (queue active)
    Mar 12 10:11:51 node238 amavis[2208]: (02208-01-8) Passed CLEAN {RelayedInbound,RelayedOpenRelay}, [xx.xxx.xx.xx]:50196 [xx.xx.xx.xx] <[email protected]> -> <[email protected]>,<[email protected]>, Queue-ID: 8273362CA0, Message-ID: <DA763EB2BFF64738A10D3CC2DD0B3AF5@xxxxxx>, mail_id: jlioMdvvoLeD, Hits: -2.891, size: 22447, queued_as: E867062BB7, 12661 ms
    Mar 12 10:11:51 node238 postfix/smtp[650]: 8273362CA0: to=<[email protected]>, relay=127.0.0.1[127.0.0.1]:1024, conn_use=8, delay=212, delays=8.3/191/0/13, dsn=2.0.0, status=sent (250 2.0.0 from MTA(smtp:[127.0.0.1]:10025): 250 2.0.0 Ok: queued as E867062BB7)
    Mar 12 10:11:51 node238 postfix/smtp[650]: 8273362CA0: to=<[email protected]>, relay=127.0.0.1[127.0.0.1]:10024, conn_use=8, delay=212, delays=8.3/191/0/13, dsn=2.0.0, status=sent (250 2.0.0 from MTA(smtp:[127.0.0.1]:10025): 250 2.0.0 Ok: queued as E867062BB7)
    Mar 12 10:11:51 node238 postfix/qmgr[3536]: 8273362CA0: removed
    """"""""""""""""""""""""""""""""""""""""""

    Same times, same message on others server, come to the server at the 10:08:22, and put to the mailbox at the 10:08:23 :-(
     
  3. till

    till Super Moderator Staff Member ISPConfig Developer

    It is possible that amavis is not able to reach all blacklist servers, so it waits until the connection times out before it finishs processing the mail. Try to run:

    sa-update -D

    and then restart amavis.

    Normally amavis neeeds 10 - 20 seconds and processes 2 emails at the same time. You can also increase the numbers of amavis processes and set the new process limit in master.cf too to process more then 2 mails at a time. Some of our customers with huge mailservers run 60 amavis processes on each cluster node.
     
  4. themark

    themark Member

    In my case solved with increase from 2 to 10 amavis concurrency istance, from master configuration file

    Thank's till to put me in the right directions.
     

Share This Page