From: Pavel Urban (urbanpmlp.cz)
Date: Fri Jul 20 2007 - 14:52:35 CDT

• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

Victor Duchovni wrote:
> On Fri, Jul 20, 2007 at 07:53:55PM +0200, Pavel Urban wrote:
>
>> Hello,
>>
>> I've tried to reinstall one of our more busy servers. It is running
>> RedHat Linux 4, x86_64 arch, Postfix 2.4.3. I've noticed quite
>> unpleasant behavior.
>>
>> At first, incoming queue got very large (about 300.000 mails). Active
>> queue has been hovewer very small, about 200 mails. During this time,
>> system load has been quite low. After several hours, incoming queue
>> slowly drained to 20000 - and suddenly, it quickly poured into active
>> queue and has been processed. Load jumped up, but wasn't critical.
>>
>> What did I wrong? I've upgraded from an older version (2.2.10) and
>> didn't change the configuration except for fallback_relay ->
>> smtp_fallback_relay and some master.cf modification ('retry' transport).
>> I just cannot find a reason for such behavior...
>>
>
> Clock rolled back? Messages from the incoming queue are not read until
> their modification time is in the past...
>

Most probably, no. I'm using ntpdate as an hourly cronjob, but I'll
double-check it again. Thanks for the tip.

--
***********************************************************************
Pavel Urban (pavel.urbano2.com)
O2 system disaster
Telefonica O2 Czech Republic, a.s. - www.cz.o2.com
***********************************************************************
    Vegetables should not operate electronic equipment.
           Computer Stupidities, http://rinkworks.com/stupid/
***********************************************************************

• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]