From: email builder (emailbuilder88yahoo.com)
Date: Sun Oct 03 2004 - 17:52:37 CDT

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

Hello,

  I recently had the pleasure of bringing a postfix server back from the brink of...
  
  The short of it is that we had an incoming queue with 250,000 messages (don't ask how
they got there). Once we fixed the flood of incoming mail, we needed to empty that out
ASAP. We couldn't really see a bottleneck per se, so we simply increased the active
queue sizing from 20,000 to 70,000. I don't fully understand why, but this was what did
it: we went from 250,000 to 30,000 in about 16 hours. Great! Once we were at 30,000, we
changed the active queue size to 20,000 again and watched it drain a little bit more
until there were still 30,000 incoming and 20,000 active. (Deferred queue is largely
empty this whole time, as is maildrop.)

  What has me completely boggled is that it leveled off at that count for about 6 hours.
It kept up with incoming messages just fine (average around 75/min which has been
consistent the whole time, draining or not), but never drained any more. Interestingly,
I watched it drop 10,000 out of the incoming queue at the end of that 6 hours, but now it
has leveled off again and is not draining.

  I find it hard to believe that we have a bottleneck somewhere, given the rate at which
we saw things drain when the incoming queue was so big. In fact, during this "level
state", our LDA (Courier Maildrop) has been largely inactive (before, we had LOTS of
maildrop processes very very busy).

  So can anyone suggest a reason why the active queue might appear to be getting lazy?
Is there anything I can do to give it a bump?

Thanks!!

                
_______________________________
Do you Yahoo!?
Declare Yourself - Register online to vote today!
http://vote.yahoo.com

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