Wietse:
On 5/13/2013 5:29 PM, Wietse Venema wrote:
Yes, at the time of each incident, there are a few threads that
eventually time out and throw a few emails into the deferred queue. That
does not concern me. What concerns me is that while Postfix is waiting
for these few threads to time out, the active queue is completely
ignored and is growing rapidly.
The queue manager WILL NOT READ THE QUEUE when the maximum
number of processes is delivering mail,
Understood. During these incidents, we are no where close to the
maximum number of processes.
or when the maximum
concurrency per destination is reached.
So while those processes are timing out talking to a broken device,
the queue manager WILL NOT READ THE QUEUE.
In the cases where this has happened, the processes that are talking to
the broken device/destination are only utilizing a handful of smtp
processes, during which time, the active queue is building up a huge
backlog with messages that *do* have deliverable destinations.
After this small handful of processes finally time out, that is when the
flood gates open, and Postfix again starts delivering successfully
again... it is then that it starts using 110 processes in an effort to
catch up. I am trying to determine what is causing Postfix to wait
until every last process times out before it starts spawning new threads.
I'm out of time today. Hopefully I'll have time in the morning to
respond to the list of questions from Viktor Dukhovni... perhaps in the
process of answering his questions I'll stumble into what's causing my
issue.
Thanks for your efforts... they are appreciated.
Curtis
Is this clear now?
Wietse
