On Wed, Nov 29, 2017 at 09:20:21AM +0530, an...@isac.gov.in wrote:
> We have a internal relay configured in our setup. The relay host hadstopped
> accepting mails as the system was hanging and we could resolve
> it only after about 8 hrs. Meanwhile the queue built up and there
> were about 80,000 mails in queue to relay to the relay host. Once the
> system was made up, mails started delivering to the relay host at a
> very slow rate. It took almost 10 hours to clear the queue.
This is almost certainly because the relay host could not process
the resumed mail flow faster. Another possibility is that there
are residual DNS issues that slow down each delivery.
> What I could understand by googling was, postfix detected the relay
> host to be dead and due to the parameters like minimal_backoff_time
> and maximal_backoff_time, the delivery rate to the relay host was slow.
This is wrong. The dead host detection state is cached only for a short
time, and cannot account for slow delivery once the host is back up.
> What is the correct way to come out of this situation and deliver the
> mails faster after recovery of relay host.
Make sure the relay host can process email quickly, find and resolve
all DNS-related issues.
To understand why delivery was slow you need to analyze your logs
and examine the "delays=" log entries to see whether the delays
in connecting to the relay host or waiting for deliveries to
complete.
--
Viktor.
