Re: [exim] Sending mail fails (using TLS via company server)…

On Mon, Aug 17, 2015 at 7:20 PM, Jeremy Harris <jgh@???> wrote:
> So it's delivered a message, and is searching the
> queue for others suitable to go down the same
> connection. Somehow, one of the sender-addresses is
> null.
>

Thank you very much, with your help I was able to track this down
further. The crashing process looks for non-existing spool files in
the queue (email-addresses have been anonymized):

2015-08-17 12:06:03 1ZRHIo-0007ZK-V3 <= sender@???
2015-08-17 12:06:05 1ZRHIo-0007ZK-V3 Spool file 1ZQR7Z-0007Zm-Sr-D not found
2015-08-17 12:06:05 1ZRHIo-0007ZK-V3 == recipient@???
R=dnslookup T=remote_smtp defer (-1): smtp transport process returned
non-zero status 0x008b: terminated by signal 11
2015-08-17 12:06:05 1ZRHIo-0007ZK-V3 Frozen

Message 1ZQR7Z-0007Zm-Sr had temp delivery errors but finally got
delivered two days (!) ago:

2015-08-15 04:52:04 1ZQR7Z-0007Zm-Sr == recipient@???
R=dnslookup T=remote_smtp defer (-53): retry time not reached for any
host
2015-08-15 05:01:37 1ZQRiy-0005Tx-Vm <= <> R=1ZQR7Z-0007Zm-Sr U=exim
P=local S=3131
2015-08-15 05:01:37 1ZQR7Z-0007Zm-Sr Completed

For some magic reason the message ID seems to be still in the
wait-remote queue ('grep' is definitely not the right tool here, but I
hope it's ok for a quick test):

# grep 1ZQR7Z-0007Zm-Sr /var/spool/exim/db/*
Binary file /var/spool/exim/db/wait-remote_smtp matches

Now I can reproducibly crash exim-4.86 by sending emails to this
domain which has had delivery errors in the past.

When I used exim-4.84 after that, there was no 'Spool file
1ZQR7Z-0007Zm-Sr-D not found' message in the logs and the email just
got delivered. After that, the wait-remote-queue had been 'repaired'
(grep showed no match anymore). Now I can use 4.86 again to send
emails to this domain.

Summary:
1.) exim-4.84 seems to be more resilient against non-existing
message-IDs in the wait-queue
2.) why there are non-existing message-IDs in the wait-queue at all?

Chris