Hi,
We experienced a very strange behavior that seems contrary to the
specification (the logs indicate that). I hope someone here can
shed some light on what happened. Am I missing something here?!
Thanks in advance
and kind regards,
Dominik Schramm
Scenario:
We have two identically configured mail servers
(server1.mydomain.de, 10.16.24.97, and server2.mydomain.com,
10.16.24.196) with two queues -- incoming and outgoing -- each.
They are running the Ubuntu 7.10 exim4-daemon-heavy version 4.67-5build1.
The servers are both reachable from the Internet, they have the
same MX priority and are thus roughly equally loaded. The
outgoing daemons use a system filter to set up unseen deliveries.
Problem:
A message coming into our network should be sent on to
copyaddress@??? by an unseen delivery (see below); this
fails without any consistent error being logged, the message in
question simply disappears from the queue after being delivered
to the original recipients. Normally this works fine (it has for
six months), only yesterday did we have errors of the same kind
on both mail servers, which work indepently of each other.
The only unusual thing about yesterday was that the
destination mail servers for the domain "outside.com" were
all down between 10:55 am and 11:25 am, and errors were logged
accordingly ("Connection timed out"). I would have expected that
the mail message to "outside.com" remain on the queue and be
processed by subsequent queue runners according to the retry
rules. Instead, there was one mainlog message "retry time not
reached" or "defer (110): Connection timed out", and after that
"Completed", and the messages was nowhere to be found any more.
Here are two complete log examples from server1 (extracted with
exigrep, but checking the logs line by line reveals that exigrep
did not miss anything) -- rejectlog and paniclog are empty, and
the other syslog log files contain nothing interesting:
2008-06-18 10:57:25 1K8tTp-0002DV-Ir <= edi@... [4.5.6.7] P=esmtp S=6728 id=z05dafk5I8tKRu.RZmta@...
2008-06-18 10:57:28 1K8tTp-0002DV-Ir => order.copy <system-filter> R=procmail T=procmail_pipe
2008-06-18 10:57:28 1K8tTp-0002DV-Ir => order@??? R=imap_pop3_server2 T=remote_smtp H=10.16.24.196 [10.16.24.196]
2008-06-18 11:00:37 1K8tTp-0002DV-Ir mailserver-2.outside.com [9.10.11.12] Connection timed out
2008-06-18 11:03:10 1K8tTp-0002DV-Ir Spool file is locked (another process is handling this message)
2008-06-18 11:03:46 1K8tTp-0002DV-Ir mailserver-1.outside.com [9.10.11.11] Connection timed out
2008-06-18 11:03:46 1K8tTp-0002DV-Ir == copyaddress@??? <system-filter> R=dnslookup T=remote_smtp defer (110): Connection timed out
2008-06-18 11:05:01 1K8tTp-0002DV-Ir Completed
2008-06-18 11:10:00 1K8tg0-00038u-BU <= nospam@... [1.2.3.4] P=esmtp S=4364
2008-06-18 11:10:01 1K8tg0-00038u-BU => order.copy <system-filter> R=procmail T=procmail_pipe
2008-06-18 11:10:01 1K8tg0-00038u-BU == copyaddress@??? <system-filter> R=dnslookup T=remote_smtp defer (-53): retry time not reached for any host
2008-06-18 11:10:01 1K8tg0-00038u-BU => order@??? R=imap_pop3_server2 T=remote_smtp H=10.16.24.196 [10.16.24.196]
2008-06-18 11:15:01 1K8tg0-00038u-BU Completed
Configuration details:
root@server2:~# cat /etc/exim4/exim4_outgoing.conf
...
system_filter = /etc/exim4/cfg_system_filter.txt
...
root@server2:~# cat /etc/exim4/cfg_system_filter.txt
# Exim filter
if error_message or not first_delivery
or $sender_host_address is "10.16.24.97"
or $sender_host_address is "10.16.24.196"
then
finish
endif
if foranyaddress $recipients ( $thisaddress is order@???
or $thisaddress is order.2@??? )
then
unseen deliver order.copy@??? errors_to it-operations@???
endif
if foranyaddress $recipients ( $thisaddress is order@??? )
then
unseen deliver copyaddress@??? errors_to it-operations@???
endif
