Edward Raymond Kryda wrote:
>
> I had a message that took around 19 hours to actually get delivered.
> Anyone have any ideas why? Here's the intitial log entry when it was
> recieved:
> Feb 2 07:45:20 sys16 exim[30853]: 2006-02-02 07:45:20 1F4dpu-00081d-2h
> <= testuser@??? U=testuser P=local S=6343
> id=200602021245.HAA21997@???
> Feb 2 07:45:20 sys16 exim[30853]: 2006-02-02 07:45:20 1F4dpu-00081d-2h
> no immediate delivery: load average 21.99
> The load average went back down around 7:46:13 and there were no other
> non-delivery messsages because of load average after that.
> Here's the actual delivery log messages:
> Feb 3 03:36:25 sys16 exim[8636]: 2006-02-03 03:36:25 1F4dpu-00081d-2h
> => testuser@??? R=dnslookup T=remote_smtp
> H=link.acns.msu.edu [35.8.2.138]
> Feb 3 03:36:25 sys16 exim[8636]: 2006-02-03 03:36:25 1F4dpu-00081d-2h
> Completed
> I didn't see any "retry time not reached for any host" messages which
> seems odd.
> Some of the valid exim config settings that I'm using that might be
> helpful:
First, temporarily enhance your logging and look again:
log_selector = +all
That will even show and track the TCP connections, IP's ports,
etc. Makes it easier to grep usefully.
- add a 'warn' verb with 'logwrite' (do if traversed) and
log_message (do only if actioned) to the acl's with an ID code,
so as to get a time-stamped audit trail in ~/mainlog and see
what you are doing, with what, when, and to whom.
- also makes editing faster: "Find <questionable acl ID code>"
Comment these out after you have solved the problem.....
Second, 'valid' these settings may be, taken one-at-a-time.
But the combination of settings you use strikes me as working at
cross-puposes in a couple of places..
What every pilot hates to hear: "....what are your intentions"?
> ignore_bounce_errors_after = 1d
> timeout_frozen_after = 3d
> auto_thaw = 1h
> smtp_accept_queue_per_connection = 70
> smtp_accept_max = 256
> smtp_accept_max_per_host = 10
> smtp_connect_backlog = 50
> return_size_limit = 10K
> helo_accept_junk_hosts = *
> deliver_queue_load_max = 20
> queue_only_load = 20
> # max msg size
> message_size_limit = 50M
> # recieve timeout
> smtp_receive_timeout = 60s
> #retry configuration
> *@+real_local_domains * F,2h,15m; F,8h,30m; F,7d,60m;
> F,14d,2h
> *@* refused_A F,2h,15m; G,16h,30m,1.5
> *@* timeout_connect_A F,2h,15m; G,16h,30m,1.5
> *@* * F,2h,15m; G,16h,30m,1.5; F,5d,8h
> Oh, and lastly exim is run using this command line:
> /usr/sbin/exim -bd -q30m
That is a last resort' if/as/when you are NOT in 'queue_only'
mode. Otherwise, each connection will invoke a queue run on
finalizing a usable router/transport set.
(Temporarily) enhanced logging and tail -f with grep/exigrep are
your friends....
Bill Hacker
