Re: [exim] option -MCd

Top Page
Delete this message
Reply to this message
Author: Marc Haber
Date:  
To: exim-users
Subject: Re: [exim] option -MCd
On Fri, 19 Jun 2020 10:04:09 +0100, Jeremy Harris via Exim-users
<exim-users@???> wrote:
>Exim uses the system time as part of generating unique identifiers.
>To do that it waits, if needed, for the granularity of the time
>it is using for that purpose. That should be something in the
>millisecond range. To end up having to wait for six minutes
>suggests that your system time jumped backwards by that much.
>
>Exim is not designed to work in that environment. It assumes that
>time only goes forwards, and that it does move. If you are
>deliberately changing the system time backwards I suggest that
>at the very least you need to stop Exim first and restart after.
>Even doing that could result in unintended behaviour for the
>doubly-covered apparent period.


Debian has a similiar issue in the latest exim version, bug #962847,
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=962847. There, the
issue is probably related to a Linux system sent to Suspend and waking
up later.

I see the issue on my own notebook as well when delivering a message
to the local exim via SMTP to localhost. exim has the message on the
queue, but the SMTP session is hung, after "." in the data phase.
pkill -9 exim and restarting exim will have the message sent by the
next queue runner, but the calling software receives a connection
abort and will eventually resend a second copy.

A contributor on the bug report says:
|I am pretty sure that the problem is caused by the commit 6906c131d1d07d07831f8fbabae6290a3cba6ca3                      
|(Use a monotonic clock, if available, for ID generation).                                                               
|The change contains measuring of the difference between CLOCK_MONOTONIC and realtime once                               
|at startup (exim_clock_init), but as far as I understand CLOCK_MONOTONIC                                                
|on Linux does not increase during suspend/hibernate (possibly wrognly [1]),                                             
|so the difference grows then, unaccounted for.                                                                          

|                                                                                                                        

|[1]: https://stackoverflow.com/a/3527632/1236045                                                                        


Is this the possible cause of the issue showing up on at least three
Debian systems since we upgraded to exim 4.94?

Greetings
Marc
-- 
-------------------------------------- !! No courtesy copies, please !! -----
Marc Haber         |   " Questions are the         | Mailadresse im Header
Mannheim, Germany  |     Beginning of Wisdom "     | 
Nordisch by Nature | Lt. Worf, TNG "Rightful Heir" | Fon: *49 621 72739834