Stephan Krinetzki writes:
And then there is a logrotate: And yes, this restarts mailman at midnight. Maybe i should optimize the logrotate.
Not sure what you mean by "optimize". Because of the bug I mentioned earlier, in older versions of Mailman 3 the master process would fail to close some of logfiles. After rotation, the master process would continue writing to the still-open file handle, defeating the purpose of logrotate. Unless that bug has been fixed in your version, you should leave the restart stanza in the logrotate for configuring mailman.
That the shunt queue is collecting digests seems weird. It's not surprising that the message object in the queue file is empty, that's by design. It seems like the digest process is happening at the same time as the restart. This shouldn't be a problem, but it might clarify things if you make sure the periodic digest delivery is offset from the midnight restart in the /etc/cron.d/mailman3 file. Eg
# cron time is UTC, put stuff on desks at start of day JST # core sends mail 22 5 * * * mailman /opt/mailman3/.v/bin/mailman notify 22 6 * * * mailman /opt/mailman3/.v/bin/mailman digests --periodic
(note 22:00 UTC is 07:00 JST).
My only other comment on your mail is that I see that you do have multiple out slices. So that explains the case of the .pck.bak (in process) file being younger than a .pck file (waiting in queue). It's not unusual in your configuration.
-- GNU Mailman consultant (installation, migration, customization) Sirius Open Source https://www.siriusopensource.com/ Software systems consulting in Europe, North America, and Japan