This is definitely still an issue, a serious one, and trivial to fix.
The problem is that the courier packages are basically orphaned. There
have been some new people trying to pick it up, but basically we need a
new maintainer.
On 3/6/16 12:11, Thomas Mayer wrote:
May be obsolete now, but I can confirm that some of courier's services
randomly don't get started after rebooting a virtual server
(presumably OpenVZ) after it was updated to Debian jessie and systemd.
The system also got moved to a newer host system which now provides
kernel 3.16. Besides courier, the system runs totally stable.
All courier services run fine if I start services manually after reboots:
service courier-authdaemon start
service courier-mta start
service courier-mta-ssl start
service courier-pop start
service courier-pop-ssl start
...
The system is up-to-date so it seems that the issue still needs to be
fixed (Ondřej Surý pointed that out already if I got it right).
I can confirm that adding courier-authdaemon as a requirement solves
this issue for me (tried 5 times whereas unstarted courier-services
originally were quite common at about 50% of tries).
The changed line 5 of /etc/init.d/courier-mta now reads:
# Required-Start: $remote_fs $syslog courier-authdaemon
So the fix should be small and trivial and solves a big problem which
occurs randomly while people might not even be aware of it.
I think this issue could be fixed without the necessity of larger
changes in package management.
