Bug#1031854: orphan-sysvinit-scripts: should cope with defective rsyslog-rotate

[Matthew Vernon]

On 24/02/2023 11:20, Lorenzo wrote:
> On Fri, 24 Feb 2023 09:20:29 +0000
> Matthew Vernon <matthew at debian.org> wrote:
> 
>>
>> rsyslog-rotate should HUP rsyslog as appropriate for the installed
>> init system. It doesn't, so orphan-sysvinit-scripts should work around
>> this on sysvinit systems, without causing problems on systemd systems.
>>
> 
> Isn't a conflict with systemd-sysv package enough?

No (and in general I don't want to end up with orphan-sysvinit-scripts 
conflicting with anything). rsyslog-rotate is part of the rsyslog 
package, and the problem relates to its integration with logrotate.

> As systemd starts rsyslog with '-iNONE' another approach could be to
> test for the PID file before HUP

Testing for "running init is systemd" has a standard canned answer, 
which we can use.

Regards,

Matthew