Bug#760616: I think this bug depends on #705254 being fixed

[Robert Euhus]

Am 27.02.2015 06:23, schrieb Martin Pitt:
> This is only true under SysV init, where systemctl indeed has to ask
> update-rc.d is-enabled (or something similar). Under systemd it would
> actually be able to figure that out, as the sysv-generator checks
> exactly that -- "is the init.d script enabled". It generates
> corresponding runlevel?.target.wants/ symlinks.

>From my understanding this is not the quite problem discussed here. What
I think is happening here is the following:

I am running a jessie system, which uses systemd. Some for some daemon
(keepalived in my case) there is still a sysv init script available and
no systemd service unit (keepalived.service) available yet. So whenever
I use systemctl, it uses a kind of sysv-wrapper, which in turn uses the
old update-rc.d utility to do it's job (enable, disable).

Here is a short log excerpt:
> root at proxy-1-c30:~# systemctl enable keepalived
> Synchronizing state for keepalived.service with sysvinit using update-rc.d...
> Executing /usr/sbin/update-rc.d keepalived defaults
> Executing /usr/sbin/update-rc.d keepalived enable

But for the 'systemctl is-enabled' command there exists no debian
update-rc.d equivalent yet, so systemctl falls back to chkconfig as a
fallback, which does not exist on a debian system.

So to solve this problem two things need to be done:
1. provide an "is-enabled" option for update-rc.d (#705254)
2. provide a patch for systemd to use this on debian (would be this bug)

I don't think anyone would want to use systemctl running under a SysV
init system without systemd. At least I would not expect it to work at all.

My current workaround is to use my own keepalived.service, which I have
adapted from Arch Linux. Then using 'systemctl enable' generates the
runlevel?.target.wants/ and also invokes update-rc.d for the init script
to syncronize the state. But the init script is just ignored by systemd
as long as there is a service unit of the same name.

Robert.