Comment 21 for bug 1309617

On 26 July 2014 02:33, Steve Langasek <email address hidden> wrote:
>> In the case when plymouthd has been started from
>> the initramfs, we can be running the bridge straight
>> away, otherwise it needs to wait for plymouthd to start.
>
> I had assumed that the plymouth-upstart-bridge was designed to wait for
> plymouth to become available. Otherwise, it will lose events from
> anything that starts before plymouth does. Granted, that's very little,
> so *maybe* we don't care about this - in that case, the bridge should
> just be 'start on started plymouth'. However, otherwise the current
> start condition is correct and the daemon should simply be fixed to wait
> for plymouthd to start (probably using inotify).
>

plymouth-upstart-bridge returns exit code 1, if
ply_boot_client_connect or ply_upstart_monitor_new fail.
Plymouthd uses abstract socket, so we can't be doing inotify watch on
the socket file.

How about this instead, see patch. Add start on or started plymouth,
or started plymouth-shutdown, and add pre-start exec to ping plymouth
and stop job from running if there is no plymouthd to talk to.

--
Regards,

Dimitri.