[Pkg-postgresql-public] Bug#759725: Acknowledgement (postgresql-common: non-synchronous service postgresql start/stop/reload)

[Christoph Berg]

Re: beuc at debian.org 2014-09-08 <20140908182249.GA25490 at mail.beuc.net>
> Can you share a bit your motivations for decreasing the severity of
> this bug?
> 
> I don't think FusionForge is the only package that will be kicked in
> the balls when installed on a brand new, systemd-enabled Debian 8
> box next year :(
> 
> Is that asynchronous behavior Debian-specific?

Hi Sylvain, hi Richard,

frankly I was too fast with the severity change, sorry. This problem
needs to be fixed for Jessie.

It's insofar Debian-specific as stock PostgreSQL doesn't have
postgresql-common.


Re: Richard Hughes 2014-09-15 <1394146913.20140915225826 at rhughes.net>
> A service which depends on postgresql (using the recommended
> Requires=/After= pair) is also affected by this, because the service
> cannot immediately connect to Postgres at startup.
> 
> This patch seems to make start/stop work for me, although it doesn't
> do anything about reload:
> 
> --- a/postgresql at .service 2014-08-11 14:40:56.000000000 +0100
> +++ b/postgresql at .service 2014-09-09 15:44:32.489366245 +0100
> @@ -8,10 +8,12 @@
>  ConditionPathExists=/etc/postgresql/%I/postgresql.conf
>  PartOf=postgresql.service
>  ReloadPropagatedFrom=postgresql.service
> +Before=postgresql.service
> 
>  [Service]
>  # @: use "postgresql@%i" as process name
> -ExecStart=@/usr/bin/pg_ctlcluster postgresql@%i --foreground --stdlog %i start
> +Type=forking
> +ExecStart=@/usr/bin/pg_ctlcluster postgresql@%i --stdlog %i start
>  ExecStop=/usr/bin/pg_ctlcluster -m fast %i stop
>  ExecReload=/usr/bin/pg_ctlcluster %i reload
>  PIDFile=/var/run/postgresql/%i.pid

Thanks for the patch, applied. (Without --stdlog as that's only useful
with --foreground.)

Christoph
-- 
cb at df7cb.de | http://www.df7cb.de/