On systems where /var/run is on tmpfs, the postgresql-server init script fails to start the postgres server because it does not recreate /var/run/postgresql if it is missing. I have /var/run on a tmpfs to save writes on my SSD. Please consider adding the following lines to the checkconfig() function in the init script. Thanks ! if [ ! -d "/var/run/postgresql" ] ; then mkdir /var/run/postgresql fi chown postgres:postgres /var/run/postgresql Reproducible: Always
Even better: checkpath -d -m 0770 -o postgres:postgres /var/run/postgresql This is the only thing needed to get this to work. And please consider that with the idea of migrating /run, this starts to be more a common situation than you could ignore.
(In reply to comment #1) > Even better: > > checkpath -d -m 0770 -o postgres:postgres /var/run/postgresql > > This is the only thing needed to get this to work. And please consider that > with the idea of migrating /run, this starts to be more a common situation than > you could ignore. Yeah, that's what I'm going to go with now. I just saw that this morning.
08 Dec 2011; Aaron W. Swenson <titanofold@gentoo.org> +postgresql-server-8.2.23.ebuild, +postgresql-server-8.3.17.ebuild, +postgresql-server-8.4.10.ebuild, +postgresql-server-9.0.6.ebuild, +postgresql-server-9.1.2.ebuild: Version bump. Fixes bugs 391851, 383471, and 378865.