On 14/12/17 17:25 +0100, Jan Pokorný wrote: > Anyway, the change is seemingly straightfoward, but few things > should be answered/investigated first: > - After=dbus.service or rather After=dbus.socket (or both)?
In theory, dbus.socket would be more flexible should anyone want to redirect it to a different DBus implementation via Service=. This may be the reason why systemd-logind.service (and only that on my system) uses that instead. > - mere After=, or combined with other relationship modifying > directives > - Wants=, Requires= ... what's the expected behaviour when > DBus crashes (or is it merely hypothetical because then > whole systemd is in trouble then?) Note that systemd (systemctl) continues to work just fine on a headless system when dbus.service is stopped (and no, it didn't immediately got started again in my testing). Furthermore, DBus gets reactivated automatically via said dbus.socket activation that happens (unless the whole system is going to shutdown at that very moment) upon initiating a connection (which is what pacemaker seems to be doing when disconnected state is spotted), so we should definitely be fine just with After= for the purpose of not having it terminated before pacemaker finishes its own termination sequence first. > - should DBus reloads somehow affect pacemaker? If that reload was to roll out a hard bump of org.freedesktop.systemd1 to org.freedesktop.systemd2, we are currently lost anyway, otherwise the compatibility should be preserved (http://0pointer.de/blog/projects/versioning-dbus.html) meaning it's not actionable for us regardless. > - (insert some other questions that are not currently > coming to mind of rather inexpert person) -- Jan (Poki)
pgp3mZCmlALLR.pgp
Description: PGP signature
_______________________________________________ Users mailing list: Users@clusterlabs.org http://lists.clusterlabs.org/mailman/listinfo/users Project Home: http://www.clusterlabs.org Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf Bugs: http://bugs.clusterlabs.org