On Tue, 19.05.15 20:13, Andrei Borzenkov (arvidj...@gmail.com) wrote: > В Tue, 19 May 2015 14:07:21 +0200 > Lennart Poettering <lenn...@poettering.net> пишет: > > > On Tue, 19.05.15 14:26, Andrei Borzenkov (arvidj...@gmail.com) wrote: > > > > > On Tue, May 19, 2015 at 2:15 PM, Uoti Urpala <uoti.urp...@pp1.inet.fi> > > > wrote: > > > > As for Evert's original problem, I think it's that RESTART is propagated > > > > to all RequiredBy units unconditionally - even if those are currently > > > > stopped! This affects both Requires= and Requisite= in exactly the same > > > > way. > > > > > > Exactly. At least I do not see anything obvious in code that would > > > check whether dependent units are running or not. And try-restart is > > > pure forntend option which is not propagated to systemd manager > > > itself. > > > > Hmm? We do have JOB_TRY_RESTART while putting together the > > transaction... What do you mean? > > > > I was referring to this comment. May be I misunderstood it. But I do not > see TRY_RESTART where dependent units are processed either. > > /* JOB_TRY_RESTART can never appear in a transaction, because > * it always collapses into JOB_RESTART or JOB_NOP before entering. > * Thus we never need to merge it with anything. */
True. The code I commited now uses job_type_collapse to fix the job type up as necessary. Lennart -- Lennart Poettering, Red Hat _______________________________________________ systemd-devel mailing list systemd-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/systemd-devel
