>>> Dejan Muhamedagic <deja...@fastmail.fm> schrieb am 09.02.2016 um 20:58 in Nachricht <20160209195816.GD2437@walrus.homenet>: [...] >> Particularly, imho RAs should not run validate_all on stop >> action. > > I'd disagree here. If the environment is no good (bad > installation, missing configuration and similar), then the stop > operation probably won't do much good. Ultimately, it may depend [...]
Hi! I think whether to use validate on start/stop/monitor is implementation defined, but obviously with a illegal configuration start/stop/monitor can't be expected to to something reasonable. For an implementer of an RA it's simply tempting to use the code that is needed (and present, hopefully!) for validate before trying to start/stop/monitor. OTOH if validate is called on all nodes before any cluster change, there SHOULD be no need to revalidate. Exception: If some node was added after the configuration change. Another question would be whether validate should check syntax only, or also check the semantics. Questions like "will it work when I'll do that?" that reference the future are always hard to decide (even if actually trying now), however... Regards, Ulrich _______________________________________________ Users mailing list: Users@clusterlabs.org http://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