On Wed, Oct 29, 2014 at 11:17:40AM +0100, Ulrich Windl wrote: > >>> Dejan Muhamedagic <deja...@fastmail.fm> schrieb am 29.10.2014 um 10:49 in > Nachricht <20141029094927.GA32415@walrus.homenet>: > [...] > > Even though nodes may have ids different from their names, people > > are prone to reference them by name and not by some number such > > as 167906357. Well, we can't really blame them, can we. So, > > the node uname is effectively its name, for all intents and > > purposes. The fact that some programs need an unwieldy number to > > reference nodes do not change our perspective. > > > >> This also breaks syntax of some existing commands, as Dejan > >> says, f.e. > > > > It doesn't break the syntax, it just makes it impossible to > > reference CIB elements. If you have a node named xyz and then > > also a resource named xyz, how should a stupid program such as > > crmsh know what do you mean when you say "xyz"... > > > > I wonder whether "prefix tags" could help here: When names are not unique in > a context the user could use the prefix "host:name" to mean the host named > "name" and "rsc:r" to mean resource "r" (Or a tag with la less ugly, but > longer name).
Yes, but I hope we won't need to go there. > Isn't a similar mechnism already used wthen specify ing the RA? If you mean [class:provider:]name, then yes. However, that doesn't help us here much, as the whole thing happens on a different level. Note that the element names (or IDs) are the basic information. Any changes there are probably going to have quite some impact on crmsh. Thanks, Dejan > Regards, > Ulrich > > > _______________________________________________ > Linux-HA mailing list > Linux-HA@lists.linux-ha.org > http://lists.linux-ha.org/mailman/listinfo/linux-ha > See also: http://linux-ha.org/ReportingProblems _______________________________________________ Linux-HA mailing list Linux-HA@lists.linux-ha.org http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems