[ClusterLabs] Antw: Re: Reusing resource set in multiple constraints

Ulrich Windl Ulrich.Windl at rz.uni-regensburg.de
Mon Jul 29 03:39:37 EDT 2019

I think the problem is XML as a descriptive language describing the
configuration _state_. What you are asking for is a description for a
configuration _change_.
IMHO that should not be part of the CIB.

Generally I'm using this manual procedure when adding a new resource with
similar behavior as an existing one:
Look up the name of the similar resource in the existing configuration.
Then add the new resource to all the places where the similar resource is
Of course that requires that "iteraltion zero" (= the initial configuration)
is configured correctly right from the start.

I could imagine automating that procedure by writing some template in a
language still unknown to me...


>>> Andrei Borzenkov <arvidjaar at gmail.com> schrieb am 28.07.2019 um 13:09 in
Nachricht <84d21bbe-e415-55aa-ecda-b7b0cc9aae2b at gmail.com>:
> 27.07.2019 11:04, Andrei Borzenkov пишет:
>> Is it possible to have single definition of resource set that is later
>> references in order and location constraints? All syntax in
>> documentation or crmsh presumes inline set definition in location or
>> order statement.
>> In this particular case there will be set of filesystems that need to be
>> colocated and ordered against other resources; these filesystems will be
>> extended over time and I would like to add new definition in just one
> I tried to play with id-refs, but it is not really usable.
> 1. It is impossible to define "standalone" resource set - set definition
> is accepted only as part of constraint definition.
> 2. It is impossible to change set attributes (like sequential) when
> using id-ref. Because attributes in order and colocation constraints
> have different semantic, it is quite possible that for the same set of
> resources different "sequential" value is needed in both.
> 3. crmsh does not support id-refs at all. It neither allows defining set
> id (it auto-generates names), nor understand id-ref at all. Because
> auto-generated names are position dependent, any change in configured
> sets would invalidate any reference to them.
> Need to make changes in several places obviously does not facilitate
> fool proof management (or requires building some additional layer on top).
> _______________________________________________
> Manage your subscription:
> https://lists.clusterlabs.org/mailman/listinfo/users 
> ClusterLabs home: https://www.clusterlabs.org/ 

More information about the Users mailing list