[Pacemaker] colocation that doesn't

[Alan Jones]

I was thinking along the lines of your second suggestion.
It involves teaching my "external source" about relationships between
resources which I don't like but is possible.
A phrase from the '08 election cycle involving lipstick comes to mind ... ;)
Thanks for following up Tim!
Alan

On Mon, Nov 29, 2010 at 7:16 PM, Tim Serong <tserong at novell.com> wrote:
> On 11/30/2010 at 10:11 AM, Alan Jones <falancluster at gmail.com> wrote:
>> On Thu, Nov 25, 2010 at 6:32 AM, Tim Serong <tserong at novell.com> wrote:
>> > Can you elaborate on why you want this particular behaviour?  Maybe
>> > there's some other way to approach the problem?
>>
>> I have explained the issue as clearly as I know how.  The problem is
>> fundamental
>> to the design of the policy engine in Pacemaker.  It performs only two
>> passes to
>> resolve constraints, when what is required for general purpose
>> constraint resolution
>> is an iterative model.  These problems have been addressed in the literature
>> for
>> decades.
>
> What I meant by "maybe there's some other way to approach the problem" is
> "maybe there's some other way we can figure out how to get something *like*
> the behaviour you desire, given the fact that Pacemaker's colocation
> constraints behave the way they do".
>
> If you have:
>
>  primitive resX ocf:pacemaker:Dummy
>  primitive resY ocf:pacemaker:Dummy
>  location resX-loc resX 1: nodeA.acme.com
>  location resY-loc resY 1: nodeB.acme.com
>  colocation resX-resY -2: resX resY
>
> And you have "-inf constraints coming from an external source", as you
> said before, can you change the external source so that it generates
> different constraints?
>
> e.g., instead of generating either of:
>
>  location resX-nodeA resX -inf: nodeA.acme.com
>  location resY-nodeB resY -inf: nodeB.acme.com
>
> (where only the second one works, because of the dependency inherent
> in the colocation contraint) can your external source specify these
> constraints only in terms of resY, which is the one that's capable of
> "dragging" resX around the place?  e.g.:
>
>  location resX-nodeA resY inf: nodeA.acme.com
>  location resY-nodeB resY -inf: nodeB.acme.com
>
> Or, if that sounds completely deranged, how about this:
>
> On the assumption your "external source" will only ever inject one
> -inf rule, for one resource, why not make it change the colocation
> constraint as well?  e.g.: generate either of:
>
>  location resX-nodeA resX -inf: nodeA.acme.com
>  colocation resY-resX -2: resY resX
>  (and delete resX-resY if present)
>
>  -- or --
>
>  location resY-nodeB resY -inf: nodeB.acme.com
>  colocation resX-resY -2: resX resY
>  (and delete resY-resX if present)
>
> Are there any more details about your application you can share?
>
> Regards,
>
> Tim
>
>
> --
> Tim Serong <tserong at novell.com>
> Senior Clustering Engineer, OPS Engineering, Novell Inc.
>
>
>
>
> _______________________________________________
> Pacemaker mailing list: Pacemaker at oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>
> Project Home: http://www.clusterlabs.org
> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
> Bugs: http://developerbugs.linux-foundation.org/enter_bug.cgi?product=Pacemaker
>