> -----Original Message-----
> From: [EMAIL PROTECTED] [mailto:linux-ha-
> [EMAIL PROTECTED] On Behalf Of Andrew Beekhof
> Sent: Tuesday, 3 June 2008 4:05 PM
> To: General Linux-HA mailing list
> Subject: Re: [Linux-HA] How to create a rsc_order 'OR' condition
> 
> On Tue, Jun 3, 2008 at 2:09 AM, Alex Strachan
> <[EMAIL PROTECTED]> wrote:
> > State1
> > ======
> > nodeA   resourceA   ------- resourceC
> > nodeB   resourceB
> >
> > State2
> > ======
> > nodeA   resourceA
> > nodeB   resourceB   ------- resourceC
> >
> > Rule:  ResourceC can only start AFTER (resourceA or resourceB), a
> preference
> > for resourceA is needed.
> >
> > Attempted config
> > ================
> >
> > <rsc_colocation id="colocation_resourceC_resourceA" from="resourceA"
> > to="resourceC" score="900000"/>
> >
> > <rsc_order id="order_resourceC_resourceA" from="resourceC"
> > to="resource_resourceA" action="start" type="after"
> symmetrical="false"/>
> >
> > <rsc_colocation id="colocation_resourceC_resourceB" from="resourceB"
> > to="resourceC" score="300000"/>
> >
> > <rsc_order id="order_resourceC_resourceB" from="resourceC"
> > to="resource_resourceB" action="start" type="after"
> symmetrical="false"/>
> >
> > Due to the two rsc_order's, when I put nodeB to standby - resourceC
> stops on
> > nodeA.  If I remove the rsc_order for resourceB then resourceC is
> attempted
> > to start before resourceB which is not possible.
> 
> maybe if you set score="0" for both rsc_order constraints... but then
> you'll be toast if neither A nor B can start.
> nor will C be recovered if A or B is restarted.
> 
> basically there is no way to perfectly express what you're after.

[AS] I set the score=0 for both of the rsc_order constraints - this gave me
what I was after.  resourceC is recovered if resourceA or B is restarted!

Next tasks is to calculate the resource stickiness weighting correctly.
Currently if resource C is running on node B, and resourceA is started on
nodeA then resourceC  stops then starts on nodeA.

Thank you for the score=0 idea.

> _______________________________________________
> Linux-HA mailing list
> [email protected]
> http://lists.linux-ha.org/mailman/listinfo/linux-ha
> See also: http://linux-ha.org/ReportingProblems


_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems

Reply via email to