[Linux-HA] What is the proper way to order NFS & IP resources

Andrew Beekhof beekhof at gmail.com
Wed Feb 14 02:02:41 MST 2007


On 2/14/07, Jim Wong <jwong at sharpcast.com> wrote:
> All,
>
> Two questions:
>
> 1) If we only have a single, uni-directional co-location constraint,
> will the constrained resources not move together in the event one or
> another fails (or if we try to migrate one of them)?

if you had:
   rsc_colocate from=A to=B

then A follows B around the cluster.
if A cant run where B is, then IIRC in 2.0.8, A would just be stopped.

in the current development version (which is pretty stable if you want
to try it - stable enough to be in the next SLES service pack):
* A can pass on its location preferences to B
* One can also set symmetrical=true on a colocation constraint to get
the pre-2.0.8 behavior
* It wont complain when you set symmetrical=true

>
> 2) What if we're using 2.0.7?

In .7 they would move around together but if one couldn't run, then
nothing that it was collocated with could either.

>
> Thanks...
>
> -----Original Message-----
>
> in 2.0.8, if you put (both):
>   rsc_colocate from=A to=B
>   rsc_colocate from=B to=A
>
> then it will do the right thing regardless of which resource it is that
> must be moved.
> it may complain a bit/lot^, but it should do the right thing.
>
> ^ IIRC the warnings were removed after 2.0.8 went out
> _______________________________________________
> Linux-HA mailing list
> Linux-HA at 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 at lists.linux-ha.org
> http://lists.linux-ha.org/mailman/listinfo/linux-ha
> See also: http://linux-ha.org/ReportingProblems
>


More information about the Linux-HA mailing list