[Linux-HA] Don't want automatic back-after-failover
vchepkov at gmail.com
Wed Jun 16 06:33:15 MDT 2010
On Jun 16, 2010, at 7:21 AM, Alain.Moulle wrote:
> I have two nodes in HA with Pacemaker/corosync, with resources
> on node1 and resources on node2.
> Once node1 fails, resources from node1 are failovered successfully on
> node 2. Then, after reboot of node1, when starting Pacemaker, the previously
> failovered resources are coming back automatically on node1.
> Note that I don't have change the resource-stickiness, so its value is
> the default one : INFINITY.
> Is there a parameter to avoid this automatic back-after-failover ?
> I was thinking to pass a order migrate order on node2 before re-starting
> pacemaker on node1 (just to add the cli location preference in cib, even
> if resources are already migrated on node2 by the failure of node1)
> but perhaps I 've missed a simpler way to maintain resources on the node
> where they currently are ?
The default stickiness is 0, not INFINITY.
There is no reason to set it to INFINITY either, just to some positive number,
otherwise you wouldn't be able to put some additional placements checks, like pingd, for example.
More information about the Linux-HA