[Linux-HA] Failover/failback

Andrew Beekhof beekhof at gmail.com
Mon Oct 8 01:12:50 MDT 2007


On 10/7/07, Stefano Colombo <stefano.colombo at gruppocdm.it> wrote:
> I have a question regarding how to configure a group to correctly fail
> to the second node and prevent automatic failback
> I wrote an OCF agent to control vmware machines. My problems are:
> -       How can I let the administrator to manually fail the resource to
> the other node ?

later versions support crm_resource -F for failing resources but that
wont necessarily move it to another node (see
http://linux-ha.org/v2/faq/forced_failover)

to migrate it, use crm_resource -M

> -       I tried rebooting the active node and the resources correctly
> fail to the other . But when the first node got back online the
> resources automatically failback . How can I prevent it from happening ?

to prevent failback, look up the default-resource-stickiness option
  # pengine metadata

> -       When failing back I found the resource in the unmanaged state .
> What can this be due to ?

either it was already running when heartbeat started, or the status
action is broken, or the stop action is broken, or a combination of
these



More information about the Linux-HA mailing list