On Mar 14, 2008, at 8:57 PM, Todd Lyons wrote:
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
On Thu, Mar 13, 2008 at 08:21:20PM -0500, Michael Brennen wrote:
We have a two node v2 heartbeat setup. We have in both ha.cf files:
auto_failback off
If the services are running on node2 and it's the dc, we can tell
node2 to
go into standby in hb_gui. The services migrate to node1 and
everything
is fine. We tell node2 to go active again and all the services
migrate
back to node1.
If you are running crm2, with the graphic interface, I think
auto_failback
in ha.cf is ignored. Failback rules must be configured in the xml.
That makes sense. In the absence of any locations, orders, or
colocations, I expected services to stay put on whatever machine they
are currently running on, by my experiments are telling me that it
wants
to migrate towards the dc if it's active. Does that sound correct?
Nope.
The location of the DC is irrelevant to resource placement (any pretty
much everything else). Its just where the decisions happen.
In the absence of any other information, the cluster will pick a
location (seemingly) at random, so its a good idea to add some
(co)location constraints
_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems
