[Pacemaker] adding a third node for quorum without running resource?

[Andrew Beekhof]

On Mon, Apr 18, 2011 at 10:35 AM, Jelle de Jong
<jelledejong at powercraft.nl> wrote:
> Dear Andrew,
>
> On 18-04-11 09:05, Andrew Beekhof wrote:
>> On Fri, Apr 15, 2011 at 9:54 PM, Jelle de Jong
>> <jelledejong at powercraft.nl> wrote:
>>> How should I change my configuration[1] so it understands that it can
>>> not run resources on the third node (machine ebony) but should use it
>>> for quorum only?
>>
>> node thirdnode \
>>         attributes standby="on"
>
> http://pastebin.com/EwqmZxjc
>
> Thank you for taking the time to help-out. This is much appreciated. I
> added the third-node ebony to the crm configuration and started corosync
> with the crm service on the node ebony. The other host saw the added
> node in standby node.
>
> I tried to move the main group resource, but it behaves the same as
> before the complete resources is failing.

This is unrelated to the third node being in standby.

    drbd_r3:0_demote_0 (node=godfrey, call=299, rc=6,
status=complete): not configured
    drbd_r2:0_demote_0 (node=godfrey, call=298, rc=6,
status=complete): not configured
    drbd_r1:0_demote_0 (node=godfrey, call=297, rc=6,
status=complete): not configured
    drbd_r0:0_demote_0 (node=godfrey, call=296, rc=6,
status=complete): not configured

For some reason drbd thinks its incorrectly configured.
Maybe the linbit guys can suggest something - but they'd likely want
to see some logs.

> I expect because the drbd
> monitor needes the third node to do somehting that it can't do.
>
> If I am still doing something wrong please let me know. I will now try
> to add a ping monitor to the configuration to detect and countermeasure
> short network failures.
>
> Thanks in advance,
>
> Kind regards,
>
> Jelle de Jong
>