[Pacemaker] Manually controlled cluster

[Vadym Chepkov]

On Nov 4, 2010, at 6:30 AM, Michael Schwartzkopff wrote:

> On Thursday 04 November 2010 11:23:18 Michael Schwartzkopff wrote:
>> Hi,
>> I want to create a cluster with DRBD, Filesystem, a service and an IP
>> address. Failover should only be triggered by a sys admin and not happen
>> automatically. I was wondering if I could solve this with a
>> 
>> symmetric-cluster="false"
>> 
>> property of the cluster. How does the CRM start resources on nodes? Do I
>> have to set
>> 
>> crm resource start <resource>
>> 
>> manually? Or how does it work. Thanks.
> 
> Ok, I read the 1.1 pacemaker doc section 6.2.2. This should do the trick.
> See: http://www.clusterlabs.org/doc/en-
> US/Pacemaker/1.1/html/Pacemaker_Explained/ch06s02s02.html
> 
> I can make the DRBD master run on the first node and the secondary on the 
> second node. How do I prevent the DRBD being promoted automatically on the 
> second node if the first node fails?
> 
> Thanks for any hints.
> 

Odd task, but how about something like this:

crm configure location manual resource rule -inf: not_defined runonlyhere

attrd_updater -n runonlyhere -l reboot -U because


:)

Vadym