[Pacemaker] How are Scores computed regarding roles?

Andrew Beekhof andrew at beekhof.net
Thu Feb 7 01:20:13 EST 2013


On Wed, Feb 6, 2013 at 10:09 AM, Jake Smith <jsmith at argotec.com> wrote:
>
> ----- Original Message -----
>> From: "Marcos Tiago" <Marcos.Tiago at openet.com>
>> To: pacemaker at oss.clusterlabs.org
>> Sent: Tuesday, February 5, 2013 3:11:13 PM
>> Subject: [Pacemaker] How are Scores computed regarding roles?
>>
>>
>>
>>
>>
>> Guys,
>>
>>
>>
>> I am having trouble configuring a collocation in crm (corosync), but
>> I am not sure if what I want is possible. This is the scenario:
>>
>>
>>
>> Let's say it's a 2 node cluster, with one resource in each. RES1 is
>> with location inf in Node1, and RES2 is with location inf in Node2.
>> Both resources are in Started role.
>>
>>
>>
>> I have a third resource (IP address) that has: colocation 20 with
>> RES1 and colocation 10 with RES2, so that if RES1 fails, it moves to
>> node2.
>>
>>
>>
>> When RES1 is restarted after a failure, the IP resource moves to Node
>> 1, as expected.
>>
>>
>>
>> But the IP is moving to Node 1 * before * RES1 is in Started state,
>> which is not good for my situation because RES1 is not ready to
>> receive requests yet.
>>
>>
>>
>> I would like the IP to move back to Node 1 only when RES1 is in
>> Started role.
>>
>>
>>
>> In other words, I would like the score associated with RES1 to be
>> computed only when RES1 is Started.
>>
>>
>>
>> Is this even possible? If it is, how? If it’s not, then how could I
>> achieve such behavior?
>>
>
> There may be a cleaner way but one option would be to create two Dummy resources - DUM1 DUM2.
> Order RES1 then DUM1, collocate DUM1 with RES1.
> Do the same for DUM2.
> Now collocate your IP with the dummy resources instead of RES.
> Dummy won't be started until RES is started because of the order statement.  Then IP will move once Dummy has been placed.

I'd be surprised if this worked.
Collocation constraints only stop the IP from running if we there is
nowhere for DUM1/2 to be active.
At the point we _find_ somewhere to start them, the IP would be
allowed to start too.  The actual starting of DUM1/2 doesn't come into
it.

The only thing that controls start order is... ordering constraints.
You probably want optional ones between the IP and RES1, and the IP and RES2.



>
> Kinda ugly but I think it would be effective.
>
> HTH
>
> Jake
>
>>
>>
>> Thanks!
>>
>> Marcos
>>
>> Sao Paulo – Brazil
>>
>>
>>
>>
>>
>> This email and any files transmitted with it are confidential and
>> intended solely for the use of the individual or entity to whom they
>> are addressed. If you are not the intended recipient, please note
>> that any review, dissemination, disclosure, alteration, printing,
>> circulation, retention or transmission of this e-mail and/or any
>> file or attachment transmitted with it, is prohibited and may be
>> unlawful. If you have received this e-mail or any file or attachment
>> transmitted with it in error please notify postmaster at openet.com.
>> Although Openet has taken reasonable precautions to ensure no
>> viruses are present in this email, we cannot accept responsibility
>> for any loss or damage arising from the use of this email or
>> attachments.
>>
>> _______________________________________________
>> Pacemaker mailing list: Pacemaker at oss.clusterlabs.org
>> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>>
>> Project Home: http://www.clusterlabs.org
>> Getting started:
>> http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
>> Bugs: http://bugs.clusterlabs.org
>>
>
> _______________________________________________
> Pacemaker mailing list: Pacemaker at oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>
> Project Home: http://www.clusterlabs.org
> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
> Bugs: http://bugs.clusterlabs.org




More information about the Pacemaker mailing list