[Pacemaker] Cleanup on Master/Slave DRBD resource changes, the roles

[Georgios Kasapoglou]

Hi Andreas,
I've tried your proposal but unfortunately the behavior is the same.
However, this procedure was on a test environment with two
plain x86 servers, while I was waiting my x86_64 servers.
My final installation will be on x86_64, so I've received
the new servers a couple days ago, and repeated the whole
installation on my new environment. The steps were exactly
the same, I've cross checked all configurations and are
exactly the same. The only difference (apart from HW) is
that my test environment is with x86 SW, while the new one
is with x86_64 SW.
The outcome is that I'm not facing this problem anymore on
my new environment. More specific, I'm running "cleanup"
on Master resource of drbd_r0, the resource stops and starts
again but in role Master as it should be. I've tried this
a lot of times in order to be sure, and not even once have
the roles swapped.

For me, there is no issue anymore since the behavior is
"normal" on my target environment. But maybe some people
with x86 servers come across it.

Thanks again for your time and effort.

Best regards,
Georgios

----------------------------------------------------------------------

Message: 1
Date: Wed, 07 Dec 2011 02:15:45 +0100
From: Andreas Kurz<andreas at hastexo.com>
To:pacemaker at oss.clusterlabs.org
Subject: Re: [Pacemaker] Cleanup on Master/Slave DRBD resource
	changes, the roles
Message-ID:<4EDEBE41.9090402 at hastexo.com>
Content-Type: text/plain; charset="iso-8859-1"

Hello  Georgios,

sorry for the late reply ... overlooked your reply ...

On 12/02/2011 12:19 PM, Georgios Kasapoglou wrote:

> Hello Andreas, first of all, thanks for the quick reply. I've noticed
> that stop and start is executed on any resource when I'm running cleanup
> on. Which also affects all other resources when it is in an "order".
> Please find attached the configuration. Regards, Georgios
is the behaviour different if you remove that resource-set and use
simple colocation ... and remove the target-role from drbd_master:

ms drbd_master drbd_r0 \
	meta master-max="1" master-node-max="1" clone-max="2"
clone-node-max="1" notify="true"
colocation coloc1 inf: unims_apps drbd_master:Master

... if not, please send your complete cib ... "cibadmin -Q"


Regards,
Andreas

-- Need help with Pacemaker? http://www.hastexo.com/now

> ------------------------------ Message: 4 Date: Fri, 02 Dec 2011
> 11:42:11 +0100 From: Andreas Kurz<andreas at hastexo.com>  To:
> pacemaker at oss.clusterlabs.org  Subject: Re: [Pacemaker] Cleanup on
> Master/Slave DRBD resource changes the roles Message-ID:
> <4ED8AB83.5050604 at hastexo.com>  Content-Type: text/plain;
> charset="iso-8859-1" Hello Georgios, On 12/02/2011 10:54 AM, Georgios
> Kasapoglou wrote:
>
>> Hi all,
>> I've a 2-nodes cluster using pacemaker 1-2.
>> I've set a DRBD resource, according to
>> http://www.clusterlabs.org/wiki/PostgresHowto#4._Configuring_DRBD
>>
>> Everything works fine, except when I'm trying to cleanup the "master".
>> E.g., when I ran cleanup on drbd_r0:0 which runs on node1 while it is
>> Master, then it will change to Slave
>> and drbd_r0:1 on node2 will become Master.
> please show us your complete Pacemaker configuration
>
>
>> I've checked the /var/log/messages and saw the message below:
>> r0 already Primary, demoting.
>>
>> I found that  drbd_start function in
>> "/usr/lib/ocf/resource.d/linbit/drbd" is responsible for this.
>>
>> I've searched in the documentation, how cleanup works but found nothing.
>> Maybe I have not searched in the
>> right documents. I suspect that it calls stop and start.
>> Is that true?
> cleanup removes all information from the status section of the resource
> and triggers a reprobe followed by any other action that brings the
> resource into the desired state.
>
> Regards,
> Andreas
>
>
>
> _______________________________________________
> 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
> -------------- next part --------------
> A non-text attachment was scrubbed...
> Name: signature.asc
> Type: application/pgp-signature
> Size: 286 bytes
> Desc: OpenPGP digital signature
> URL:<http://oss.clusterlabs.org/pipermail/pacemaker/attachments/20111207/4fea848d/attachment-0001.sig>
>
> ------------------------------
> -- 
> Georgios Kasapoglou
> Telecom and Systems Engineer
> Operation/Business Support Systems
>
> INTRACOM TELECOM
> 19.7 km Markopoulou Ave.
> 19002 Peania, Athens, Greece
> tel: +30 210 667 7667
> fax: +30 210 667 7137
> e-mail: gkasap at intracom.com
> www.intracom-telecom.com
>