Login | Register For Free | Help
Search for: (Advanced)

Mailing List Archive: Linux-HA: Users

Active/Active Cluster for DRBD partition

 

 

Linux-HA users RSS feed   Index | Next | Previous | View Threaded


Yount.William at menloworldwide

Jul 24, 2012, 1:21 AM

Post #1 of 3 (396 views)
Permalink
Active/Active Cluster for DRBD partition

I have two servers: KNTCLFS001 and KNTCLFS002
I have a drbd partition named nfs, on each server. They are mirrored. The mirroring works perfectly.

What I want is to serve this drbd partition up and have it so that if one server goes down, the drbd partition is still available on the other server. I am trying to do this in an active/active cluster. I have a cloned IP address that is supposed to be running on both machines at the same time.

I can get the resources setup according to the "Clusters from Scratch" guide. The issue is that as soon as I take KNTCLFS001 down, all my resources go down as well. They won't even stay running on KNTCLFS002.

Someone suggested adding constraints but that doesn't really make sense. I want the services running on all systems at once. Constraints are preconditions for bringing the service up on another machine, but the service should already be running so in theory, constraints shouldn't be necessary.

Any help will be greatly appreciated. I have attached my cib.xml to give anyone a better idea of what I am trying to do. I have been going through the "Pacemaker 1.1 Configuration Explained" but so far I haven't found anything.


William
Attachments: cib.xml (5.30 KB)


andreas at hastexo

Jul 25, 2012, 2:18 AM

Post #2 of 3 (369 views)
Permalink
Re: Active/Active Cluster for DRBD partition [In reply to]

On 07/24/2012 10:21 AM, Yount, William D wrote:
> I have two servers: KNTCLFS001 and KNTCLFS002
> I have a drbd partition named nfs, on each server. They are mirrored. The mirroring works perfectly.
>
> What I want is to serve this drbd partition up and have it so that if one server goes down, the drbd partition is still available on the other server. I am trying to do this in an active/active cluster. I have a cloned IP address that is supposed to be running on both machines at the same time.
>
> I can get the resources setup according to the "Clusters from Scratch" guide. The issue is that as soon as I take KNTCLFS001 down, all my resources go down as well. They won't even stay running on KNTCLFS002.
>

Your configuration does not look like you followed "Clusters from
Scratch" ... drbd has to be a master-slave resource and not a clone, the
filesystem and IP needs to be ordered to start after drbd promotion and
they need colocation constraints ... that is all step by step explained
in this document. And you miss STONITH, a must-have for dual-primary drbd.

Regards,
Andreas

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

> Someone suggested adding constraints but that doesn't really make sense. I want the services running on all systems at once. Constraints are preconditions for bringing the service up on another machine, but the service should already be running so in theory, constraints shouldn't be necessary.
>
> Any help will be greatly appreciated. I have attached my cib.xml to give anyone a better idea of what I am trying to do. I have been going through the "Pacemaker 1.1 Configuration Explained" but so far I haven't found anything.
>
>
> William
>
>
>
> _______________________________________________
> Linux-HA mailing list
> Linux-HA [at] lists
> http://lists.linux-ha.org/mailman/listinfo/linux-ha
> See also: http://linux-ha.org/ReportingProblems
>
Attachments: signature.asc (0.22 KB)


Yount.William at menloworldwide

Jul 25, 2012, 3:30 AM

Post #3 of 3 (365 views)
Permalink
Re: Active/Active Cluster for DRBD partition [In reply to]

I didn't miss Stonith, I just didn't want it turning my server off while I work on the configuration. Surprisingly you actually managed to help me here. I had Master/Slave set for the IP address originally but that didn't work so I switched to clone. I will switch back to Master/Slave and go from there.



-----Original Message-----
From: linux-ha-bounces [at] lists [mailto:linux-ha-bounces [at] lists] On Behalf Of Andreas Kurz
Sent: Wednesday, July 25, 2012 4:18 AM
To: linux-ha [at] lists
Subject: Re: [Linux-HA] Active/Active Cluster for DRBD partition

On 07/24/2012 10:21 AM, Yount, William D wrote:
> I have two servers: KNTCLFS001 and KNTCLFS002 I have a drbd partition
> named nfs, on each server. They are mirrored. The mirroring works perfectly.
>
> What I want is to serve this drbd partition up and have it so that if one server goes down, the drbd partition is still available on the other server. I am trying to do this in an active/active cluster. I have a cloned IP address that is supposed to be running on both machines at the same time.
>
> I can get the resources setup according to the "Clusters from Scratch" guide. The issue is that as soon as I take KNTCLFS001 down, all my resources go down as well. They won't even stay running on KNTCLFS002.
>

Your configuration does not look like you followed "Clusters from Scratch" ... drbd has to be a master-slave resource and not a clone, the filesystem and IP needs to be ordered to start after drbd promotion and they need colocation constraints ... that is all step by step explained in this document. And you miss STONITH, a must-have for dual-primary drbd.

Regards,
Andreas

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

> Someone suggested adding constraints but that doesn't really make sense. I want the services running on all systems at once. Constraints are preconditions for bringing the service up on another machine, but the service should already be running so in theory, constraints shouldn't be necessary.
>
> Any help will be greatly appreciated. I have attached my cib.xml to give anyone a better idea of what I am trying to do. I have been going through the "Pacemaker 1.1 Configuration Explained" but so far I haven't found anything.
>
>
> William
>
>
>
> _______________________________________________
> Linux-HA mailing list
> Linux-HA [at] lists
> http://lists.linux-ha.org/mailman/listinfo/linux-ha
> See also: http://linux-ha.org/ReportingProblems
>




_______________________________________________
Linux-HA mailing list
Linux-HA [at] lists
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems

Linux-HA users RSS feed   Index | Next | Previous | View Threaded
 
 


Interested in having your list archived? Contact Gossamer Threads
 
  Web Applications & Managed Hosting Powered by Gossamer Threads Inc.