[Linux-cluster] qdisk

[Stepan Kadlec]

oh sorry, I have pasted wrong configuration (the one that I use for 
non-qdisk configuration) - for testing the qdisk I am using the 
expected_votes="3" and two_node="0" like:

  <?xml version="1.0"?>
  <cluster name="xen" config_version="1">
     <cman expected_votes="3" two_node="0"/>
     <clusternodes>
      ...
     <clusternodes>

     <quorumd label="xen" votes="1">
     </quorumd>

     <fencedevices>
     </fencedevices>
  </cluster>

Harri.Paivaniemi at tietoenator.com wrote:
> When using qdisk in 2-node cluster,
> 
> you should have: two_node="0". The quorum vote becomes like a third node... like you can see in "cman_tool nodes"...
> 
> Have you?
> 
> -hjp
> 
> 
> 
> -----Original Message-----
> From: linux-cluster-bounces at redhat.com on behalf of Stepan Kadlec
> Sent: Thu 11/27/2008 0:41
> To: linux clustering
> Subject: [Linux-cluster] qdisk
>  
> hi,
> any idea why the qdiskd doesn't work? I am running fresh compilation of 
> 2.03.09 cman on 2.6.25 xen kernel on twonode cluster. without quorum, 
> the cluster works fine, but the qdisk would be appreciated. the qorum 
> disk is shared over iSCSI. the config looks like:
> 
> <?xml version="1.0"?>
> <cluster name="xen" config_version="1">
>    <cman expected_votes="1" two_node="1"/>
>    <clusternodes>
>     ...
>    <clusternodes>
> 
>    <quorumd label="xen" votes="1">
>    </quorumd>
> 
>    <fencedevices>
>    </fencedevices>
> </cluster>
> 
> when the cluster goes up and the quorum disc becomes online, it takes 
> exactly 10 seconds until all cluster tools hang on execution (eg. 
> clustat, cman_tool nodes, or the qdiskd) - all stacked in blocking 
> cman_dispatch call (awaiting data with recv).
> 
> is there anyone having ever similar problem?
> 
> thanks steve.
> 
> 
> 
> 
> ------------------------------------------------------------------------
> 
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster