On Wed, 05 Nov 2008 14:06:03 +0000, kmeuskens wrote:
> (FYI I cross-posted this in OES:Linux - Storage as well but this seems
> like a better forum. I will update both on resolution.)
>
> I have 2 nodes in a working cluster. They are both using a SBD volume on
> a shared iSCSI LUN using RDMs. (My servers are all virtual under
> VMware.)
> My problem is that I cannot join a new server to this cluster because I
> cannot map to the shared disks - they don't show up in the 'connect'
> list of the VM config. Is there something I'm missing, does NCS mark
> these LUNs as unavailable for further connects or something? I know that
> NCS supports up to 32 nodes in a cluster so this is confusing. I'm
> thinking of disconnecting the SBD partition LUN from the existing
> servers, then seeing if I can connect to it from all three....any
> comments?
> Am I limited to 1 node per ESX host per cluster because of RDMs? I have
> tried SCSI bus sharing without success.
NCS does not mark them or hide them in any way. Nor is there a licensing
blackout, so that is out too.
Before you try to join the cluster (I assume you are doing this manually
as opposed to within the autoyast build?) can you see the SBD LUN - not
looking at the physical blocks, but is the LUN visible in the list of
devices available?
If you can see, then there is a very nasty case sensitivity bug in
getting NCS to work (another reason why AutoYast is so good) that is it
is not identical that can blow things out.
HTH
T
--