Rejoin of failed node to cluster

  • Thread starter Thread starter Gary
  • Start date Start date
G

Gary

Hi,
We had a node fail in our cluster and had to start from
scratch with OS install, etc. It was evicted from the
cluster and then rejoined when repaired. The repaired node
has never seen the shared drives since the non-failed node
and cluster was never off-line. My question: The rebuilt
node joined the cluster fine but will it support the drive
resources not having ever assigned drive letters for them
in its disk manager? I'm afraid to fail over the cluster
resources to the new node to test.

Thanks, Gary
 
-----Original Message-----
Hi,
We had a node fail in our cluster and had to start from
scratch with OS install, etc. It was evicted from the
cluster and then rejoined when repaired. The repaired node
has never seen the shared drives since the non-failed node
and cluster was never off-line. My question: The rebuilt
node joined the cluster fine but will it support the drive
resources not having ever assigned drive letters for them
in its disk manager? I'm afraid to fail over the cluster
resources to the new node to test.
Click to expand...

Clustering takes care of the disk configuration
internally but you should definitely fail the resource
groups back and forth between the old and the newly
rebuilt node a couple of times to make sure everything is
OK. We always test clustering by copying a couple of 1GB
files across the network to the cluster shares while
owned by each node in turn, this usually throws up any
oddities which might have sneaked into the
configuration. You really don't want to wait for any
errors to crop up when the primary has blown up and you
desperately need the second node to work.

Cheers
Matt
 
Gary,

It may support the drive, but it is possible that the node may change the
drive letters on you at some point.

I think that you should take down your cluster and verify the drive letter
settings on the "new" node before going any further to prevent potential
issues in the future. When you take down the "new" node, disable the
"cluster disk" driver so that you can see the disks properly.

Regards,
JT
 
Back
Top