Good Morning Checkyourlogs Fans,

Today, I had a customer call me and let me know that they would like to move a Storage Spaces Direct Cluster into a new rack in the Data Center. They have a full maintenance outage window so we are just going to move the whole thing all at once. The process to do this is similar to taking a full outage for servicing.


  1. Ensure that the Network Ports in the new location have the same VLANs configured. To be sure you could ask your network team for a running configuration. If you don’t get this right, your virtual machines won’t be able to talk on the new switches.
  2. MAKE Sure that if you are connecting into new switches for RoCE / RDMA traffic that things are configured properly. I normally like to have dedicated East/West Switches for my Storage Spaces Direct cluster so I would say just move the switches over with you. If you are in a 2-node configuration just bring your SPF+ or QSFP+ cables with you.
  3. Plan your maintenance window.
  4. Shut Down all of the Virtual Machines in the Cluster

  5. Make sure that you don’t have any stand-alone virtual machines running on the individual nodes. We have seen this before where VM’s live outside of the cluster and still use the Cluster Shared Volumes.
  6. Stop the cluster to take the storage pool offline. Run the Stop-Cluster cmdlet or use Failover Cluster Manager to stop the cluster.
  7. Set the cluster service to Disabled in Services.msc on each node. This prevents the cluster service from starting up while being patched or rebooted

  8. You can now power down the nodes. Note: This is also where you could run a complete Windows Update cycle and reboot as many times as required for the cluster. You can do all of the nodes at once here.
  9. After the nodes are re-racked in their new location make sure that the Virtual Adapters can communicate with all of the other nodes. I like to do a ping sweep for this when I move things.
  10. Set the cluster service back to Automatic on each node.

  11. Start the cluster. Run the Start-Cluster cmdlet or use Failover Cluster Manager.

  12. Run netstat -xan to make sure all the RDMA traffic looks good.

  13. Run Get-PhysicalDisk

  14. Run Get-VirtualDisk

  15. Run Get-StoragePool

  16. Run Get-StorageJob

  17. Start back up the VM’s

  18. Double check the VM’s to make sure they can see their respective networks.

And there we have it a Storage Spaces Direct Cluster moved to a new home.

I hope this has helped you,

Dave

Advertisements