I have a question around the permanent removal of nodes from a cluster, such as during a hardware refresh and I'm trying to establish the correct procedures/checklists to go through to ensure that during a hardware upgrade nothing is overlooked and user disruption is kept to a minimum.
Scenario: (All nodes running ONTAP 9.1 P7 and are HA pairs)
2 x FAS 8020 units (cluster nodes 1 to 4)
1 x AFF A200 unit (cluster nodes 5 and 6)
Assuming that all user data volumes have been migrated to the aggregates on the AFF nodes 5+6 then this should, I believe, just leave the following steps however I would welcome any comments to correct or enhance the process.
• Migrate Data SVM root volumes from FAS8020 aggregates on to AFF nodes (we already have LS-mirror copies but the ‘live’ version is still on the 8020s)
• Data LIFs for CIFS and iSCSI (hosted on the 8020s) have to be removed, or migrated to the new AFF nodes.
• Make the 8020 nodes ineligible for cluster RDB operations (i.e. remove eligibility forcing it to use one of the AFF nodes) - using command: node modify -node nodenametoberemoved -eligibility false
• Perform a 'cluster leave' operation on the 8020’s - does this need to be performed on each node in a HA pair?
• Physically disconnect the node from cluster network
• Zero the data disks
Thanks in advance,