I'm wondering if it is possible to unjoin a ha-pair of a cluster that still hold data volumes. According to what I have seen in the ONTAP documentation all data has to be evacuated before you can unjoin nodes.
Problem: not enough space on the shelves of the new nodes to evacuate all the data and not enough rack space for more temporary shelves. Also a downtime is not possible, at least not for all of the data.
The plan would be to add the new nodes to the cluster, migrate those data volumes that must stay online to the new nodes/shelves in the cluster. Then unjoin the old nodes, turn the old controller into a disk shelf (internal disks are not partitioned), add these shelves to the new controllers shelf stack and reassign the disks using the disk reassign -s -d command on the node shell.
Current ONTAP version is 9.4P3. Old nodes are FAS2554, new ones are FAS2720.
Reassigning Disks worked in 7-mode, not sure if it still does in cDOT (because of databases, cluster ring, ...). The disk reassign command is still present on node shell.
Any input/idea on this would be highly appreciated.
Sorry, I don't believe there is a way to do what you're asking. unjoin forces the reinitialization of the HA pair that you're removing. And converting from a controller to a shelf also requires re-init / zero out.
Are their any shelves on the 2554?
On the 2750, do you have all space saving mechanisms in place?
thanks for the reply. Aggregate Relocate is not possible because of internal disks. And a head swap would result in a downtime which is an absolut no go for the customer. My current plan is to convince the customer to use more temporary shelves and do a vol move. Hope this works ...
we found a workaround. We are evacuating most of the data to a backup netapp and the rest via vol move to the new nodes. So we can delete the data aggregates and unjoin the nodes savely. Seems like the saves way to us.