Ontap 8040 rebuild node in cluster?

[snyderkv]

Greetings,

Node A on our 8040 cluster is dead and being replaced with a new controller.

Node B is finicky and although shows Healthy and targets pingable, any commands issued to it say that no nodes are healthy. Storage disk show, looks like it owns the disks but we aren't sure. CF show shows empty for the good node and "in takeover" Giveback deferred" for the bad node.

My question is, keeping in mind we can't blow away storage during re-imaging of NodeA, will it just automatically join the Node B cluster if I use the same name and management IP or will I have to force disjoin the bad node first and then rejoin the cluster? Or perhaps a standalone Node can forcibly take ownership? I'd lab this but I've got less than 24 hours to do this without losing data.

Any help is appreciated.

[nboggs · Accepted solution]

This documentation should help with replacing the system controller.

https://library.netapp.com/ecm/ecm_download_file/ECMP1199896

[snyderkv]

Thanks Boggs

[SpindleNinja]

I'm assuming no support?  if you have support open a ticket ASAP.  

otherwise, yeah.  you do a controller replacement and the system thinks the new controller is the old controller.  

[snyderkv]

 That would be great, will give it a shot.

We do have support which determined we need a new controller and gave us a controller replacement document but obviously theirs a little more going on here so I thought I'd ask. Thanks Ninja

[SpindleNinja]

Unless you're "parts only"  They should also send a tech out to do the swap, it's not an easy process 🙂   

In failover, you should see the surviving controller own most of what the dead controller had.   (disks, aggrs etc..)