P4500 - Adding repaired node incorrectly

[bwicks]

Long story will keep it to highlights.. let me know if you need more detail

single mgmt group

single cluster

4 p4500 nodes running version 9.0

1. had issue with multiple disks failing causing the raid to go into failed

2. used repair option to fix the disks/raid. causing the ghost node

3. after the raid is fixed added back to the management group

-------

4. incorrectly added the repaired node back to the cluster by right clicking an selecting add to existing management group

        a. I did not right click management group and swap ghost with repaired system.. oops

Issue

1. i see both ghost and now new node in the cluster

2. data is restriping .. however on sucessfull restriped volumes the status still says "data protection degraded" .. not sure why?

Question - how do i resolve not adding the node correctly by swapping ghost for node

my thoughts

Wait until all volumes are done restriping and  then try to delete the ghost node.  curious if when done that the volumes will return to a normal status.. not degraded protection.  not sure even if i will be able to delete the ghost with volume status not normal

your thought?  Thanks for your time

[oikjn]

doh!  Thats what exchange/swap nodes are for ;)

What you suggest makes sense, but you might want to check with support on this one.  I"m actually shocked it LET you add a node to the cluster with a missing node as you cannot restripe to it.

That said, it makes sense that data protection is degrated because the missing node is still missing and its part of the stripes are all missing.

I THINK you should be able to remove the RIP node from the cluster once the resync is complete, but I would contact support just in case.  If their logic is screwed up, you might need to remove an existing node from the cluster, have it resync, then do the exchange the correct way to get that dead node out of the cluster for real.  All those resyncs are kinda scarry ideas with some old/failing hardware, so hopefully you are all backed up as well

[psan359]

Were you able to resolve this issue, and what approach did you use to fix the issue and remove the ghost?

Thank you in advance for your response

[Torsten.]

If a node is repaired (e.g. a new system board with a new MAC address) you usually use the replace node option.

But if this fails, you should try to re-detect the nodes, this normally helps.