Powerscale - Smartfail Last three nodes, all at once ? or 1by1

[Coolgoose]

We have a cluster with NL & X . Filepooldefault policy is used for evacuating NL Data. ( NL is almost empty)

To remove last 3 nodes from the cluster, Smartfail all 3 NL nodes at once. Any experience on this ? is this faster and efficient?

If done 1-1, 

smartfail one NLnode ( 1-2 days), Other 2 NL nodes remaining in the cluster will fill up first right, then smartpool will move them to x Nodes ?

Smartfail remaining two NL nodes. so all data will go to X nodes. ( again 1-2 days)

[mandar kolhe]

Are their any other nodes in that node pool? If not smartfail one node and smartfail remaining 2 nodes at same time

--
You received this message because you are subscribed to the Google Groups "Isilon Technical User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to isilon-user-gr...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/isilon-user-group/33932a5e-5bb5-4c1c-9789-4a63f21fac47n%40googlegroups.com.

[Paul Carrington]

Have you used a file pool policy to migrate data to other node pools first?

Is global spillover enabled?

If the nodes still have high usage in terms of capacity its likely to be deduce shadowstores. The data can be migrated manually if its excessive. I found with nl nodes that the performance impact was quite high when failing the last few nodes out. I've always failed out one by one until last two then do the last two in tandem as the nodepool won't have quorum once you drop below two nodes.

--

[Coolgoose]

Only one node type on each node pool.  thank you ..  

[Coolgoose]

thank you all,

Smartfailed  last 2 nodes at same time. finished in 3 days.