Unsolved
This post is more than 5 years old
6 Posts
0
1228
February 28th, 2017 09:00
EMC Isilon x400 to x410 migration
I am running a project to replace existing Isilon 10 x X400 nodes with 7 x X410 nodes (Adding extra 84TB and saving 12U rack space) on both our Isilons.
I am aware I need to add each X410 node into the existing cluster and create a new X410 smart pool with +2d:1n protection.
What is the process to remove each X400 node out of the existing cluster to ensure a successful migration?
Some guidance and direction on how to complete would be great? If you could point me in the direction of X Series node migrations that would be even better?
Many Thanks
Jonathan Gardner | Storage & Backup
EMEA Infrastructure and Virtualisation
CBRE | Digital & Tech
St Martin's Court | 10 Paternoster Row | London | EC4M 7HP
Ext: 6977 | DDI: +44 20 3257 6977 | Mobile: +44 7920212519
EMEA Enterprise Hardware Team Hotline +44 20 3257 6799


Eric_W1
33 Posts
0
February 28th, 2017 13:00
we just did a NL400 -> NL410 migration but I think some of the caveats might be similar.
Do you have SmartPools? If you do not have smartpools licensed then your data will begin to balance as soon as the new pool is formed.
If you have manual protection policies on anything it may bite you as you try to smartfail the old nodes out. If you have ever played with manual protection on files or directories you may wish to review your tree or force them all back to defaults before you start taking nodes out. We got conflicting info on "global spillover" and in the end it was necessary to have that ON in order to complete the smartfail when a handful of random files had manual protection policies holding them to the old pool.
Check every node when you are done to make sure there is a consistent view of the cluster.
gvaidman
20 Posts
0
February 28th, 2017 17:00
The migration process is described in KB 304058. At a high level, you're
This process is the same no matter what node types you have. It's important you follow the process in the order described (specifically that you do the data migration before smartfailing the old nodes); otherwise you will be moving the data multiple times.
A couple of things you won't find in that article, though (I recently asked EMC Support that this be added):