I have a customer that has some data on X node of a Isilon cluster that has mix of X nodes and NL nodes, customer now wants to move all the data (~15TB) from x node to NL node and dedicate the x node for a different application, I haven't done this before so wanted to check what is the best way to go about it and if there is any tool available to move the data between different node type in same Isilon Cluster?
Solved! Go to Solution.
The nice thing about SmartPools and its file pool policies is that one doesn't need to change paths or directory names for data, just create (or modify) the policy rules. In particular, do not actually use path names like "/ifs/xnodes" -- this would be cumbersome to track and update when data gets migrated to another pool.
A bit more practical example (simplified, omitting the recommended "clustername" in the path):
Let's say projects A-C are the "old" ones on X nodes so far,
to be migrated to NL nodes,
and project D has been freshly created for the "new purpose".
By a clear scheme, there would be three rules for projects A-C -> NL nodes
and one rule for projectD -> X nodes.
One might take a short cut and define the SmartPools "default policy" to be -> NL nodes
and then have only one explicit policy for project D -> X nodes (less typing or clicking in the UI).
Thanks for the response Peter, after taking a look at the document , it looks like customer maybe able to run Smart Pools Job as listed under the File Pool Policy Engine section of the document? so if I'm understanding it correctly we need to create a File Policy that should have source data (on X nodes) and target (NL node directory path) and then run that smart pools job?
Absolutely -- on a mixed cluster SmartPools should be licensed already, and the job would be
automatically scheduled once a day.
Thanks again, just to be clear, the objective here is to move the data completely from x-node and repurpose these nodes for other usage, does the process that I outlined still good ?
You mean, repurpose the X nodes with a completely different cluster?
To remove the nodes from the current cluster, "smartfail" them
after the SmartPools job has finished:
(That documents starts with discussing nodes in "down" status,
but the described smartfailing procedure applies to decommissioning nodes, too.)
no we are not removing the x nodes from cluster, Its just that the customer wants to use the storage on x-node for a different purpose and they want to drain the data from x-nodes to NL-nodes storage
OK, then just set up the file pool policies simply by path:
/ifs/path/to/old/data -> NL nodes pool
/ifs/path/to/new/purpose -> X nodes pool
Hi Peter ,
Quick question , in your file pool policies recommendation , I'm getting little mixed up - the first one /ifs/path/to/old/data --> NL nodes pool seems ok since this is where the data needs to go but the 2nd one /ifs/path/to/new/purpose--> x nodes pool , do you mean this should be the *from* x-node, sorry but trying to understand the correct flow , lets assume that our xnode data (source) is at /ifs/xnode/data and it needs to go to NL nodes (target) /ifs/nlnode/data1 , what should be the file policy configuration ?