This post is more than 5 years old
30 Posts
1
3464
May 8th, 2015 07:00
Add node to Islon cluster with smartfail already progress?
Hello,
Does any of you know if a node can be joined to an Isilon cluster that already has a smartfail operation in progress?
We have to perform a test at a customer (on his behalf):
a) Remove a node
b) Some read/write tests
c) Add the node back
I want to know if we can process with adding the node back, before the smartfail process (from step a) has finished.
We are running latest 7.2.
Thanks
Francis
No Events found!
paul_schuette
3 Posts
1
May 12th, 2015 14:00
The answer to this question is no, not without interrupting the running smartfail.
The reason for this is that adding a node changes the cluster group statement, which would cause the running flexprotect job (smartfail) to be system cancelled. The flexprotect job would then have to be restarted.
Anonymous User
170 Posts
0
May 8th, 2015 10:00
I've got a cluster running 7.1.1.2 and happen to have 2 nodes being SmartFail'ed right now.
In the gui, the nodes that are being SmartFailed are not available to be added.
You can, however, abort the smartfail although it is not recommended:
cadiletta
1 Rookie
•
106 Posts
1
May 12th, 2015 14:00
From a theory standpoint, yes you should be able to add a new node to a cluster at any time.
Best practice would say that you should avoid major group changes during a smartfail operation, as you want to avoid unnecessary interruptions of a critical job engine task working to protect your data.
Under dire circumstances you could pause the smartfail operation, add the node, then resume the smartfail operation.
Now another layer of caveat to the above - is the node you are adding the same node that you smartfailed? If so, the cluster maintains a record of that node and may prevent the re-introduction of that node, until the smartfail is complete.
To get around this you would want to work with support to ensure data integrity is maintained. In this case as you are doing a test the risk might be acceptable, however I would still hesitate to provide further instruction on an open forum, as the steps could very easily put production data at risk for anyone else who followed those steps.