Replacing a node but keeping data disk as-is


Would this work?

I'm planning to replace a node in the cluster by stopping it, moving the data disk from to the new one and joining it to the cluster. Will it be able to properly recognise the shards it's holding?

(Boaz Leskes) #2

Yes, that should work.

(Simon Willnauer) #3

ideally you stop indexing into the shards on that node and call synced flush before you shut down the node just to be on the safe side


Yea forgot to mention it but it all worked ok. Didn't know about synced flush so I didn't do that, I just stopped elasticsearch on the node, moved the disk and all indices were moved successfully.

I'm not sure no data was lost at all but even if that happened it was in small numbers.

(Simon Willnauer) #5

my suggestion could only help in reducing the time to recover, your data should be safe!

(system) #6

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.