Elasticsearch rebalancing on single node

(Thomas Ardal) #1

When running a single node setup, it always bugged me, that Elasticsearch uses a lot of time rebalancing all of the shards after a restart. The shards will always go to the same node as before restarting the node.

Is it recommended to set cluster.routing.rebalance.enable to none in single node clusters?

(Mark Walkom) #2

Where are you seeing the rebalancing?
It should be reallocating, because if you restart everything becomes unallocated, but that is it.

(Thomas Ardal) #3

Sorry, but I probably mean reallocating then. Why unallocate everything on shutdown and use time to reallocate on startup? The data is the same as when the node shut down and no new node is in the cluster.

(Mark Walkom) #4

What version are you on?
It should be pretty quick if you are on 1.6 and above.

(Thomas Ardal) #5

You're right about the fact that it takes a long time on < 1.6. But on 1.7.1 it still takes a couple of minutes with 4.000 shards.

(Mark Walkom) #6

That's pretty quick for a lot of shards!

(system) #7