Number of shards in one node of cluster is stuck


(Oleg Bragin) #1

Hello,

I have a cluster with 2 nodes. Each node has 2248 (1124 * 2) shards.
When I check cluster health using head plugin on one server (node) I see that indexing is going on and number of shards is increasing.
On another cluster using head plugin also I see that number of shards is stuck somewhere in the middle.
Another thing to notice is that on a "good" node when all shards are processed the cluster gets the green state, which is ok. My concerns is another node with stuck index. is it ok? Should the number of shards be equal on all nodes throughout the cluster?


(Mark Walkom) #2

That's quite a lot, you should look at reducing that quite dramatically for ES to run optimally.

What do you mean, what does this have to do with the problem?


(Oleg Bragin) #3

Hi,

You were right, the problem wasn't with configuration. Thank you!


(Oleg Bragin) #4

Hi again :smile:

I have a default configuration of elasticsearch, which sets 5 shards + 1 replica for the index.
Currently I have 234 indices, so that's where I got more that 2200 shards with replicas.
Am I correct that the only one way to reduce the that number is by reducing the number or shards per node?


(Mark Walkom) #5

You should really change the mapping to reduce the number of shards, ideally it's one shard per node per index.


(system) #6