Data balancing problem

Elastic Stack Elasticsearch
1

we have 8 nodes in an ES cluster and an index which we have set to 4 shards
with 1 replica. i would've assumed that this would mean 4 roughly equal
chunks on 4 nodes and replicas of those 4 chunks on the other 4 nodes but
that doesn't seem to be the case. for example, one node is primary for one
chunk and is a non-primary replica of two others and some nodes have no
data at all. is the rebalancer disabled by default? how can i check the
status? is there a way to manually redistribute the data?

thanks for any guidance!
-r

--

2

The balancing is not very intelligent, and it tends to mess up if a node
leaves and enters the cluster.
You can, however, control how many shards (primary & replicas) that you
have on each node. I use this to ensure that each node has a primary shard
and one replica.
Take a look at

and index.routing.allocation.total_shards_per_node.

Den torsdagen den 13:e september 2012 kl. 13:30:05 UTC+2 skrev revdev:

we have 8 nodes in an ES cluster and an index which we have set to 4
shards with 1 replica. i would've assumed that this would mean 4 roughly
equal chunks on 4 nodes and replicas of those 4 chunks on the other 4 nodes
but that doesn't seem to be the case. for example, one node is primary for
one chunk and is a non-primary replica of two others and some nodes have no
data at all. is the rebalancer disabled by default? how can i check the
status? is there a way to manually redistribute the data?

thanks for any guidance!
-r

--