High IO when use replica

Elastic Stack Elasticsearch
1

Dear All:
I have a ES 6.0.1 cluster with 64 machine and 256 nodes .with about 200000 records will be indexed to ES index every second. And Some problems occur.

  1. if the index has 256 shards ,0 replicas , get a normal IO util(less than 10)
  2. if the index has 64 shards ,0 replicas , get a normal IO util(less than 10)
  3. if the index has 32 shards ,1 replicas , get a normal IO util(less than 10)
  4. if the index has 16 shards ,3 replicas , get a normal IO util(less than 10)
  5. if the index has 64 shards ,1 replicas , get an abnormal IO util(more than 50)
  6. if the index has 128 shards ,1 replicas , get an abnormal IO util(more than 50)
  7. if the index has 32 shards ,3 replicas , get an abnormal IO util(more than 50)
    can anyone tell me why? thanks a lot!
2

6.2.0 is available ,you can try it

3

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