What would make cpu usage differs too excessively between nodes?

Elastic Stack Elasticsearch
1

We have a cluster of three nodes with same hardwares. We have made some http benchmark test on:

  • one index of 4 million docs, 5 shards, 1 replica
  • use golang client gopkg.in/olivere/elastic.v6 connect to all three nodes. The coordinate node should not be focused on one node
  • es version: 6.5.4
  • this is a pure benchmark test environment, no other application processes and no other es query / index requests

image
image862×166 19.1 KB

2

And this is out output of hot thread

3

Does your data distribute evenly across three nodes?

5

The shards distribution seems to be evenly.

index          shard prirep state     docs   store node
quick_match_v1 3     p      STARTED 847408 570.7mb matchtest_es-node-172.16.0.223
quick_match_v1 3     r      STARTED 847408 570.7mb matchtest_es-node-172.16.0.222
quick_match_v1 2     r      STARTED 847348 559.9mb matchtest_es-node-172.16.0.223
quick_match_v1 2     p      STARTED 847348 559.9mb matchtest_es-node-172.16.0.221
quick_match_v1 4     p      STARTED 846896 568.2mb matchtest_es-node-172.16.0.222
quick_match_v1 4     r      STARTED 846896 546.8mb matchtest_es-node-172.16.0.221
quick_match_v1 1     p      STARTED 845880 587.8mb matchtest_es-node-172.16.0.223
quick_match_v1 1     r      STARTED 845880 575.7mb matchtest_es-node-172.16.0.221
quick_match_v1 0     r      STARTED 848310 570.9mb matchtest_es-node-172.16.0.223
quick_match_v1 0     p      STARTED 848310 570.9mb matchtest_es-node-172.16.0.222
6

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