Replicas

Elastic Stack Elasticsearch
1

setup.template.settings: index.number_of_shards: 3 index.number_of_replicas: 3
If i use above configuration in beats, is it create 3 replicas for each shards. or one replica for each shards.

I used 3 shards and 3 replicas. After that Kibana dashboards become slow. I am using 3 data&master nodes and one co-ordinate node. Also getting below INFO logs on one node.

[2021-07-27T05:28:09,161][INFO ][o.e.i.r.PeerRecoveryTargetService] [USLA-PAPP-ELK04] recovery of [issqmetric-2021.07.27-000015][0] from [{USLA-PAPP-ELK06}{K0CWQ8D9Tw-ScRUabDeWHw}{zRpCYtPBRpuAmtkur79f6A}{USLA-PAPP-ELK06}{10.7.0.66:9300}{cdhilmrstw}{ml.machine_memory=33510563840, ml.max_open_jobs=20, xpack.installed=true, ml.max_jvm_size=17179869184, transform.node=true}] interrupted by network disconnect, will retry in [5s]; cause: [[USLA-PAPP-ELK04][10.7.0.64:9300] Node not connected]
2

Your calculation is slightly off. Configuring the number of replicas to 3 means that each shard will have 3 copies... so you will end up with 12 shards in total for this index.

Can you define what slow means in your context? Is it the query response time? Does this change after some time? WIth which configuration was it faster before? How much difference is there between fast and slow?

Adding some more context and especially comparing numbers would be greatly appreciated!

Thanks!

3

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