Logstash to Elasticsearch resilience

Hello everyone,

Im having a little bit of trouble finding a way to approach resilience on my Elasticsearch cluster.

Currently i have 2 Logstash nodes for resilience.
Each one of them are pointing to my 4 Elasticsearch ingest/data_hot nodes.
Plus, i have more 3 master nodes coordinating it all.

My problem is, if one of my ingest/data_nodes fails or go down, Logstash is stopping sending data to elastic giving me the error "Elasticsearch Unreachable: [...] connect timed out"

Is there any way to avoid this errors and keep sending data to the Elasticsearch alive nodes?
Thank you for your help!!

What is your output in logstash pipelines?

Logstash should load balance if you have more than one node in the output configuration, it would give you his error, but would retry to send using another node.

Currently this is the configured logstash output (something like this, edited for privacy):

elasticsearch {
      hosts => ["https://node1:9200", "https://node2:9200","https://node3:9200","https://node4:9200"]
      index => "filebeat-index-%{+YYYY.ww}"
      pipeline => "filebeat-pipeline"
      user => "logstash"
      password => "${pw}"
      ssl => true
      cacert => "/ca.cer"
      }

Currently im also not using replicas, so there are some primary shards missing because of the node that is down.
In case the logstash keeps indexing when a node is unavailable, maybe thats the reason im not seeing any indexing action?!?

Thank you for your help

If there are primary shards missing, then those index will not receive any data until the node with their primary shards are back.

If you want logstash to keep indexing to those index you need to use replicas.

thank you so much for your help.

Gonna implement it.
Plus, i have another question, is there any way to update the system indices number of replicas?, every one at once. would like to increase them for 3 but i cant find a way to do it!?

I mean, i would like to have 1 replicas of the normal indices, and 3 of the system.

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