Closed Infra-Red closed 5 years ago
From my point of view - having a local ES drain for Logstash was a great idea, and keeps cluster configuration simpler. To resolve memory issues, we can just increase amount of memory for instance, and maybe change heap calculating formula here. Anyway, I'd prefer to keep this conversation for a couple of weeks to hear from other active Logsearch users.
Hi @axelaris ! I changed the PR and now by default Logstash will forward to local ES host. I've added link support to ingestor_syslog
job and now operator can optionally configure forwarding to remove ES host. I've also added ops file for this ingestor-forward-to-elasticsearch-data.yml
. Is it works for you?
Merged. Thank you @Infra-Red!
Hey! This PR changes the default hosts configuration in logstash elasticsearch output. Instead of collocating ElasticSearch and Logstash on the same VM we can point Logstash output plugin to the DNS name of ElasticSearch data nodes. This will decrease the amount of RAM used by Ingestor instances, check the graph below:
I will also create a separate PR which will allow to configure Elasticsearch coordinating nodes in a high load environments.
/cc @pommi @diogomatsubara