Running Elasticsearch locally with multiple nodes

mikko.ropanen

Hi, we ran into the "inconsistent scoring" issue https://www.elastic.co/guide/en/elasticsearch/reference/7.17/consistent-scoring.html in our production environment. While the fix was pretty simple (change search_type to dfs_query_then_fetch), I currently don't have a way of verifying it since production AWS is the only place that uses more than one ES node.

Is there a way to run Elasticsearch locally with more than one node?

Oleksandr Peresypkin

You could try the following approach:

• Stop the project docker/sdk stop
• Comment out the section in the deploy file deploy.dev.yml

#    search:
#        engine: elastic
#        version: '7.10'
#        endpoints:
#            localhost:9200:
#                protocol: tcp

• Add the following environment variables to the section

image:
    …
    environment:
        …
        SPRYKER_SEARCH_HOST: host.docker.internal
        SPRYKER_SEARCH_PORT: 9200

host.docker.internal in the case of using Docker Desktop

• [OPTIONAL] The above steps could be performed in the deploy.local.yml created based on deploy.dev.yml . The former is in .gitignore
• Run the Elastic cluster based on the section Starting a multi-node cluster with Docker Compose of the documentation https://www.elastic.co/guide/en/elasticsearch/reference/7.17/docker.html#docker-compose-file
• Boot the project docker/sdk boot deploy.dev.yml (or deploy.local.yml)
• Build and run the project as usually

Oleksandr Peresypkin

You could try the following approach:

• Stop the project docker/sdk stop
• Comment out the section in the deploy file deploy.dev.yml

#    search:
#        engine: elastic
#        version: '7.10'
#        endpoints:
#            localhost:9200:
#                protocol: tcp

• Add the following environment variables to the section

image:
    …
    environment:
        …
        SPRYKER_SEARCH_HOST: host.docker.internal
        SPRYKER_SEARCH_PORT: 9200

host.docker.internal in the case of using Docker Desktop

• [OPTIONAL] The above steps could be performed in the deploy.local.yml created based on deploy.dev.yml . The former is in .gitignore
• Run the Elastic cluster based on the section Starting a multi-node cluster with Docker Compose of the documentation https://www.elastic.co/guide/en/elasticsearch/reference/7.17/docker.html#docker-compose-file
• Boot the project docker/sdk boot deploy.dev.yml (or deploy.local.yml)
• Build and run the project as usually

mikko.ropanen

The above solution works nicely 👍️

If anyone else is not using Docker Desktop and gets fixated on the "host.docker.internal" part like I did:
you can just use "es01" (the name of the first Elasticsearch node)