How are environment variables handled?

Open Source Elasticsearch and Kibana
1

Hello,

I see in the examples given through the documentation that you can define environment variables like cluster.initial_master_nodes or discovery.seed_host…

How do these variables get handled inside the container? Are they automatically resolved by Elasticsearch or do they need to be appended to the elasticsearch.yml configuration file through a manual action?

2

Yes they are added by docker entrypoint

Here to mos important lines:

# Parse Docker env vars to customize Elasticsearch
#
# e.g. Setting the env var cluster.name=testcluster
#
# will cause Elasticsearch to be invoked with -Ecluster.name=testcluster
#
# see https://www.elastic.co/guide/en/elasticsearch/reference/current/settings.html#_setting_default_settings

declare -a es_opts

while IFS='=' read -r envvar_key envvar_value
do
    # Elasticsearch settings need to have at least two dot separated lowercase
    # words, e.g. `cluster.name`, except for `processors` which we handle
    # specially
    if [[ "$envvar_key" =~ ^[a-z0-9_]+\.[a-z0-9_]+ || "$envvar_key" == "processors" ]]; then
        if [[ ! -z $envvar_value ]]; then
          es_opt="-E${envvar_key}=${envvar_value}"
          es_opts+=("${es_opt}")
        fi
    fi
done < <(env)

So pay attention to use only valid ES options as environment variables.

1 Like
3

Thank you for answering Aparo.

I have 2 questions from here:

1- Where can I find the list of all valid ES options as environment variables?
2- In case the same variable is defined as an environment variable and in the same time in the elasticsearch.yml file, which one has priority?

4

  1. you need to look to the ES documentation, but everything you put on elasticsearch.yml is a valid variable. They are simple key/value pairs.

  2. I’m 90% sure that ENV should have priority. I have no time now to check the code of Bootstrap and Environment class.
    Just try with cluster.name in docker.

2 Likes