Building elasticsearch indexes from another container - django

Question

I have a Django project that uses django-elasticsearch-dsl. The project is dockerized, so elasticsearch and the web projects leave in separate containers.
Now my goal is to recreate and repopulate the indices running
python manage.py search_index --rebuild
In order to do that, I try to run the command from the container of the web service the following way:
docker-compose exec web /bin/bash
> python manage.py search_index --rebuild
Not surprsiginly, I get an error
Failed to establish a new connection: [Errno 111] Connection refused)
apparently because python tried to connect to elasticsearch using localhost:9200.
So the question is, how do I tell the management command the host where elasticsearch lives ?
Here's my docker-compose.yml file:
version: '2'
services:
web:
build: .
restart: "no"
command: ["python3", "manage.py", "runserver", "0.0.0.0:8000"]
env_file: &envfile
- .env
environment:
- DEBUG=True
ports:
- "${DJANGO_PORT}:8000"
networks:
- deploy_network
depends_on:
- elasticsearch
- db
elasticsearch:
image: 'elasticsearch:2.4.6'
ports:
- "9200:9200"
- "9300:9300"
networks:
- deploy_network
db:
image: "postgres"
container_name: "postgres"
restart: "no"
env_file: *envfile
ports:
- "5432:5432"
volumes:
- db_data:/var/lib/postgresql/data
volumes:
db_data:
networks:
deploy_network:
driver: bridge
UPDATE:
In the Django project's settings I setup the elasticsearch dsl host:
# settings.py
ELASTICSEARCH_DSL = {
'default': {
'hosts': 'localhost:9200'
}
}

Answer 1

Since your Django project and Elasticsearch are in 2 separate containers, setting ELASTICSEARCH_DSL's host to 'localhost:9200' won't work, in this case localhost refers to localhost inside Django container.
So you need to set it like this:
# settings.py
ELASTICSEARCH_DSL = {
'default': {
'hosts': 'elasticsearch:9200'
}
}