Connection refused error (111) when using Locust for Performance Testing against docker-compose app

Question

I am using Locust, a performance testing tool, to load test an application that is setup to run within docker-compose. I get the following error (connection refused, error 111) for every request:

Error report
 # occurrences      Error
--------------------------------------------------------------------------------------------------------------------------------------------
 5                  GET /parties/: 'ConnectionError(MaxRetryError("HTTPConnectionPool(host=\'localhost\', port=8080): Max retries exceeded with url: /parties/ (Caused by NewConnectionError(\'<urllib3.connection.HTTPConnection object at 0x7fa6f294df28>: Failed to establish a new connection: [Errno 111] Connection refused\',))",),)'

I am running Locust from a docker container as follows:

docker run --volume /mydir/locustfile:/mnt/locust -e LOCUSTFILE_PATH=/mnt/locust/locustfile.py -e TARGET_URL=https://localhost:8080/parties -e LOCUST_OPTS="--clients=1 --no-web --run-time=600" locustio/locust

The weird thing is that when I use curl to hit the exact same URL it works properly.

curl http://localhost:8080/parties

Any help is appreciated!

Answer 1

localhost is probably the wrong hostname to use in your context. when you use containers, the hostname must match the container's name. so use your container's name instead of localhost.

references:

https://github.com/elastic/elasticsearch-py/issues/715

https://docs.locust.io/en/latest/running-locust-docker.html

https://docs.locust.io/en/stable/configuration.html

general assumption:

locustfile.py exists in the current working directory

a distributed example with docker compose:

for example, the following docker-compose.yml file config will work:

services:
  web:
    build: .
    command: python manage.py runserver 0:8000
    volumes:
      - .:/code/
    ports:
      - "8000:8000"
  master:
    image: locustio/locust
    ports:
      - "8089:8089"
    volumes:
      - ./:/mnt/locust
    command: -f /mnt/locust/locustfile.py --master -H http://web:8000
  worker:
    image: locustio/locust
    volumes:
      - ./:/mnt/locust
    command: -f /mnt/locust/locustfile.py --worker --master-host master

the -H flag (alias for --host) on the following line makes the trick:

command: -f /mnt/locust/locustfile.py --master -H http://web:8000

a non-distributed example with docker compose:

services:
  web:
    build: .
    command: python manage.py runserver 0:8000
    volumes:
      - .:/code/
    ports:
      - "8000:8000"
  locust:
    image: locustio/locust
    ports:
      - "8089:8089"
    volumes:
      - ./:/mnt/locust
    command: -f /mnt/locust/locustfile.py --host=http://web:8000

you can also specify a config file, instead of defining the flags on the command line. supose the file locust.conf exists in the current working directory:

locust.conf:

host: http://web:8000

so in your docker-compose.yml you do:

command: -f /mnt/locust/locustfile.py --config=/mnt/locust/locust.conf

instead of:

command: -f /mnt/locust/locustfile.py --host=http://web:8000

a non-distributed example without docker compose:

the containers must live on the same network, so they can communicate with each other. to do so, let's create a common bridge network called locustnw:

docker network create --driver bridge locustnw

now run your app container within this network. Supose it's listening on port 8000 and named web:

docker run -p 8000:8000 --network=locustnw --name web <my_image>

now run your locust container, within the same network too. Supose it's listening on port 8089:

docker run -p 8089:8089 --network=locustnw -v $PWD:/mnt/locust locustio/locust -f /mnt/locust/locustfile.py --host=http://web:8000

the --network, --name and --host flags are the keys!