Set up a distributed instance of locust.io.

You need to first set up your config to run the instances.

Create a file named config.yaml with contents like::

locust-master:
    master: true
    host: http://url.prefix.to.test/
    locust_file: |
       # The contents of the locust_file.py to use.
       # See http://docs.locust.io/en/latest/writing-a-locustfile.html
locust-slave:
    master: false
    locust_file: |
       # The same locust_file as above.

(Hopefully we'll be able to make specifying the locust_file in the slaves unnecesarry soon.)

Now you need to deploy the charm twice and relate the master to the slaves:

make sync-charm-helpers
juju deploy --repository .. local:locustio locust-master --config config.yaml
juju deploy --repository .. local:locustio locust-slave --config config.yaml
juju add-relation locust-master:slave locust-slave:master
juju expose locust-master

Once the services has started you will be able to connect to port 8089 on the public address of the master unit and use the web interface.

If you need to change the locust file, or the host that is being tested you can update the config.yaml then run

juju set locust-master --config config.yaml juju set locust-slave --config config.yaml

If you need more slaves for your testing you can run

juju add-unit locust-slave

and use the -n option to add multiple units at once if you wish.

Deploying the charm twice and making one the leader in config would be unnecessary if the units could just elect a leader themselves.

Once that is supported natively by juju then the charm could be made to use that.