Using Fabric to start Locust on multiple slaves

Question

I am rather new to Fabric but I started working with it in order to automate the execution of load tests.

I use Locust for load testing my apps but since i test with over 100k CCU this requires multiple slaves.

I've used boto3 to automate the EC2 instances creation and that work perfectly, but unfortunately i've run into a problem when trying to use fabric to deploy the Locust agent and then run it automatically on all machines.

Bad part is that since Fabric waits from a response when running Locust in distributed mode it only connects one Slave and then hangs (since no response is ever received).

Can you please help me with any suggestions regarding this issue? Did anyone have this issue and managed to fix it?

Thank you in advance.

Why do fabric waits for response ? is there any prompts which stopping execution ? — Roshan, Aug 28 '16 at 14:02

score 0 · Answer 1 · edited Feb 16 '17 at 11:36

For Locust Master:

def master_start(filename, web_port, master_port):
    with settings(warn_only=True):
        return run('nohup locust -f %s --master --port=%s --master-bind-port=%s >& /dev/null < /dev/null &'% (filename, web_port, master_port), pty=False)

For Locust Slaves, but run in parallel:

def slave_start(filename, master, master_port):
    with settings(warn_only=True):
        run('nohup locust -f %s --slave --master-host=%s --master-port=%s >& /dev/null < /dev/null &'%(filename, master, master_port),pty=False)

Using Fabric to start Locust on multiple slaves

1 Answers1