I am trying to create a program that creates multiple droplets, sends a script to each droplet, and initiates the execution all of the scripts without waiting for the output. I have tried to run it in the background, using nohup so that it isn't killed off when disconnected from terminal with the following code:
for i in len(script_names):
c = Connection(host = host[i], user = user[i], connect_kwargs = {"password" : password, "key_filename" : key_filename})
c.run("nohup python3 /root/" + script_names[i] + " &")
I have tried other variations of the same idea, including setting "pty=False", redirecting the output to dev/null with "> /dev/null < /dev/null &" yet nothing seems to work.
Is it possible to issue multiple commands to run scripts on different hosts concurrently without waiting for the output with fabric? Or should I use another package?
Fabric 2.x's groups aren't fully fleshed out yet, so they aren't well suited for this use case. In fabric 1.x I would accomplish this by using a dictionary for script_names where the keys are the host strings from your host list and the values are the names from script_names currently. Then I'd have each task perform its run commands in parallel as usual, looking up values using fabric.api.env.host_string within the task. The execution layer of fabric 2.x does not yet support this use-case afaik. This was my attempt at hacking it in, but the author rightly pointed out that this functionality should be handled in an Executor, which I could not come up with a solution for at the time: https://github.com/fabric/fabric/pull/1595