Command an armada of processes in a cluster.
This module integrates with git using propagit, to make rolling out new code to a bunch of worker drones super simple!
With npm do:
npm install -g fleet
In a fresh directory, start a fleet hub:
$ fleet hub --port=7000 --secret=beepboop
in another fresh directory start a drone:
$ fleet drone --hub=localhost:7000 --secret=beepboop
You can start as many drones as you want in fresh directories.
Now from a git repo, set a remote:
$ fleet remote add default --hub=localhost:7000 --secret=beepboop
Now deploy your code to all the drones:
$ fleet deploy
Deploying just checks out your commit to a fresh directory on every drone.
To run a process on the latest commit do:
$ fleet spawn -- node beep.js 8080
To see the process list across all your drones do:
$ fleet ps
This command creates and updates a file called
fleet.json in your git project
root so you don't need to type --hub and --secret all the time for the deploy,
spawn, stop, and ps commands.
fleet remote add NAME=default --hub=HUB --secret=SECRET
Register a remote with NAME and OPTIONS.
fleet remote rm NAME
Delete a remote.
fleet remote ls
List the remotes from your
fleet hub --port=PORT --secret=SECRET
Create a new hub listening on a port with a secret passphrase.
fleet drone --hub=HOST:PORT --secret=SECRET
Spin up a worker drone and connect to the hub.
From a git directory, push the top commit from
git log and check out the code
on all the drones in an isolated directory based on the commit hash.
List the processes running on all the drones.
fleet spawn OPTIONS -- command...
command on a random drone.
You can control which drone by specifying
Your spawned process will get the $REPO name, $COMMIT hash, $DRONE_ID, and $PROCESS_ID as environment variables from propagit.
fleet stop id
Stop a process. Get the
id by running
fleet ps. It doesn't matter which
drone the process is running on, fleet spams the request out to all the drones
and ignores the request if the drone doesn't have the requested process id.
Work in progress. This command will let you listen to all the events ever across the whole cluster.