ResqueSliders is a Resque plugin which allows you to control Resque workers from the Web-UI.
From the Resque-Web UI, you can:
- Start workers with any queue, or combination of queues on any host, and specify how many of each should be running
- Pause / Stop / Restart ALL running workers
ResqueSliders comes in two parts:
KEWatcher: A daemon that runs on any machine that needs to run Resque workers, watches over the workers and controls which ones are running
Resque-Web Plugin: A bunch of slider bars, with text-input box to specify what queues to run on the workers
Install as a gem:
$ gem install resque-sliders
This is the daemon component that runs on any host that you want to run Resque workers on. The daemon's job is to manage how many Resque workers should be running, and what they should be running. It also provides an easy way to stop all workers during maintenance or deploys.
When the daemon first runs, it will register itself, by hostname with Redis:
- Adds a few persistent settings to the hash key
- Gets any queues that need to be running on the host by looking at
Usage: kewatcher [options] Options: -c, --config CONFIG Resque Config (Yaml) -r, --rakefile RAKEFILE Rakefile location -p, --pidfile PIDFILE PID File location -f, --force FORCE KILL ANY OTHER RUNNING KEWATCHERS -v, --verbose Verbosity (Can be specified more than once, -vv) -m, --max MAX Max Children (default: 10) -w, --wait WAIT_TIME Time (in seconds) to wait for worker to die before sending TERM signal (default: 20 seconds) -t, --time MAX_TIME Max Time (in seconds) to wait for worker to die before sending KILL (-9) signal (FORCE QUIT) (default: 60) NOTE: With Resque >= 1.22.0 force quit is handled for you so by default this is the same as: RESQUE_TERM_TIMEOUT=40 or the difference of MAX_TIME and WAIT_TIME more info: http://hone.heroku.com/resque/2012/08/21/resque-signals.html -a, --async Do NOT wait for Resque workers to die completely before spawning new workers (default: false) -V, --version Prints Version
-m|--max MAX (Max Children): Maximum number of workers to run on host (default: 10) -w|--wait WAIT_TIME (Wait Time): How long to wait before sending TERM to zombies (default: 20 seconds) -t|--time TIME (Total Time): How long to wait before sending KILL to zombies (default: 60 seconds) NOTE: Resque >= 1.22.0 includes signal handling of its own to force quit, so we use it if its there, and override with our own timeout here -a|--async (Async): Should we spawn new workers before old ones have fully terminated (default: false) -r|--rakefile RAKEFILE (Rakefile): Pass along a rakefile to use when calling rake ... resque:work - shouldn't be needed if run from project directory -f|--force (Force): Force any currently running KEWatcher processes to quit, waiting for it to do so, and starting in its place RAILS_ENV: If you're using rails, you need to set your RAILS_ENV variable
Controlling the Daemon
Once the daemon is running on each host that is going to run Resque workers, you'll need to tell them which queues to run.
KEWatcher supports all the same signals as Resque:
QUIT- Shutdown. Gracefully kill all child Resque workers, and wait for them to finish before exiting
HUP- Restart all Resque workers by gracefully killing them, and starting new ones in their place
USR1- Stop all Resque workers, and don't start any more
USR2- Pause spawning of new queues, but leave current ones running
CONT- Unpause. Continue spawning/managing child Resque workers
The queue configuration is done via Resque-Web interface
See below for screenshots
Pause- Start or Pause
Stop- Stop all workers
Reload- Sends HUP signal to running KEWatcher
To enable the Resque-Web Integration you'll need to load ResqueSliders to enable the Sliders tab. Just add:
to a file, like resque-web_init.rb, and run resque-web:
resque-sliders has been tested on the following platforms:
- 1.8.7 (ree)
- probabaly more...
Want to fix a bug? See a new feature?