Inspired by ( https://github.com/schneems/puma_worker_killer ). Idea is to run separate process as a daemon to measure puma memory and restart worker when memory threshold reached.
Add this line to your application's Gemfile:
gem 'puma_doctor'
And then execute:
$ bundle
Or install it yourself as:
$ gem install puma_doctor
To run from your code:
PumaDoctor.start(frequency: 60, memory_threshold: 2000, puma_pid: 99999)
This is not very useful in production since it blocks execution, but you can play around with options locally. Available options with defaults are:
frequency: 60 # Interval in seconds
puma_pid_file: 'puma.pid' # Location of puma pid file
memory_threshold: 4000 # Amount in MB
log_file: 'puma_doctor.log' # Name and location of log file
To run as daemon you can create file with content below(Ex.: doctor.rb)
require 'puma_doctor'
require 'daemons'
pid_dir = '../' # Path to directory to store pid.
Daemons.run_proc('puma_doctor', { dir: pid_dir }) do
PumaDoctor.start(frequency: 60, memory_threshold: 1000)
end
Then control it with(for more details visit https://github.com/thuehlinger/daemons):
bundle exec ruby doctor.rb start
bundle exec ruby doctor.rb stop
bundle exec ruby doctor.rb restart
Probably the easiest way to run puma_doctor in production is to use capistrano. Require script in Capfile:
require 'puma_doctor/capistrano'
This will add hook to start/restart daemon on after deploy:finished. If you want to start/stop from capistrano manually - this tasks are available:
cap puma_doctor:check # Check if config file exixts on server
cap puma_doctor:config # Config daemon
cap puma_doctor:restart # Restart daemon
cap puma_doctor:start # Start daemon
cap puma_doctor:stop # Stop daemon
Available options with defaults:
set :puma_doctor_pid, -> { File.join(shared_path, 'tmp', 'pids', 'puma_doctor.pid') }
set :puma_doctor_frequency, 30 #seconds
set :puma_doctor_memory_threshold, 4000 #mb
set :puma_doctor_daemon_file, -> { File.join(shared_path, 'puma_doctor_daemon.rb') }
set :puma_doctor_log_file, -> { File.join(shared_path, 'log', 'puma_doctor.log') }
set :puma_pid, -> { File.join(shared_path, 'tmp', 'pids', 'puma.pid') }
You can always see what puma_doctor is doing by reading logs.
Test
- Fork it ( http://github.com/spilin/puma_doctor/fork )
- Create your feature branch (
git checkout -b my-new-feature) - Commit your changes (
git commit -am 'Add some feature') - Push to the branch (
git push origin my-new-feature) - Create new Pull Request