Skip to content
Automatically restart Unicorn workers based on 1) max number of requests and 2) max memory
Ruby
Branch: master
Clone or download

Latest commit

kzk Merge pull request #58 from faizalzakaria/master
Require 'unicorn' library
Latest commit bdec225 Nov 30, 2015

Files

Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
lib/unicorn Require 'unicorn' library Nov 21, 2015
AUTHORS v0.4.0 Kill processes from a thread Mar 12, 2013
ChangeLog v0.4.4 Nov 13, 2015
Gemfile use https for rubygems.org Feb 26, 2013
LICENSE initial commit Nov 18, 2012
README.md Update README.md Nov 13, 2015
Rakefile initial commit Nov 18, 2012
VERSION v0.4.4 Nov 13, 2015
unicorn-worker-killer.gemspec Update unicorn in gemspec Nov 1, 2015

README.md

unicorn-worker-killer

Unicorn is widely used HTTP-server for Rack applications. One thing we thought Unicorn missed, is killing the Unicorn workers based on the number of requests and consumed memories.

unicorn-worker-killer gem provides automatic restart of Unicorn workers based on 1) max number of requests, and 2) process memory size (RSS), without affecting any requests. This will greatly improves site's stability by avoiding unexpected memory exhaustion at the application nodes.

Install

No external process like god is required. Just install one gem: unicorn-worker-killer.

gem 'unicorn-worker-killer'

Usage

Add these lines to your config.ru. (These lines should be added above the require ::File.expand_path('../config/environment', __FILE__) line.

# Unicorn self-process killer
require 'unicorn/worker_killer'

# Max requests per worker
use Unicorn::WorkerKiller::MaxRequests, 3072, 4096

# Max memory size (RSS) per worker
use Unicorn::WorkerKiller::Oom, (192*(1024**2)), (256*(1024**2))

This gem provides two modules.

Unicorn::WorkerKiller::MaxRequests(max_requests_min=3072, max_requests_max=4096, verbose=false)

This module automatically restarts the Unicorn workers, based on the number of requests which worker processed.

max_requests_min and max_requests_max specify the min and max of maximum requests per worker. The actual limit is decided by rand() between max_requests_min and max_requests_max per worker, to prevent all workers to be dead at the same time. Once the number exceeds the limit, that worker is automatically restarted.

If verbose is set to true, then after every request, your log will show the requests left before restart. This logging is done at the info level.

Unicorn::WorkerKiller::Oom(memory_limit_min=(1024\*\*3), memory_limit_max=(2\*(1024\*\*3)), check_cycle = 16, verbose = false)

This module automatically restarts the Unicorn workers, based on its memory size.

memory_limit_min and memory_limit_max specify the min and max of maximum memory in bytes per worker. The actual limit is decided by rand() between memory_limit_min and memory_limit_max per worker, to prevent all workers to be dead at the same time. Once the memory size exceeds memory_size, that worker is automatically restarted.

The memory size check is done in every check_cycle requests.

If verbose is set to true, then every memory size check will be shown in your logs. This logging is done at the info level.

Special Thanks

You can’t perform that action at this time.