Option to kill other resque-pool instances on startup

Adds the `--kill-others` command line option.
When this option is set, resque-pool will send a graceful
shutdown signal to all other running resque-pools.

This is useful in "no downtime deploy" scenarios,
where you want the pool running the new code to
completely load its environment and be ready to
process work *before* shutting down the old code.
Once the new code is ready, it will shut down the
old processes for you.

See also resque#132

Author: joshuaflanagan

lib/resque/pool.rb

@@ -88,6 +88,11 @@ def self.handle_winch=(bool)
       @handle_winch = bool
     end
 
+    def self.kill_other_pools!
+      require 'resque/pool/killer'
+      Resque::Pool::Killer.run
+    end
+
     def self.single_process_group=(bool)
       ENV["RESQUE_SINGLE_PGRP"] = !!bool ? "YES" : "NO"
     end
@@ -98,6 +103,7 @@ def self.single_process_group
     end
 
     def self.run
+      kill_other_pools! if kill_other_pools
       if GC.respond_to?(:copy_on_write_friendly=)
         GC.copy_on_write_friendly = true
       end
@@ -235,6 +241,7 @@ def handle_sig_queue!
 
     class << self
       attr_accessor :term_behavior
+      attr_accessor :kill_other_pools
     end
 
     def graceful_worker_shutdown_and_wait!(signal)

lib/resque/pool/cli.rb

@@ -37,6 +37,7 @@ def parse_options
           opt.on('-c', '--config PATH', "Alternate path to config file") { |c| opts[:config] = c }
           opt.on('-a', '--appname NAME', "Alternate appname") { |c| opts[:appname] = c }
           opt.on("-d", '--daemon', "Run as a background daemon") { opts[:daemon] = true }
+          opt.on("-k", '--kill-others', "Shutdown any other Resque Pools on startup") { opts[:killothers] = true }
           opt.on('-o', '--stdout FILE', "Redirect stdout to logfile") { |c| opts[:stdout] = c }
           opt.on('-e', '--stderr FILE', "Redirect stderr to logfile") { |c| opts[:stderr] = c }
           opt.on('--nosync', "Don't sync logfiles on every write") { opts[:nosync] = true }
@@ -130,6 +131,7 @@ def set_pool_options(opts)
         if opts[:spawn_delay]
           Resque::Pool.spawn_delay = opts[:spawn_delay] * 0.001
         end
+        Resque::Pool.kill_other_pools = !!opts[:killothers]
       end
 
       def setup_environment(opts)

lib/resque/pool/killer.rb

@@ -0,0 +1,41 @@
+require 'open3'
+
+module Resque
+  class Pool
+    class Killer
+      include Logging
+
+      GRACEFUL_SHUTDOWN_SIGNAL=:INT
+
+      def self.run
+        new.run
+      end
+
+      def run
+        my_pid = Process.pid
+        pool_pids = all_resque_pool_processes
+        pids_to_kill = pool_pids.reject{|pid| pid == my_pid}
+        pids_to_kill.each do |pid|
+          log "Pool (#{my_pid}) in kill-others mode: killing pool with pid (#{pid})"
+          Process.kill(GRACEFUL_SHUTDOWN_SIGNAL, pid)
+        end
+      end
+
+
+      def all_resque_pool_processes
+        out, err, status = Open3.capture3("ps ax")
+        unless status.success?
+          raise "Unable to identify other pools: #{err}"
+        end
+        parse_pids_from_output out
+      end
+
+      def parse_pids_from_output(output)
+        pool_lines = output.split("\n").grep(/resque-pool-master/)
+        pool_lines.map{|line|
+          line.split.first.to_i
+        }.select{|pid| pid > 0}
+      end
+    end
+  end
+end

spec/killer_spec.rb

@@ -0,0 +1,41 @@
+require 'spec_helper'
+require 'resque/pool/killer'
+
+describe Resque::Pool::Killer do
+  subject(:killer) { Resque::Pool::Killer.new }
+  describe "parse pids from output" do
+    it "returns the first column, as integer, for resque pool processes" do
+      pids = killer.parse_pids_from_output(PS_OUTPUT)
+      pids.should match_array [6502, 16800]
+    end
+  end
+
+  PS_OUTPUT= <<END
+ 6472 ?        S      0:00 hald-addon-acpi: listening on acpid socket /var/run/acpid.socket
+ 6502 ?        Sl     2:51 resque-pool-master[demo]: managing [7028, 7034]
+ 6511 ?        Ss     0:00 xinetd -stayalive -pidfile /var/run/xinetd.pid
+ 6539 pts/18   Sl+    0:31 ruby /var/www/shipit/shared/bundle/ruby/2.1.0/bin/rails console
+20971 ?        Sl    91:51 resque-1.25.2: Waiting for ecwid
+14817 pts/15   Ss     0:00 -bash
+16334 ?        Ss     0:00 sshd: foo [priv]
+16336 ?        S      0:00 sshd: foo@pts/0
+16337 pts/0    Ss     0:00 -bash
+16342 pts/7    S+     0:00 /bin/bash /usr/local/bin/console
+16480 pts/7    Sl+    0:30 ruby /var/www/foo/shared/bundle/ruby/2.1.0/bin/rails console
+16486 pts/0    R+     0:00 ps ax
+16800 ?        Sl     5:21 resque-pool-master[demo]: managing [20728, 20734]
+17239 pts/19   Ss     0:00 -bash
+17407 pts/19   S+     0:00 /bin/bash /usr/local/bin/console
+17549 pts/19   Sl+    3:37 ruby /var/www/foo/shared/bundle/ruby/2.1.0/bin/rails console
+18027 ?        Ss     0:00 rpc.statd -p 32765 -o 32766
+18114 ?        S<sl  15:32 auditd
+19536 ?        Sl     2:05 resque-scheduler-4.0.0[production]: Processing Delayed Items
+20004 pts/9    Ss     0:00 -bash
+20728 ?        Sl   165:24 resque-1.25.2: Waiting for queuea,queueb
+20734 ?        Sl   164:55 resque-1.25.2: Waiting for queueb,queuea
+20799 pts/10   S+     0:00 /bin/bash /usr/local/bin/console
+20959 pts/10   Sl+    0:31 ruby /var/www/foo/shared/bundle/ruby/2.1.0/bin/rails console
+20967 ?        Sl   332:07 resque-1.25.2: queuec,queued,*
+END
+
+end