Configuration for maximum number of job execution threads

README.md

@@ -67,6 +67,17 @@ $ bundle
     ```bash
     $ bundle exec good_job
     ```
+
+### Configuring Job Execution Threads
+
+GoodJob’s executes enqueued jobs using threads. There is a lot than can be said about [multithreaded behavior in Ruby on Rails](https://guides.rubyonrails.org/threading_and_code_execution.html), but briefly:
+
+- Each GoodJob thread requires its own database connection, which are automatically checked out from Rails’s connection pool. _Allowing GoodJob to schedule more threads than are available in the database connection pool can lead to timeouts and is not recommended._ 
+- The maximum number of GoodJob threads can be configured, in decreasing precedence:
+    1. `$ bundle exec good_job --max_threads 4`
+    2. `$ GOOD_JOB_MAX_THREADS=4 bundle exec good_job`
+    3. `$ RAILS_MAX_THREADS=4 bundle exec good_job`
+    4. Implicitly via Rails's database connection pool size (`ActiveRecord::Base.connection_pool.size`)
 
 ## Development
 

lib/good_job/cli.rb

@@ -5,18 +5,23 @@ class CLI < Thor
     RAILS_ENVIRONMENT_RB = File.expand_path("config/environment.rb")
 
     desc :start, "Start jobs"
+    method_option :max_threads, type: :numeric
     def start
       require RAILS_ENVIRONMENT_RB
 
-      scheduler = GoodJob::Scheduler.new
+      max_threads = options[:max_threads] ||
+                    ENV['GOOD_JOB_MAX_THREADS'] ||
+                    ENV['RAILS_MAX_THREADS'] ||
+                    ActiveRecord::Base.connection_pool.size
+
+      $stdout.puts "GoodJob starting with max_threads=#{max_threads}"
+      scheduler = GoodJob::Scheduler.new(pool_options: { max_threads: max_threads })
 
       %w[INT TERM].each do |signal|
         trap(signal) { @stop_good_job_executable = true }
       end
       @stop_good_job_executable = false
 
-      $stdout.puts "GoodJob waiting for jobs..."
-
       Kernel.loop do
         sleep 0.1
         break if @stop_good_job_executable || scheduler.shutdown?

lib/good_job/scheduler.rb

@@ -4,8 +4,6 @@
 
 module GoodJob
   class Scheduler
-    MAX_THREADS = Concurrent.processor_count
-
     DEFAULT_TIMER_OPTIONS = {
       execution_interval: 1,
       timeout_interval: 1,
@@ -15,18 +13,18 @@ class Scheduler
     DEFAULT_POOL_OPTIONS = {
       name: 'good_job',
       min_threads: 0,
-      max_threads: MAX_THREADS,
+      max_threads: Concurrent.processor_count,
       auto_terminate: true,
       idletime: 0,
       max_queue: 0,
       fallback_policy: :abort, # shouldn't matter -- 0 max queue
     }.freeze
 
-    def initialize(query = GoodJob::Job.all, **_options)
+    def initialize(query = GoodJob::Job.all, timer_options: {}, pool_options: {})
       @query = query
 
-      @pool = Concurrent::ThreadPoolExecutor.new(DEFAULT_POOL_OPTIONS)
-      @timer = Concurrent::TimerTask.new(DEFAULT_TIMER_OPTIONS) do
+      @pool = Concurrent::ThreadPoolExecutor.new(DEFAULT_POOL_OPTIONS.merge(pool_options))
+      @timer = Concurrent::TimerTask.new(DEFAULT_TIMER_OPTIONS.merge(timer_options)) do
         idle_threads = @pool.max_length - @pool.length
         create_thread if idle_threads.positive?
       end
@@ -64,20 +62,21 @@ def shutdown?
 
     def create_thread
       future = Concurrent::Future.new(args: [ordered_query], executor: @pool) do |query|
-        Rails.application.executor.wrap do
-          loop do
+        loop do
+          executed_job = false
+
+          Rails.application.executor.wrap do
             good_job = query.with_advisory_lock.first
             break unless good_job
 
+            executed_job = true
             ActiveSupport::Notifications.instrument("job_started.good_job", { good_job: good_job })
-
             JobWrapper.new(good_job).perform
-
             good_job.advisory_unlock
           end
-        end
 
-        true
+          break unless executed_job
+        end
       end
       future.add_observer(TaskObserver.new)
       future.execute

spec/good_job/cli_spec.rb

@@ -43,5 +43,21 @@
       expect(GoodJob::Scheduler).to have_received(:new)
       expect(scheduler_mock).to have_received(:shutdown)
     end
+
+    describe 'max threads' do
+      it 'defaults to --max_threads, GOOD_JOB_MAX_THREADS, RAILS_MAX_THREADS, database connection pool size' do
+        allow(Kernel).to receive(:loop)
+
+        cli = described_class.new([], { max_threads: 4 }, {})
+        stub_const 'ENV', ENV.to_hash.merge({ 'RAILS_MAX_THREADS' => 3, 'GOOD_JOB_MAX_THREADS' => 2 })
+        allow(ActiveRecord::Base.connection_pool).to receive(:size).and_return(1)
+
+        expect do
+          cli.start
+        end.to output.to_stdout
+
+        expect(GoodJob::Scheduler).to have_received(:new).with(pool_options: { max_threads: 4 })
+      end
+    end
   end
 end