Que migration v8: que_jobs_ext view; add first_run_at column to que_jobs

FIX bulk_insert_jobs and insert_jobs don't pass redundant arg for first_run_at (use run_at)

cleanup. specs. documentation

migrations and specs support backwards compatibility

PR change requests.

remove unnecessary db_version check in introspection spec.

Add view and job table enhancements for improved observability

docs/README.md

@@ -56,6 +56,7 @@
 - [Enqueueing jobs in bulk](#enqueueing-jobs-in-bulk)
 - [Expired jobs](#expired-jobs)
 - [Finished jobs](#finished-jobs)
+- [Que Jobs Ext view](#que-jobs-ext-view)
 
 <!-- /MarkdownTOC -->
 
@@ -879,3 +880,17 @@ SET LOCAL que.skip_notify TO true;
 DELETE FROM que_jobs WHERE finished_at < (select now() - interval '7 days');
 COMMIT;
 ```
+
+## Que Jobs Ext view
+
+This view extends the functionality of the que job management system by providing an enriched view of que jobs. It combines data from the 'que_jobs' table and the 'pg_locks' table to present a comprehensive overview of que jobs, including their status, associated information, and locking details.
+
+This view is designed to facilitate the monitoring and management of que jobs, allowing you to track job statuses, locking details, and job-related information.
+
+Additional Columns:
+
+- lock_id: Unique identifier for the lock associated with the job.
+- que_locker_pid: Process ID (PID) of the que job locker.
+- sub_class: The job class extracted from the job arguments.
+- updated_at: The most recent timestamp among 'run_at,' 'expired_at,' and 'finished_at.'
+- status: The status of the job, which can be 'running,' 'completed,' 'failed,' 'errored,' 'queued,' or 'scheduled.'

lib/que/job.rb

@@ -12,7 +12,7 @@ class Job
     SQL[:insert_job] =
       %{
         INSERT INTO public.que_jobs
-        (queue, priority, run_at, job_class, args, kwargs, data, job_schema_version)
+        (queue, priority, run_at, job_class, args, kwargs, data, job_schema_version, first_run_at)
         VALUES
         (
           coalesce($1, 'default')::text,
@@ -22,7 +22,8 @@ class Job
           coalesce($5, '[]')::jsonb,
           coalesce($6, '{}')::jsonb,
           coalesce($7, '{}')::jsonb,
-          #{Que.job_schema_version}
+          #{Que.job_schema_version},
+          coalesce($3, now())::timestamptz
         )
         RETURNING *
       }
@@ -33,7 +34,7 @@ class Job
           SELECT * from json_to_recordset(coalesce($5, '[{args:{},kwargs:{}}]')::json) as x(args jsonb, kwargs jsonb)
         )
         INSERT INTO public.que_jobs
-        (queue, priority, run_at, job_class, args, kwargs, data, job_schema_version)
+        (queue, priority, run_at, job_class, args, kwargs, data, job_schema_version, first_run_at)
         SELECT
           coalesce($1, 'default')::text,
           coalesce($2, 100)::smallint,
@@ -42,7 +43,8 @@ class Job
           args_and_kwargs.args,
           args_and_kwargs.kwargs,
           coalesce($6, '{}')::jsonb,
-          #{Que.job_schema_version}
+          #{Que.job_schema_version},
+          coalesce($3, now())::timestamptz
         FROM args_and_kwargs
         RETURNING *
       }
@@ -75,7 +77,8 @@ class << self
         :maximum_retry_count,
         :queue,
         :priority,
-        :run_at
+        :run_at,
+        :first_run_at
 
       def enqueue(*args)
         args, kwargs = Que.split_out_ruby2_keywords(args)

lib/que/migrations.rb

@@ -4,7 +4,7 @@ module Que
   module Migrations
     # In order to ship a schema change, add the relevant up and down sql files
     # to the migrations directory, and bump the version here.
-    CURRENT_VERSION = 7
+    CURRENT_VERSION = 8
 
     class << self
       def migrate!(version:)

lib/que/migrations/8/down.sql

@@ -0,0 +1,4 @@
+DROP VIEW IF EXISTS public.que_jobs_ext;
+
+ALTER TABLE que_jobs
+  DROP COLUMN first_run_at;

lib/que/migrations/8/up.sql

@@ -0,0 +1,61 @@
+-- Add 'first_run_at' column to 'que_jobs'. The 'first_run_at' column will store the timestamp with time zone (timestamptz) 
+-- of the initial scheduled execution time for que jobs. The column default is set to now, but realistically this will always 
+-- be defaulted to the initial run_at value. This enhancement helps re-trace the execution of the job when reviewing failures.
+
+ALTER TABLE que_jobs
+  ADD COLUMN first_run_at timestamptz NOT NULL DEFAULT now();
+
+
+-- This view extends the functionality of the que job management system by providing an enriched view of que jobs. 
+-- It combines data from the 'que_jobs' table and the 'pg_locks' table to present a comprehensive overview of que jobs, 
+-- including their status, associated information, and locking details. The view is designed to facilitate the monitoring
+--  and management of que jobs, allowing you to track job statuses, locking details, and job-related information.
+
+-- Columns:
+-- - lock_id: Unique identifier for the lock associated with the job.
+-- - que_locker_pid: Process ID (PID) of the que job locker.
+-- - sub_class: The job class extracted from the job arguments.
+-- - updated_at: The most recent timestamp among 'run_at,' 'expired_at,' and 'finished_at.'
+-- - status: The status of the job, which can be 'running,' 'completed,' 'failed,' 'errored,' 'queued,' or 'scheduled.'
+
+CREATE OR REPLACE VIEW public.que_jobs_ext
+AS
+SELECT
+    locks.id AS lock_id,
+    locks.pid as que_locker_pid,
+    (que_jobs.args -> 0) ->> 'job_class'::text AS sub_class,
+    greatest(run_at, expired_at, finished_at) as updated_at,
+
+    case
+      when locks.id is not null then 'running'
+      when finished_at is not null then 'completed'
+      when expired_at is not null then 'failed'
+      when error_count > 0 then 'errored'
+      when run_at < now() then 'queued'
+      else 'scheduled'
+    end as status,
+
+    -- que_jobs.*:
+    que_jobs.id,
+    que_jobs.priority,
+    que_jobs.run_at,
+    que_jobs.first_run_at,
+    que_jobs.job_class,
+    que_jobs.error_count,
+    que_jobs.last_error_message,
+    que_jobs.queue,
+    que_jobs.last_error_backtrace,
+    que_jobs.finished_at,
+    que_jobs.expired_at,
+    que_jobs.args,
+    que_jobs.data,
+    que_jobs.kwargs,
+    que_jobs.job_schema_version
+
+  FROM que_jobs
+    LEFT JOIN (
+      SELECT
+        (classid::bigint << 32) + objid::bigint AS id
+        , pid
+          FROM  pg_locks
+          WHERE pg_locks.locktype = 'advisory'::text) locks USING (id);

lib/que/utils/queue_management.rb

@@ -12,7 +12,7 @@ def clear!
       # Very old migrations may use Que.create! and Que.drop!, which just
       # created and dropped the initial version of the jobs table.
       def create!; migrate!(version: 1); end
-      def drop!;   migrate!(version: 0); end
+      def drop!; migrate!(version: 0); end
     end
   end
 end

spec/que/job.bulk_enqueue_spec.rb

@@ -44,6 +44,7 @@ def assert_enqueue(
       assert_equal expected_job_class.to_s, job[:job_class]
       assert_equal expected_args[i], job[:args]
       assert_equal expected_kwargs[i], job[:kwargs]
+      assert_equal job[:run_at], job[:first_run_at]
     end
 
     jobs_dataset.delete

spec/que/job.enqueue_spec.rb

@@ -39,6 +39,7 @@ def assert_enqueue(
     assert_equal expected_job_class.to_s, job[:job_class]
     assert_equal expected_args, job[:args]
     assert_equal expected_job_schema_version, job[:job_schema_version]
+    assert_equal job[:first_run_at], job[:run_at]
 
     jobs_dataset.delete
   end

spec/que/utils/introspection_spec.rb

@@ -65,10 +65,10 @@
       locker.stop!
 
       state = states.first
-      assert_equal \
-        %i(priority run_at id job_class error_count last_error_message queue
-          last_error_backtrace finished_at expired_at args data job_schema_version kwargs ruby_hostname ruby_pid),
-        state.keys
+      expected_keys = %i(priority run_at id job_class error_count last_error_message queue
+      last_error_backtrace finished_at expired_at args data job_schema_version kwargs ruby_hostname ruby_pid first_run_at)
+
+      assert_equal expected_keys.sort, state.keys.sort
 
       assert_equal 2, state[:priority]
       assert_in_delta state[:run_at], Time.now, QueSpec::TIME_SKEW

spec/que/views/que_jobs_ext_spec.rb

@@ -0,0 +1,199 @@
+# frozen_string_literal: true
+
+require 'spec_helper'
+
+describe "Que Jobs Ext View", skip: true do
+
+  class TestJob < Que::Job
+    include Que::JobMethods
+
+    def default_resolve_action
+      # prevents default deletion of complete jobs for testing purposes
+      finish
+    end
+  end
+
+  class TestFailedJob < TestJob
+    def run 
+      raise Que::Error, 'Test Error'
+    end
+  end
+
+  describe 'job.enqueue' do 
+    it "should mirror enqueued job" do
+      assert_equal 0, jobs_dataset.count
+      assert_equal 0, jobs_ext_dataset.count
+
+      TestJob.enqueue(
+          1,
+          'two',
+          string: "string",
+          integer: 5,
+          array: [1, "two", {three: 3}],
+          hash: {one: 1, two: 'two', three: [3]},
+          job_options: { 
+            priority: 4,
+            queue: 'special_queue_name',
+            run_at: Time.now
+          }
+        )
+
+      assert_equal 1, jobs_dataset.count
+      assert_equal 1, jobs_ext_dataset.count
+
+      job = jobs_dataset.first
+      ext_job = jobs_ext_dataset.first
+      assert_equal ext_job[:queue], job[:queue]
+      assert_equal ext_job[:priority], job[:priority]
+      assert_equal ext_job[:run_at], job[:run_at]
+      assert_equal ext_job[:first_run_at], job[:first_run_at]
+      assert_equal ext_job[:job_class], job[:job_class]
+      assert_equal ext_job[:args], job[:args]
+      assert_equal ext_job[:job_schema_version], job[:job_schema_version]   
+
+      jobs_dataset.delete
+
+      assert_equal 0, jobs_dataset.count
+      assert_equal 0, jobs_ext_dataset.count
+    end
+
+    it "should include additional lock data" do
+      locker_settings.clear
+      locker_settings[:listen] = false
+      locker_settings[:poll_interval] = 0.02
+      locker
+
+      TestJob.enqueue
+
+      sleep_until { locked_ids.count.positive? && locked_ids.first == jobs_ext_dataset.first[:lock_id] }
+
+      locker.stop!
+
+      jobs_dataset.delete
+    end
+
+    it "should add additional updated_at" do
+      TestJob.enqueue
+
+      ext_job = jobs_ext_dataset.first
+
+      assert_equal ext_job[:run_at], ext_job[:updated_at]
+
+      locker
+
+      sleep_until_equal(1) { finished_jobs_dataset.count }
+
+      locker.stop!
+
+      ext_job = jobs_ext_dataset.first
+
+      assert_equal ext_job[:finished_at], ext_job[:updated_at]
+
+      jobs_dataset.delete
+    end
+
+    describe "should include additional status" do
+
+      let(:notified_errors) { [] }
+
+      it "should set status to scheduled when run_at is in the future" do
+        TestJob.enqueue(job_options: { run_at: Time.now + 1 })
+
+        assert_equal jobs_ext_dataset.first[:status], 'scheduled'
+
+        jobs_dataset.delete
+      end
+
+      it "should set status to queued when run_at is in the past and the job is not currently running, completed, failed or errored" do
+        TestJob.enqueue(job_options: { run_at: Time.now - 1 })
+
+        assert_equal jobs_ext_dataset.first[:status], 'queued'
+
+        jobs_dataset.delete
+      end
+
+      it "should set status to running when the job has a lock associated with it" do  
+        locker_settings.clear
+        locker_settings[:listen] = false
+        locker_settings[:poll_interval] = 0.02
+        locker
+
+        TestJob.enqueue
+
+        sleep_until { locked_ids.count.positive? && locked_ids.first == jobs_ext_dataset.first[:lock_id] && jobs_ext_dataset.first[:status] == 'running' }
+
+        locker.stop!
+
+        jobs_dataset.delete
+      end
+
+      it "should set status to complete when finished_at is present" do
+        TestJob.enqueue
+
+        locker
+
+        sleep_until_equal(1) { DB[:que_lockers].count }
+
+        sleep_until { finished_jobs_dataset.count.positive? }
+
+        locker.stop!
+
+        assert_equal jobs_ext_dataset.first[:status], 'completed'
+
+        jobs_dataset.delete
+      end
+
+      it "should set status to errored when error_count is positive and expired_at is not present" do
+        Que.error_notifier = proc { |e| notified_errors << e }
+
+        TestFailedJob.class_eval do 
+          self.maximum_retry_count = 100 # prevent from entering failed state on first error
+        end
+
+        locker
+
+        sleep_until_equal(1) { DB[:que_lockers].count }
+
+        TestFailedJob.enqueue
+
+        sleep_until { errored_jobs_dataset.where(expired_at: nil).count.positive? }
+
+        locker.stop!
+
+        ext_job = jobs_ext_dataset.first
+
+        assert_equal ext_job[:status], 'errored'
+        assert_equal notified_errors.count, 1
+        assert_equal notified_errors.first.message, 'Test Error'
+
+
+        jobs_dataset.delete
+      end
+
+      it "should set status to failed when expired_at is present" do
+        TestFailedJob.class_eval do 
+          self.maximum_retry_count = 0
+        end
+
+        Que.error_notifier = proc { |e| notified_errors << e }
+
+        locker
+
+        sleep_until_equal(1) { DB[:que_lockers].count }
+
+        TestFailedJob.enqueue
+
+        sleep_until { expired_jobs_dataset.count.positive? }
+
+        locker.stop!
+
+        assert_equal jobs_ext_dataset.first[:status], 'failed'
+        assert_equal notified_errors.count, 1
+        assert_equal notified_errors.first.message, 'Test Error'
+
+
+        jobs_dataset.delete
+      end
+    end
+  end
+end