Add a "stopped" JobStatus and the machinery to properly handle it (#1394)

* Add a "stopped" JobStatus and the machinery to properly handle it

fixes #1389

* Apply requested changes

docs/docs/jobs.md

@@ -99,7 +99,7 @@ print('Status: %s' % job.get_status())
 ```
 
 Some interesting job attributes include:
-* `job.get_status()` Possible values are `queued`, `started`, `deferred`, `finished`, and `failed`
+* `job.get_status()` Possible values are `queued`, `started`, `deferred`, `finished`, `stopped`, and `failed`
 * `job.origin` queue name of this job
 * `job.func_name`
 * `job.args` arguments passed to the underlying job function
@@ -136,6 +136,8 @@ redis = Redis()
 send_stop_job_command(redis, job_id)
 ```
 
+Unlike failed jobs, stopped jobs will *not* be automatically retried if retry is configured. Subclasses of `Worker` which override `handle_job_failure()` should likewise take care to handle jobs with a `stopped` status appropriately.
+
 ## Job / Queue Creation with Custom Serializer
 
 When creating a job or queue, you can pass in a custom serializer that will be used for serializing / de-serializing job arguments.

rq/command.py

@@ -72,6 +72,9 @@ def handle_stop_job_command(worker, payload):
     job_id = payload.get('job_id')
     worker.log.debug('Received command to stop job %s', job_id)
     if job_id and worker.get_current_job_id() == job_id:
+        # Sets the '_stopped_job_id' so that the job failure handler knows it
+        # was intentional.
+        worker._stopped_job_id = job_id
         worker.kill_horse()
     else:
         worker.log.info('Not working on job %s, command ignored.', job_id)

rq/job.py

@@ -35,6 +35,7 @@ JobStatus = enum(
     STARTED='started',
     DEFERRED='deferred',
     SCHEDULED='scheduled',
+    STOPPED='stopped',
 )
 
 # Sentinel value to mark that some of our lazily evaluated properties have not
@@ -175,6 +176,10 @@ class Job(object):
     def is_scheduled(self):
         return self.get_status() == JobStatus.SCHEDULED
 
+    @property
+    def is_stopped(self):
+        return self.get_status() == JobStatus.STOPPED
+
     @property
     def _dependency_id(self):
         """Returns the first item in self._dependency_ids. Present to

rq/worker.py

@@ -208,6 +208,8 @@ class Worker(object):
         self._is_horse = False
         self._horse_pid = 0
         self._stop_requested = False
+        self._stopped_job_id = None
+
         self.log = logger
         self.log_job_description = log_job_description
         self.last_cleaned_at = None
@@ -797,8 +799,14 @@ class Worker(object):
         job_status = job.get_status()
         if job_status is None:  # Job completed and its ttl has expired
             return
-        if job_status not in [JobStatus.FINISHED, JobStatus.FAILED]:
-
+        elif job_status == JobStatus.STOPPED:
+            # Work-horse killed deliberately
+            self.log.warning('Job stopped by user, moving job to FailedJobRegistry')
+            self.handle_job_failure(
+                job, queue=queue,
+                exc_string="Job stopped by user, work-horse terminated."
+            )
+        elif job_status not in [JobStatus.FINISHED, JobStatus.FAILED]:
             if not job.ended_at:
                 job.ended_at = utcnow()
 
@@ -895,13 +903,18 @@ class Worker(object):
                     job_class=self.job_class
                 )
             job.worker_name = None
-            # Requeue/reschedule if retry is configured
-            if job.retries_left and job.retries_left > 0:
-                retry = True
-                retry_interval = job.get_retry_interval()
-                job.retries_left = job.retries_left - 1
+
+            # check whether a job was stopped intentionally and set the job
+            # status appropriately if it was this job.
+            job_is_stopped = self._stopped_job_id == job.id
+            retry = job.retries_left and job.retries_left > 0 and not job_is_stopped
+
+            if job_is_stopped:
+                job.set_status(JobStatus.STOPPED, pipeline=pipeline)
+                self._stopped_job_id = None
             else:
-                retry = False
+                # Requeue/reschedule if retry is configured, otherwise
+                if not retry:
                     job.set_status(JobStatus.FAILED, pipeline=pipeline)
 
             started_job_registry.remove(job, pipeline=pipeline)
@@ -920,6 +933,8 @@ class Worker(object):
                 )
 
             if retry:
+                retry_interval = job.get_retry_interval()
+                job.retries_left = job.retries_left - 1
                 if retry_interval:
                     scheduled_datetime = datetime.now(timezone.utc) + timedelta(seconds=retry_interval)
                     job.set_status(JobStatus.SCHEDULED)

tests/test_commands.py

@@ -92,6 +92,10 @@ class TestCommands(RQTestCase):
 
         send_stop_job_command(connection, job_id=job.id)
         time.sleep(0.25)
+
+        # Job status is set appropriately
+        self.assertTrue(job.is_stopped)
+
         # Worker has stopped working
         worker.refresh()
         self.assertEqual(worker.get_state(), WorkerStatus.IDLE)