feat: avoided "zombie" processes after killing work horse (#1348)

* feat: avoided "zombie" processes after killing work horse by setting work horse process group and killing this group

* fixed tests

* tests: added test to check that all workhorse subprocesses are killed

* tests: updated guthub run tests dependencies since they are not using (dev-)requirements.txt

Co-authored-by: Ruslan Mullakhmetov <ruslan@twentythree.net>

.github/workflows/workflow.yml

@@ -33,7 +33,7 @@ jobs:
       run: |
         python -m pip install --upgrade pip
         pip install redis==${{ matrix.redis-py-version }}
-        pip install pytest pytest-cov sentry-sdk codecov mock
+        pip install pytest pytest-cov sentry-sdk codecov mock psutil
         pip install -e .
 
     - name: Test with pytest

dev-requirements.txt

@@ -1,2 +1,3 @@
 mock
 pytest
+psutil

rq/worker.py

@@ -394,7 +394,7 @@ class Worker(object):
         Kill the horse but catch "No such process" error has the horse could already be dead.
         """
         try:
-            os.kill(self.horse_pid, sig)
+            os.killpg(os.getpgid(self.horse_pid), sig)
             self.log.info('Killed horse pid %s', self.horse_pid)
         except OSError as e:
             if e.errno == errno.ESRCH:
@@ -699,6 +699,7 @@ class Worker(object):
         os.environ['RQ_WORKER_ID'] = self.name
         os.environ['RQ_JOB_ID'] = job.id
         if child_pid == 0:
+            os.setsid()
             self.main_work_horse(job, queue)
             os._exit(0) # just in case
         else:

tests/fixtures.py

@@ -10,6 +10,7 @@ import os
 import time
 import signal
 import sys
+import subprocess
 
 from rq import Connection, get_current_job, get_current_connection, Queue
 from rq.decorators import job
@@ -66,6 +67,17 @@ def create_file_after_timeout(path, timeout):
     time.sleep(timeout)
     create_file(path)
 
+def create_file_after_timeout_and_setsid(path, timeout):
+    os.setsid()
+    create_file_after_timeout(path, timeout)
+
+def launch_process_within_worker_and_store_pid(path, timeout):
+
+    p = subprocess.Popen(['sleep', str(timeout)])
+    with open(path, 'w') as f:
+        f.write('{}'.format(p.pid))
+
+    p.wait()
 
 def access_self():
     assert get_current_connection() is not None

tests/test_worker.py

@@ -4,6 +4,7 @@ from __future__ import (absolute_import, division, print_function,
 
 import json
 import os
+import psutil
 import shutil
 import signal
 import subprocess
@@ -23,9 +24,10 @@ from mock import Mock
 
 from tests import RQTestCase, slow
 from tests.fixtures import (
-    access_self, create_file, create_file_after_timeout, div_by_zero, do_nothing,
+    access_self, create_file, create_file_after_timeout, create_file_after_timeout_and_setsid, div_by_zero, do_nothing,
     kill_worker, long_running_job, modify_self, modify_self_and_error,
-    run_dummy_heroku_worker, save_key_ttl, say_hello, say_pid, raise_exc_mock
+    run_dummy_heroku_worker, save_key_ttl, say_hello, say_pid, raise_exc_mock,
+    launch_process_within_worker_and_store_pid
 )
 
 from rq import Queue, SimpleWorker, Worker, get_current_connection
@@ -37,7 +39,6 @@ from rq.utils import utcnow
 from rq.version import VERSION
 from rq.worker import HerokuWorker, WorkerStatus
 
-
 class CustomJob(Job):
     pass
 
@@ -1166,12 +1167,16 @@ class WorkerShutdownTestCase(TimeoutTestCase, RQTestCase):
         sentinel_file = '/tmp/.rq_sentinel_work_horse_death'
         if os.path.exists(sentinel_file):
             os.remove(sentinel_file)
-        fooq.enqueue(create_file_after_timeout, sentinel_file, 100)
+        fooq.enqueue(launch_process_within_worker_and_store_pid, sentinel_file, 100)
         job, queue = w.dequeue_job_and_maintain_ttl(5)
         w.fork_work_horse(job, queue)
         job.timeout = 5
         w.job_monitoring_interval = 1
         now = utcnow()
+        time.sleep(1)
+        with open(sentinel_file) as f:
+            subprocess_pid = int(f.read().strip())
+        self.assertTrue(psutil.pid_exists(subprocess_pid))
         w.monitor_work_horse(job, queue)
         fudge_factor = 1
         total_time = w.job_monitoring_interval + 65 + fudge_factor
@@ -1180,6 +1185,7 @@ class WorkerShutdownTestCase(TimeoutTestCase, RQTestCase):
         failed_job_registry = FailedJobRegistry(queue=fooq)
         self.assertTrue(job in failed_job_registry)
         self.assertEqual(fooq.count, 0)
+        self.assertFalse(psutil.pid_exists(subprocess_pid))
 
 
 def schedule_access_self():
@@ -1283,9 +1289,10 @@ class HerokuWorkerShutdownTestCase(TimeoutTestCase, RQTestCase):
         w = HerokuWorker('foo')
 
         path = os.path.join(self.sandbox, 'shouldnt_exist')
-        p = Process(target=create_file_after_timeout, args=(path, 2))
+        p = Process(target=create_file_after_timeout_and_setsid, args=(path, 2))
         p.start()
         self.assertEqual(p.exitcode, None)
+        time.sleep(0.1)
 
         w._horse_pid = p.pid
         w.handle_warm_shutdown_request()