Database still in use after a selenium test in Django
Asked Answered
M

3

10

I have a Django project in which I'm starting to write Selenium tests. The first one looking like this:

from django.contrib.staticfiles.testing import StaticLiveServerTestCase
from selenium import webdriver
from selenium.webdriver.common.keys import Keys

from core.models import User
from example import settings

BACH_EMAIL = "[email protected]"
PASSWORD = "password"


class TestImportCRMData(StaticLiveServerTestCase):
    @classmethod
    def setUpClass(cls):
        super().setUpClass()
        cls.webdriver = webdriver.Chrome()
        cls.webdriver.implicitly_wait(10)

    @classmethod
    def tearDownClass(cls):
        cls.webdriver.close()
        cls.webdriver.quit()
        super().tearDownClass()

    def setUp(self):
        self.admin = User.objects.create_superuser(email=BACH_EMAIL, password=PASSWORD)

    def test_admin_tool(self):
        self.webdriver.get(f"http://{settings.ADMIN_HOST}:{self.server_thread.port}/admin")

        self.webdriver.find_element_by_id("id_username").send_keys(BACH_EMAIL)
        self.webdriver.find_element_by_id("id_password").send_keys(PASSWORD)
        self.webdriver.find_element_by_id("id_password").send_keys(Keys.RETURN)
        self.webdriver.find_element_by_link_text("Users").click()

When I run it, the test pass but still ends with this error:

Traceback (most recent call last):
  File "C:\Users\pupeno\Documents\Eligible\code\example\venv\lib\site-packages\django\db\backends\utils.py", line 83, in _execute
    return self.cursor.execute(sql)
psycopg2.OperationalError: database "test_example" is being accessed by other users
DETAIL:  There is 1 other session using the database.


The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "C:\Program Files\JetBrains\PyCharm 2018.2.4\helpers\pycharm\django_test_manage.py", line 168, in <module>
    utility.execute()
  File "C:\Program Files\JetBrains\PyCharm 2018.2.4\helpers\pycharm\django_test_manage.py", line 142, in execute
    _create_command().run_from_argv(self.argv)
  File "C:\Users\pupeno\Documents\Eligible\code\example\venv\lib\site-packages\django\core\management\commands\test.py", line 26, in run_from_argv
    super().run_from_argv(argv)
  File "C:\Users\pupeno\Documents\Eligible\code\example\venv\lib\site-packages\django\core\management\base.py", line 316, in run_from_argv
    self.execute(*args, **cmd_options)
  File "C:\Users\pupeno\Documents\Eligible\code\example\venv\lib\site-packages\django\core\management\base.py", line 353, in execute
    output = self.handle(*args, **options)
  File "C:\Program Files\JetBrains\PyCharm 2018.2.4\helpers\pycharm\django_test_manage.py", line 104, in handle
    failures = TestRunner(test_labels, **options)
  File "C:\Program Files\JetBrains\PyCharm 2018.2.4\helpers\pycharm\django_test_runner.py", line 255, in run_tests
    extra_tests=extra_tests, **options)
  File "C:\Program Files\JetBrains\PyCharm 2018.2.4\helpers\pycharm\django_test_runner.py", line 156, in run_tests
    return super(DjangoTeamcityTestRunner, self).run_tests(test_labels, extra_tests, **kwargs)
  File "C:\Users\pupeno\Documents\Eligible\code\example\venv\lib\site-packages\django\test\runner.py", line 607, in run_tests
    self.teardown_databases(old_config)
  File "C:\Users\pupeno\Documents\Eligible\code\example\venv\lib\site-packages\django\test\runner.py", line 580, in teardown_databases
    keepdb=self.keepdb,
  File "C:\Users\pupeno\Documents\Eligible\code\example\venv\lib\site-packages\django\test\utils.py", line 297, in teardown_databases
    connection.creation.destroy_test_db(old_name, verbosity, keepdb)
  File "C:\Users\pupeno\Documents\Eligible\code\example\venv\lib\site-packages\django\db\backends\base\creation.py", line 257, in destroy_test_db
    self._destroy_test_db(test_database_name, verbosity)
  File "C:\Users\pupeno\Documents\Eligible\code\example\venv\lib\site-packages\django\db\backends\base\creation.py", line 274, in _destroy_test_db
    % self.connection.ops.quote_name(test_database_name))
  File "C:\Users\pupeno\Documents\Eligible\code\example\venv\lib\site-packages\django\db\backends\utils.py", line 68, in execute
    return self._execute_with_wrappers(sql, params, many=False, executor=self._execute)
  File "C:\Users\pupeno\Documents\Eligible\code\example\venv\lib\site-packages\django\db\backends\utils.py", line 77, in _execute_with_wrappers
    return executor(sql, params, many, context)
  File "C:\Users\pupeno\Documents\Eligible\code\example\venv\lib\site-packages\django\db\backends\utils.py", line 85, in _execute
    return self.cursor.execute(sql, params)
  File "C:\Users\pupeno\Documents\Eligible\code\example\venv\lib\site-packages\django\db\utils.py", line 89, in __exit__
    raise dj_exc_value.with_traceback(traceback) from exc_value
  File "C:\Users\pupeno\Documents\Eligible\code\example\venv\lib\site-packages\django\db\backends\utils.py", line 83, in _execute
    return self.cursor.execute(sql)
django.db.utils.OperationalError: database "test_example" is being accessed by other users
DETAIL:  There is 1 other session using the database.

The problem of course is that the next run of the tests, the database still exists, so, the tests don't run without confirming deletion of the database.

If I comment out the last line:

self.webdriver.find_element_by_link_text("Users").click()

then I don't get this error. I guess just because the database connection is not established. Sometimes it's 1 other session, sometimes it's up to 4. In one of the cases of 4 sessions, these were the running sessions:

select * from pg_stat_activity where datname = 'test_example';

100123  test_example    29892   16393   pupeno  ""  ::1     61967   2018-11-15 17:28:19.552431      2018-11-15 17:28:19.562398  2018-11-15 17:28:19.564623          idle            SELECT "core_user"."id", "core_user"."password", "core_user"."last_login", "core_user"."is_superuser", "core_user"."email", "core_user"."is_staff", "core_user"."is_active", "core_user"."date_joined" FROM "core_user" WHERE "core_user"."id" = 1
100123  test_example    33028   16393   pupeno  ""  ::1     61930   2018-11-15 17:28:18.792466      2018-11-15 17:28:18.843383  2018-11-15 17:28:18.851828          idle            SELECT "django_admin_log"."id", "django_admin_log"."action_time", "django_admin_log"."user_id", "django_admin_log"."content_type_id", "django_admin_log"."object_id", "django_admin_log"."object_repr", "django_admin_log"."action_flag", "django_admin_log"."change_message", "core_user"."id", "core_user"."password", "core_user"."last_login", "core_user"."is_superuser", "core_user"."email", "core_user"."is_staff", "core_user"."is_active", "core_user"."date_joined", "django_content_type"."id", "django_content_type"."app_label", "django_content_type"."model" FROM "django_admin_log" INNER JOIN "core_user" ON ("django_admin_log"."user_id" = "core_user"."id") LEFT OUTER JOIN "django_content_type" ON ("django_admin_log"."content_type_id" = "django_content_type"."id") WHERE "django_admin_log"."user_id" = 1 ORDER BY "django_admin_log"."action_time" DESC  LIMIT 10
100123  test_example    14128   16393   pupeno  ""  ::1     61988   2018-11-15 17:28:19.767225      2018-11-15 17:28:19.776150  2018-11-15 17:28:19.776479          idle            SELECT "core_firm"."id", "core_firm"."name", "core_firm"."host_name" FROM "core_firm" WHERE "core_firm"."id" = 1
100123  test_example    9604    16393   pupeno  ""  ::1     61960   2018-11-15 17:28:19.469197      2018-11-15 17:28:19.478775  2018-11-15 17:28:19.478788          idle            COMMIT

I've been trying to find the minimum reproducible example of this problem, but so far I haven't succeeded.

Any ideas what could be causing this or how to find out more about what the issue could be?

Miscall answered 15/11, 2018 at 16:23 Comment(12)
So you have no direct access to the database and no other tests being run that interact with the database, and yet you're getting this error?Trainband
@natn2323: correct. I'm running this and only this test class. There's nothing else connecting or it would cause the problem regardless of me commenting out the last line.Miscall
Maybe you are limited? check SELECT rolname, rolconnlimit FROM pg_roles WHERE rolconnlimit <> -1; if it doesn't show your your user then it's not limited... let me know firstOrthoepy
@ShlomiBazel: it doesn't. The user I'm using is a superuser.Miscall
Is the issue permanent or intermittent?Colure
@TarunLalwani: intermittent. It also seems to be happening on Mac.Miscall
@pupeno, I would also add from django.db import connections and connections.close_all() in the tear down tearDownClass and see if it helpsColure
@pupeno You use f"http://{settings.ADMIN_HOST}:{self.server_thread.port}/admin" to connect to your server. Are you sure that this value is the same as you'd get with f"{self.live_server_url}/admin"? If not, then you have something at a different address from the live server that the testing framework starts which is answering queries.Rotifer
@pupeno, It could be that there's still a pending request in the page when the test is disposed of. Check for any error in the browser console and try to add a sleep at the end to see if it's the case. You should also try to move the webdriver instantiation from setUpClass to setUp. In any case you should handle the state of the database in the setup if the test rely on it.Pentateuch
Have you tried wrapping your cls.webdriver.close() and cls.webdriver.quit() lines in a try: ... except: ...? I've had this happen with normal tests if I've done something like sent SIGKILL to the test process before it finishes, and I'm presuming you may have something similar happening if an error occurs killing the chrome webdriver.Dagenham
Does this issue show up even when running the test with an "in memory" database ? For example: DATABASES = {'default': {'ENGINE': 'django.db.backends.sqlite3', 'NAME': ':memory:', 'USER': '', 'PASSWORD': '', 'HOST': '', 'PORT': '', 'ADMINUSER': 'admin', } }Subtropics
What version of Django are you using? 2.0 or greater? And what database, SQLite? It's likely you're hitting this Django bug: code.djangoproject.com/ticket/29062Barrens
C
3

There has been an issue reported long back for the same (bug #22424).

One thing you need to make sure is that CONN_MAX_AGE is set to 0 and not None

Also, you can use something like below in your teardown

from django.db import connections
from django.db.utils import OperationalError

@classmethod
def tearDownClass(cls):
    # Workaround for https://code.djangoproject.com/ticket/22414
    # Persistent connections not closed by LiveServerTestCase, preventing dropping test databases
    # https://github.com/cjerdonek/django/commit/b07fbca02688a0f8eb159f0dde132e7498aa40cc
    def close_sessions(conn):
        close_sessions_query = """
            SELECT pg_terminate_backend(pg_stat_activity.pid) FROM pg_stat_activity WHERE
                datname = current_database() AND
                pid <> pg_backend_pid();
        """
        with conn.cursor() as cursor:
            try:
                cursor.execute(close_sessions_query)
            except OperationalError:
                # We get kicked out after closing.
                pass

    for alias in connections:
        connections[alias].close()
        close_sessions(connections[alias])

    print "Forcefully closed database connections."

Above code is from below URL

https://github.com/cga-harvard/Hypermap-Registry/blob/cd4efad61f18194ddab2c662aa431aa21dec03f4/hypermap/tests/test_csw.py

Colure answered 12/7, 2019 at 5:7 Comment(1)
Even though that bug report is closed as fixed, I still found myself needing this workaround in this answer post. Thank you!Riker
E
3

This error message...

django.db.utils.OperationalError: database "test_example" is being accessed by other users
DETAIL:  There is 1 other session using the database.

...implies that there is an existing session using the database and the new session can't access the database.


Some more information regarding the Django version, Database type and version along with Selenium, ChromeDriver and Chrome versions would have helped us to debug this issue in a better way.

However, you need to take care of a couple of things from Selenium perspective as follows:

  • As you are initiating a new session on the next run of the tests you need to remove the line of code cls.webdriver.close() as the next line of code cls.webdriver.quit() will be enough to terminate the existing session. As per the best practices you should invoke the quit() method within the tearDown() {}. Invoking quit() DELETEs the current browsing session through sending "quit" command with {"flags":["eForceQuit"]} and finally sends the GET request on /shutdown EndPoint. Here is an example below :

    1503397488598   webdriver::server   DEBUG   -> DELETE /session/8e457516-3335-4d3b-9140-53fb52aa8b74 
    1503397488607   geckodriver::marionette TRACE   -> 37:[0,4,"quit",{"flags":["eForceQuit"]}]
    1503397488821   webdriver::server   DEBUG   -> GET /shutdown
    
  • So on invoking quit() method the Web Browser session and the WebDriver instance gets killed completely. Hence you don't have to incorporate any additional steps which will be an overhead.

You can find a detailed discussion in Selenium : How to stop geckodriver process impacting PC memory, without calling driver.quit()?

  • The currently build Web Applications are gradually moving towards dynamically rendered HTML DOM so to interact with the elements within the DOM Tree, ImplicitWait is no more that effective and you need to use WebDriverWait instead. At this point it is worth to mention that, mixing up Implicit Wait and Explicit Wait can cause unpredictable wait times
  • So you need to:

    • Remove the implicitly_wait(10):

      cls.webdriver.implicitly_wait(10)
      
    • Induce WebDriverWait while interacting with elements:

      WebDriverWait(driver, 20).until(EC.element_to_be_clickable((By.ID, "id_username"))).send_keys(BACH_EMAIL)
      self.webdriver.find_element_by_id("id_password").send_keys(PASSWORD)
      self.webdriver.find_element_by_id("id_password").send_keys(Keys.RETURN)
      WebDriverWait(driver, 20).until(EC.element_to_be_clickable((By.LINK_TEXT, "Users"))).click()
      

Now, as per the discussion Persistent connections not closed by LiveServerTestCase, preventing dropping test databases this issue was observed, reported, discussed within Djangov1.6 and fixed. The main issue was:

Whenever a PostgreSQL connection is marked as persistent (CONN_MAX_AGE = None) and a LiveServerTestCase is executed, the connection from the server thread is never closed, leading to inability to drop the test database.

This is exactly the reason why you see:

select * from pg_stat_activity where datname = 'test_example';

100123  test_example    29892   16393   pupeno  ""  ::1     61967   2018-11-15 17:28:19.552431      2018-11-15 17:28:19.562398  2018-11-15 17:28:19.564623          idle            SELECT "core_user"."id", "core_user"."password", "core_user"."last_login", "core_user"."is_superuser", "core_user"."email", "core_user"."is_staff", "core_user"."is_active", "core_user"."date_joined" FROM "core_user" WHERE "core_user"."id" = 1
100123  test_example    33028   16393   pupeno  ""  ::1     61930   2018-11-15 17:28:18.792466      2018-11-15 17:28:18.843383  2018-11-15 17:28:18.851828          idle            SELECT "django_admin_log"."id", "django_admin_log"."action_time", "django_admin_log"."user_id", "django_admin_log"."content_type_id", "django_admin_log"."object_id", "django_admin_log"."object_repr", "django_admin_log"."action_flag", "django_admin_log"."change_message", "core_user"."id", "core_user"."password", "core_user"."last_login", "core_user"."is_superuser", "core_user"."email", "core_user"."is_staff", "core_user"."is_active", "core_user"."date_joined", "django_content_type"."id", "django_content_type"."app_label", "django_content_type"."model" FROM "django_admin_log" INNER JOIN "core_user" ON ("django_admin_log"."user_id" = "core_user"."id") LEFT OUTER JOIN "django_content_type" ON ("django_admin_log"."content_type_id" = "django_content_type"."id") WHERE "django_admin_log"."user_id" = 1 ORDER BY "django_admin_log"."action_time" DESC  LIMIT 10
100123  test_example    14128   16393   pupeno  ""  ::1     61988   2018-11-15 17:28:19.767225      2018-11-15 17:28:19.776150  2018-11-15 17:28:19.776479          idle            SELECT "core_firm"."id", "core_firm"."name", "core_firm"."host_name" FROM "core_firm" WHERE "core_firm"."id" = 1
100123  test_example    9604    16393   pupeno  ""  ::1     61960   2018-11-15 17:28:19.469197      2018-11-15 17:28:19.478775  2018-11-15 17:28:19.478788          idle            COMMIT

It was further observed that, even with CONN_MAX_AGE=None, after LiveServerTestCase.tearDownClass(), querying PostgreSQL's pg_stat_activity shows a lingering connection in state idle (which was the connection created by the previous test in your case). So it was pretty evident that the idle connections are not closed when the thread terminates and the needle of supection was at:

  • LiveServerThread(threading.Thread) which control the threads for running a live http server while the tests are running:

    class LiveServerThread(threading.Thread):
    
        def __init__(self, host, static_handler, connections_override=None):
            self.host = host
            self.port = None
            self.is_ready = threading.Event()
            self.error = None
            self.static_handler = static_handler
            self.connections_override = connections_override
            super(LiveServerThread, self).__init__()
    
        def run(self):
            """
            Sets up the live server and databases, and then loops over handling
            http requests.
            """
            if self.connections_override:
                # Override this thread's database connections with the ones
                # provided by the main thread.
                for alias, conn in self.connections_override.items():
                    connections[alias] = conn
            try:
                # Create the handler for serving static and media files
                handler = self.static_handler(_MediaFilesHandler(WSGIHandler()))
                self.httpd = self._create_server(0)
                self.port = self.httpd.server_address[1]
                self.httpd.set_app(handler)
                self.is_ready.set()
                self.httpd.serve_forever()
            except Exception as e:
                self.error = e
                self.is_ready.set()
    
        def _create_server(self, port):
            return WSGIServer((self.host, port), QuietWSGIRequestHandler, allow_reuse_address=False)
    
        def terminate(self):
            if hasattr(self, 'httpd'):
                # Stop the WSGI server
                self.httpd.shutdown()
                self.httpd.server_close()
    
  • LiveServerTestCase(TransactionTestCase) which basically does the same as TransactionTestCase but also launches a live http server in a separate thread so that the tests may use another testing framework to be used by Selenium, instead of the built-in dummy client:

    class LiveServerTestCase(TransactionTestCase):
    
        host = 'localhost'
        static_handler = _StaticFilesHandler
    
        @classproperty
        def live_server_url(cls):
            return 'http://%s:%s' % (cls.host, cls.server_thread.port)
    
        @classmethod
        def setUpClass(cls):
            super(LiveServerTestCase, cls).setUpClass()
            connections_override = {}
            for conn in connections.all():
                # If using in-memory sqlite databases, pass the connections to
                # the server thread.
                if conn.vendor == 'sqlite' and conn.is_in_memory_db(conn.settings_dict['NAME']):
                    # Explicitly enable thread-shareability for this connection
                    conn.allow_thread_sharing = True
                    connections_override[conn.alias] = conn
    
        cls._live_server_modified_settings = modify_settings(
            ALLOWED_HOSTS={'append': cls.host},
        )
        cls._live_server_modified_settings.enable()
        cls.server_thread = cls._create_server_thread(connections_override)
        cls.server_thread.daemon = True
        cls.server_thread.start()
    
        # Wait for the live server to be ready
        cls.server_thread.is_ready.wait()
        if cls.server_thread.error:
            # Clean up behind ourselves, since tearDownClass won't get called in
            # case of errors.
            cls._tearDownClassInternal()
            raise cls.server_thread.error
    
        @classmethod
        def _create_server_thread(cls, connections_override):
            return LiveServerThread(
                cls.host,
                cls.static_handler,
                connections_override=connections_override,
        )
    
        @classmethod
        def _tearDownClassInternal(cls):
            # There may not be a 'server_thread' attribute if setUpClass() for some
            # reasons has raised an exception.
            if hasattr(cls, 'server_thread'):
                # Terminate the live server's thread
                cls.server_thread.terminate()
                cls.server_thread.join()
    
            # Restore sqlite in-memory database connections' non-shareability
            for conn in connections.all():
                if conn.vendor == 'sqlite' and conn.is_in_memory_db(conn.settings_dict['NAME']):
                    conn.allow_thread_sharing = False
    
        @classmethod
        def tearDownClass(cls):
            cls._tearDownClassInternal()
            cls._live_server_modified_settings.disable()
            super(LiveServerTestCase, cls).tearDownClass()
    

The solution was to close only the non-overriden connections and was incorporated from this pull request / commit. The changes were:

  • In django/test/testcases.py add:

    finally:
        connections.close_all() 
    
  • Add a new file tests/servers/test_liveserverthread.py:

    from django.db import DEFAULT_DB_ALIAS, connections
    from django.test import LiveServerTestCase, TestCase
    
    
    class LiveServerThreadTest(TestCase):
    
        def run_live_server_thread(self, connections_override=None):
            thread = LiveServerTestCase._create_server_thread(connections_override)
            thread.daemon = True
            thread.start()
            thread.is_ready.wait()
            thread.terminate()
    
        def test_closes_connections(self):
            conn = connections[DEFAULT_DB_ALIAS]
            if conn.vendor == 'sqlite' and conn.is_in_memory_db():
                self.skipTest("the sqlite backend's close() method is a no-op when using an in-memory database")
            # Pass a connection to the thread to check they are being closed.
            connections_override = {DEFAULT_DB_ALIAS: conn}
    
            saved_sharing = conn.allow_thread_sharing
        try:
            conn.allow_thread_sharing = True
            self.assertTrue(conn.is_usable())
            self.run_live_server_thread(connections_override)
            self.assertFalse(conn.is_usable())
        finally:
            conn.allow_thread_sharing = saved_sharing
    
  • In tests/servers/tests.py remove:

    finally:
        TestCase.tearDownClass()
    
  • In tests/servers/tests.py add:

    finally:
        if hasattr(TestCase, 'server_thread'):
            TestCase.server_thread.terminate()
    

Solution

Steps:

  • Ensure you have updated to the latest released version of Django package.
  • Ensure you are using the latest released version of selenium v3.141.0.
  • Ensure you are using the latest released version of Chrome v76 and ChromeDriver 76.0.

Outro

You can find a similar discussion in django.db.utils.IntegrityError: FOREIGN KEY constraint failed while executing LiveServerTestCases through Selenium and Python Django

Eczema answered 18/7, 2019 at 8:28 Comment(0)
T
0

Check active connections to know what causes the problem select * from pg_stat_activity;

You can disable extensions:

    @classmethod
    def setUpClass(cls):
        super().setUpClass()
        options = webdriver.chrome.options.Options()
        options.add_argument("--disable-extensions") 
        cls.webdriver = webdriver.Chrome(chrome_options=options)
        cls.webdriver.implicitly_wait(10)

Then in teardown:

    @classmethod
    def tearDownClass(cls):
        cls.webdriver.stop_client()
        cls.webdriver.quit()
        connections.close_all()
        super().tearDownClass()
Talithatalk answered 11/7, 2019 at 18:15 Comment(0)

© 2022 - 2024 — McMap. All rights reserved.