InterfaceError: connection already closed (using django + celery + Scrapy)

Question

I am getting this when using a Scrapy parsing function (that can take till 10 minutes sometimes) inside a Celery task.

I use: - Django==1.6.5 - django-celery==3.1.16 - celery==3.1.16 - psycopg2==2.5.5 (I used also psycopg2==2.5.4)

[2015-07-19 11:27:49,488: CRITICAL/MainProcess] Task myapp.parse_items[63fc40eb-c0d6-46f4-a64e-acce8301d29a] INTERNAL ERROR: InterfaceError('connection already closed',)
Traceback (most recent call last):
  File "/home/mo/Work/python/pb-env/local/lib/python2.7/site-packages/celery/app/trace.py", line 284, in trace_task
    uuid, retval, SUCCESS, request=task_request,
  File "/home/mo/Work/python/pb-env/local/lib/python2.7/site-packages/celery/backends/base.py", line 248, in store_result
    request=request, **kwargs)
  File "/home/mo/Work/python/pb-env/local/lib/python2.7/site-packages/djcelery/backends/database.py", line 29, in _store_result
    traceback=traceback, children=self.current_task_children(request),
  File "/home/mo/Work/python/pb-env/local/lib/python2.7/site-packages/djcelery/managers.py", line 42, in _inner
    return fun(*args, **kwargs)
  File "/home/mo/Work/python/pb-env/local/lib/python2.7/site-packages/djcelery/managers.py", line 181, in store_result
    'meta': {'children': children}})
  File "/home/mo/Work/python/pb-env/local/lib/python2.7/site-packages/djcelery/managers.py", line 87, in update_or_create
    return get_queryset(self).update_or_create(**kwargs)
  File "/home/mo/Work/python/pb-env/local/lib/python2.7/site-packages/djcelery/managers.py", line 70, in update_or_create
    obj, created = self.get_or_create(**kwargs)
  File "/home/mo/Work/python/pb-env/local/lib/python2.7/site-packages/django/db/models/query.py", line 376, in get_or_create
    return self.get(**lookup), False
  File "/home/mo/Work/python/pb-env/local/lib/python2.7/site-packages/django/db/models/query.py", line 304, in get
    num = len(clone)
  File "/home/mo/Work/python/pb-env/local/lib/python2.7/site-packages/django/db/models/query.py", line 77, in __len__
    self._fetch_all()
  File "/home/mo/Work/python/pb-env/local/lib/python2.7/site-packages/django/db/models/query.py", line 857, in _fetch_all
    self._result_cache = list(self.iterator())
  File "/home/mo/Work/python/pb-env/local/lib/python2.7/site-packages/django/db/models/query.py", line 220, in iterator
    for row in compiler.results_iter():
  File "/home/mo/Work/python/pb-env/local/lib/python2.7/site-packages/django/db/models/sql/compiler.py", line 713, in results_iter
    for rows in self.execute_sql(MULTI):
  File "/home/mo/Work/python/pb-env/local/lib/python2.7/site-packages/django/db/models/sql/compiler.py", line 785, in execute_sql
    cursor = self.connection.cursor()
  File "/home/mo/Work/python/pb-env/local/lib/python2.7/site-packages/django/db/backends/__init__.py", line 160, in cursor
    cursor = self.make_debug_cursor(self._cursor())
  File "/home/mo/Work/python/pb-env/local/lib/python2.7/site-packages/django/db/backends/__init__.py", line 134, in _cursor
    return self.create_cursor()
  File "/home/mo/Work/python/pb-env/local/lib/python2.7/site-packages/django/db/utils.py", line 99, in __exit__
    six.reraise(dj_exc_type, dj_exc_value, traceback)
  File "/home/mo/Work/python/pb-env/local/lib/python2.7/site-packages/django/db/backends/__init__.py", line 134, in _cursor
    return self.create_cursor()
  File "/home/mo/Work/python/pb-env/local/lib/python2.7/site-packages/django/db/backends/postgresql_psycopg2/base.py", line 137, in create_cursor
    cursor = self.connection.cursor()
InterfaceError: connection already closed

Answer 1

In my case it was caused by not closed transaction.atomic.

def foo():
    trans = transaction.atomic()
    trans.__enter__()
    ## some code
    if error:
        raise Exception
    trans.__exit__()

When an exception occured the transaction was kept open so the connection was. Even if I manually called connection.close() Connection already closed was occuring on the next call of the task.

I've fixed it with:

with transaction.atomic():
    ....

Check if you don't leave an open transactions ;)

Answer 2

If you are running into this when running tests, then you can either change the test to TransactionTestCase class instead of TestCase or add the mark pytest.mark.django_db(transaction=True). This kept my db connection alive from creation of the pytest-celery fixtures to the database calls.

Github issue - https://github.com/Koed00/django-q/issues/167

For context, I am using pytest-celery with celery_app and celery_worker as fixtures in my tests. I am also trying to hit the test db in the tasks referenced in these tests.

If someone would explain switching to transaction=True keeps it open, that would be great!

Answer 3

Guys and emanuelcds,

I had the same problem, now I have updated my code and created a new loader for celery:

from djcelery.loaders import DjangoLoader
from django import db

class CustomDjangoLoader(DjangoLoader):
    def on_task_init(self, task_id, task):
        """Called before every task."""
        for conn in db.connections.all():
            conn.close_if_unusable_or_obsolete()
        super(CustomDjangoLoader, self).on_task_init(task_id, task)

This of course if you are using djcelery, it will also require something like this in the settings:

CELERY_LOADER = 'myproject.loaders.CustomDjangoLoader'
os.environ['CELERY_LOADER'] = CELERY_LOADER

I still have to test it, I will update.

Answer 4

Unfortunately this is a problem with django + psycopg2 + celery combo. It's an old and unsolved problem.

Take a look on this thread to understand: https://github.com/celery/django-celery/issues/121

Basically, when celery starts a worker, it forks a database connection from django.db framework. If this connection drops for some reason, it doesn't create a new one. Celery has nothing to do with this problem once there is no way to detect when the database connection is dropped using django.db libraries. Django doesn't notifies when it happens, because it just start a connection and it receives a wsgi call (no connection pool). I had the same problem on a huge production environment with a lot of machine workers, and sometimes, these machines lost connectivity with postgres server.

I solved it putting each celery master process under a linux supervisord handler and a watcher and implemented a decorator that handles the psycopg2.InterfaceError, and when it happens this function dispatches a syscall to force supervisor restart gracefully with SIGINT the celery process.

Edit:

Found a better solution. I implemented a celery task baseclass like this:

from django.db import connection
import celery

class FaultTolerantTask(celery.Task):
    """ Implements after return hook to close the invalid connection.
    This way, django is forced to serve a new connection for the next
    task.
    """
    abstract = True

    def after_return(self, *args, **kwargs):
        connection.close()

@celery.task(base=FaultTolerantTask)
def my_task():
    # my database dependent code here

I believe it will fix your problem too.