[server/session] add missing rollback when freeing a db connection
When we let go of a cnxset and give it back to the pool, we need to make
sure it's not in the middle of a transaction.
We got bitten by a migration where the connection used by admin's
repo.connect was not rolled back before attempting the migration,
causing a deadlock because it had a lock on a table. This could
potentially also be an issue for e.g. hooks or looping tasks that didn't
explicitly rollback before calling free_cnxset or moving out of a
ensure_cnx_set block.
# copyright 2003-2013 LOGILAB S.A. (Paris, FRANCE), all rights reserved.
# contact http://www.logilab.fr/ -- mailto:contact@logilab.fr
#
# This file is part of CubicWeb.
#
# CubicWeb is free software: you can redistribute it and/or modify it under the
# terms of the GNU Lesser General Public License as published by the Free
# Software Foundation, either version 2.1 of the License, or (at your option)
# any later version.
#
# CubicWeb is distributed in the hope that it will be useful, but WITHOUT
# ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
# FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more
# details.
#
# You should have received a copy of the GNU Lesser General Public License along
# with CubicWeb. If not, see <http://www.gnu.org/licenses/>.
"""dummy wsgi server for CubicWeb web instances"""
__docformat__ = "restructuredtext en"
import socket
from cubicweb.wsgi.handler import CubicWebWSGIApplication
from cubicweb import ConfigurationError
from werkzeug.serving import run_simple
from logging import getLogger
LOGGER = getLogger('cubicweb')
def run(config):
config.check_writeable_uid_directory(config.appdatahome)
port = config['port'] or 8080
interface = config['interface']
app = CubicWebWSGIApplication(config)
repo = app.appli.repo
try:
repo.start_looping_tasks()
run_simple(interface, port, app,
threaded=True,
use_debugger=True,
processes=1) # more processes yield weird errors
finally:
repo.shutdown()