fixes #340 session cleanup for celery tasks

rhodecode/lib/celerylib/__init__.py

@@ -1,105 +1,127 b''
 # -*- coding: utf-8 -*-
 """
     rhodecode.lib.celerylib.__init__
     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
     celery libs for RhodeCode
 
     :created_on: Nov 27, 2010
     :author: marcink
     :copyright: (C) 2010-2012 Marcin Kuzminski <[e-mail]>
     :license: GPLv3, see COPYING for more details.
 """
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by
 # the Free Software Foundation, either version 3 of the License, or
 # (at your option) any later version.
 #
 # This program 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 General Public License for more details.
 #
 # You should have received a copy of the GNU General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 
 import os
 import sys
 import socket
 import traceback
 import logging
 from os.path import dirname as dn, join as jn
+from pylons import config
 
 from hashlib import md5
 from decorator import decorator
 
 from vcs.utils.lazy import LazyProperty
 from rhodecode import CELERY_ON
 from rhodecode.lib import str2bool, safe_str
 from rhodecode.lib.pidlock import DaemonLock, LockHeld
+from rhodecode.model import init_model
+from rhodecode.model import meta
+from rhodecode.model.db import Statistics, Repository, User
+
+from sqlalchemy import engine_from_config
 
 from celery.messaging import establish_connection
 
-
 log = logging.getLogger(__name__)
 
 
-
-
 class ResultWrapper(object):
     def __init__(self, task):
         self.task = task
 
     @LazyProperty
     def result(self):
         return self.task
 
 
 def run_task(task, *args, **kwargs):
     if CELERY_ON:
         try:
             t = task.apply_async(args=args, kwargs=kwargs)
             log.info('running task %s:%s', t.task_id, task)
             return t
 
         except socket.error, e:
             if isinstance(e, IOError) and e.errno == 111:
                 log.debug('Unable to connect to celeryd. Sync execution')
             else:
                 log.error(traceback.format_exc())
         except KeyError, e:
                 log.debug('Unable to connect to celeryd. Sync execution')
         except Exception, e:
             log.error(traceback.format_exc())
 
     log.debug('executing task %s in sync mode', task)
     return ResultWrapper(task(*args, **kwargs))
 
 
 def __get_lockkey(func, *fargs, **fkwargs):
     params = list(fargs)
     params.extend(['%s-%s' % ar for ar in fkwargs.items()])
 
     func_name = str(func.__name__) if hasattr(func, '__name__') else str(func)
 
     lockkey = 'task_%s.lock' % \
         md5(func_name + '-' + '-'.join(map(safe_str, params))).hexdigest()
     return lockkey
 
 
 def locked_task(func):
     def __wrapper(func, *fargs, **fkwargs):
         lockkey = __get_lockkey(func, *fargs, **fkwargs)
         lockkey_path = config['here']
 
         log.info('running task with lockkey %s', lockkey)
         try:
             l = DaemonLock(file_=jn(lockkey_path, lockkey))
             ret = func(*fargs, **fkwargs)
             l.release()
             return ret
         except LockHeld:
             log.info('LockHeld')
             return 'Task with key %s already running' % lockkey
 
     return decorator(__wrapper, func)
+
+
+def get_session():
+    if CELERY_ON:
+        engine = engine_from_config(config, 'sqlalchemy.db1.')
+        init_model(engine)
+    sa = meta.Session
+    return sa
+
+
+def dbsession(func):
+    def __wrapper(func, *fargs, **fkwargs):
+        try:
+            ret = func(*fargs, **fkwargs)
+            return ret
+        finally:
+            meta.Session.remove()
+
+    return decorator(__wrapper, func)

rhodecode/lib/celerylib/tasks.py

@@ -1,418 +1,414 b''
 # -*- coding: utf-8 -*-
 """
     rhodecode.lib.celerylib.tasks
     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
     RhodeCode task modules, containing all task that suppose to be run
     by celery daemon
 
     :created_on: Oct 6, 2010
     :author: marcink
     :copyright: (C) 2010-2012 Marcin Kuzminski <[e-mail]>
     :license: GPLv3, see COPYING for more details.
 """
 # This program is free software: you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by
 # the Free Software Foundation, either version 3 of the License, or
 # (at your option) any later version.
 #
 # This program 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 General Public License for more details.
 #
 # You should have received a copy of the GNU General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 from celery.decorators import task
 
 import os
 import traceback
 import logging
 from os.path import join as jn
 
 from time import mktime
 from operator import itemgetter
 from string import lower
 
 from pylons import config, url
 from pylons.i18n.translation import _
 
 from vcs import get_backend
 
 from rhodecode import CELERY_ON
 from rhodecode.lib import LANGUAGES_EXTENSIONS_MAP, safe_str
-from rhodecode.lib.celerylib import run_task, locked_task, str2bool, \
-    __get_lockkey, LockHeld, DaemonLock
+from rhodecode.lib.celerylib import run_task, locked_task, dbsession, \
+    str2bool, __get_lockkey, LockHeld, DaemonLock, get_session
 from rhodecode.lib.helpers import person
 from rhodecode.lib.rcmail.smtp_mailer import SmtpMailer
 from rhodecode.lib.utils import add_cache, action_logger
 from rhodecode.lib.compat import json, OrderedDict
 
-from rhodecode.model import init_model
-from rhodecode.model import meta
 from rhodecode.model.db import Statistics, Repository, User
 
-from sqlalchemy import engine_from_config
 
 add_cache(config)
 
 __all__ = ['whoosh_index', 'get_commits_stats',
            'reset_user_password', 'send_email']
 
 
-def get_session():
-    if CELERY_ON:
-        engine = engine_from_config(config, 'sqlalchemy.db1.')
-        init_model(engine)
-    sa = meta.Session
-    return sa
-
 def get_logger(cls):
     if CELERY_ON:
         try:
             log = cls.get_logger()
         except:
             log = logging.getLogger(__name__)
     else:
         log = logging.getLogger(__name__)
 
     return log
 
 
 @task(ignore_result=True)
 @locked_task
+@dbsession
 def whoosh_index(repo_location, full_index):
     from rhodecode.lib.indexers.daemon import WhooshIndexingDaemon
-
-    # log = whoosh_index.get_logger(whoosh_index)
+    log = whoosh_index.get_logger(whoosh_index)
+    DBS = get_session()
 
     index_location = config['index_dir']
     WhooshIndexingDaemon(index_location=index_location,
-                         repo_location=repo_location, sa=get_session())\
+                         repo_location=repo_location, sa=DBS)\
                          .run(full_index=full_index)
 
 
 @task(ignore_result=True)
+@dbsession
 def get_commits_stats(repo_name, ts_min_y, ts_max_y):
     log = get_logger(get_commits_stats)
-
+    DBS = get_session()
     lockkey = __get_lockkey('get_commits_stats', repo_name, ts_min_y,
                             ts_max_y)
     lockkey_path = config['here']
 
     log.info('running task with lockkey %s', lockkey)
 
     try:
-        sa = get_session()
         lock = l = DaemonLock(file_=jn(lockkey_path, lockkey))
 
         # for js data compatibilty cleans the key for person from '
         akc = lambda k: person(k).replace('"', "")
 
         co_day_auth_aggr = {}
         commits_by_day_aggregate = {}
         repo = Repository.get_by_repo_name(repo_name)
         if repo is None:
             return True
 
         repo = repo.scm_instance
         repo_size = repo.count()
         # return if repo have no revisions
         if repo_size < 1:
             lock.release()
             return True
 
         skip_date_limit = True
         parse_limit = int(config['app_conf'].get('commit_parse_limit'))
         last_rev = None
         last_cs = None
         timegetter = itemgetter('time')
 
-        dbrepo = sa.query(Repository)\
+        dbrepo = DBS.query(Repository)\
             .filter(Repository.repo_name == repo_name).scalar()
-        cur_stats = sa.query(Statistics)\
+        cur_stats = DBS.query(Statistics)\
             .filter(Statistics.repository == dbrepo).scalar()
 
         if cur_stats is not None:
             last_rev = cur_stats.stat_on_revision
 
         if last_rev == repo.get_changeset().revision and repo_size > 1:
             # pass silently without any work if we're not on first revision or
             # current state of parsing revision(from db marker) is the
             # last revision
             lock.release()
             return True
 
         if cur_stats:
             commits_by_day_aggregate = OrderedDict(json.loads(
                                         cur_stats.commit_activity_combined))
             co_day_auth_aggr = json.loads(cur_stats.commit_activity)
 
         log.debug('starting parsing %s', parse_limit)
         lmktime = mktime
 
         last_rev = last_rev + 1 if last_rev >= 0 else 0
         log.debug('Getting revisions from %s to %s' % (
              last_rev, last_rev + parse_limit)
         )
         for cs in repo[last_rev:last_rev + parse_limit]:
             last_cs = cs  # remember last parsed changeset
             k = lmktime([cs.date.timetuple()[0], cs.date.timetuple()[1],
                           cs.date.timetuple()[2], 0, 0, 0, 0, 0, 0])
 
             if akc(cs.author) in co_day_auth_aggr:
                 try:
                     l = [timegetter(x) for x in
                          co_day_auth_aggr[akc(cs.author)]['data']]
                     time_pos = l.index(k)
                 except ValueError:
                     time_pos = False
 
                 if time_pos >= 0 and time_pos is not False:
 
                     datadict = \
                         co_day_auth_aggr[akc(cs.author)]['data'][time_pos]
 
                     datadict["commits"] += 1
                     datadict["added"] += len(cs.added)
                     datadict["changed"] += len(cs.changed)
                     datadict["removed"] += len(cs.removed)
 
                 else:
                     if k >= ts_min_y and k <= ts_max_y or skip_date_limit:
 
                         datadict = {"time": k,
                                     "commits": 1,
                                     "added": len(cs.added),
                                     "changed": len(cs.changed),
                                     "removed": len(cs.removed),
                                    }
                         co_day_auth_aggr[akc(cs.author)]['data']\
                             .append(datadict)
 
             else:
                 if k >= ts_min_y and k <= ts_max_y or skip_date_limit:
                     co_day_auth_aggr[akc(cs.author)] = {
                                         "label": akc(cs.author),
                                         "data": [{"time":k,
                                                  "commits":1,
                                                  "added":len(cs.added),
                                                  "changed":len(cs.changed),
                                                  "removed":len(cs.removed),
                                                  }],
                                         "schema": ["commits"],
                                         }
 
             #gather all data by day
             if k in commits_by_day_aggregate:
                 commits_by_day_aggregate[k] += 1
             else:
                 commits_by_day_aggregate[k] = 1
 
         overview_data = sorted(commits_by_day_aggregate.items(),
                                key=itemgetter(0))
 
         if not co_day_auth_aggr:
             co_day_auth_aggr[akc(repo.contact)] = {
                 "label": akc(repo.contact),
                 "data": [0, 1],
                 "schema": ["commits"],
             }
 
         stats = cur_stats if cur_stats else Statistics()
         stats.commit_activity = json.dumps(co_day_auth_aggr)
         stats.commit_activity_combined = json.dumps(overview_data)
 
         log.debug('last revison %s', last_rev)
         leftovers = len(repo.revisions[last_rev:])
         log.debug('revisions to parse %s', leftovers)
 
         if last_rev == 0 or leftovers < parse_limit:
             log.debug('getting code trending stats')
             stats.languages = json.dumps(__get_codes_stats(repo_name))
 
         try:
             stats.repository = dbrepo
             stats.stat_on_revision = last_cs.revision if last_cs else 0
-            sa.add(stats)
-            sa.commit()
+            DBS.add(stats)
+            DBS.commit()
         except:
             log.error(traceback.format_exc())
-            sa.rollback()
+            DBS.rollback()
             lock.release()
             return False
 
         #final release
         lock.release()
 
         #execute another task if celery is enabled
         if len(repo.revisions) > 1 and CELERY_ON:
             run_task(get_commits_stats, repo_name, ts_min_y, ts_max_y)
         return True
     except LockHeld:
         log.info('LockHeld')
         return 'Task with key %s already running' % lockkey
 
 @task(ignore_result=True)
+@dbsession
 def send_password_link(user_email):
     from rhodecode.model.notification import EmailNotificationModel
 
     log = get_logger(send_password_link)
+    DBS = get_session()
 
     try:
-        sa = get_session()
         user = User.get_by_email(user_email)
         if user:
             log.debug('password reset user found %s' % user)
             link = url('reset_password_confirmation', key=user.api_key,
                        qualified=True)
             reg_type = EmailNotificationModel.TYPE_PASSWORD_RESET
             body = EmailNotificationModel().get_email_tmpl(reg_type,
                                                 **{'user':user.short_contact,
                                                    'reset_url':link})
             log.debug('sending email')
             run_task(send_email, user_email,
                      _("password reset link"), body)
             log.info('send new password mail to %s', user_email)
         else:
             log.debug("password reset email %s not found" % user_email)
     except:
         log.error(traceback.format_exc())
         return False
 
     return True
 
 @task(ignore_result=True)
+@dbsession
 def reset_user_password(user_email):
     from rhodecode.lib import auth
 
     log = get_logger(reset_user_password)
+    DBS = get_session()
 
     try:
         try:
-            sa = get_session()
             user = User.get_by_email(user_email)
             new_passwd = auth.PasswordGenerator().gen_password(8,
                              auth.PasswordGenerator.ALPHABETS_BIG_SMALL)
             if user:
                 user.password = auth.get_crypt_password(new_passwd)
                 user.api_key = auth.generate_api_key(user.username)
-                sa.add(user)
-                sa.commit()
+                DBS.add(user)
+                DBS.commit()
                 log.info('change password for %s', user_email)
             if new_passwd is None:
                 raise Exception('unable to generate new password')
         except:
             log.error(traceback.format_exc())
-            sa.rollback()
+            DBS.rollback()
 
         run_task(send_email, user_email,
                  'Your new password',
                  'Your new RhodeCode password:%s' % (new_passwd))
         log.info('send new password mail to %s', user_email)
 
     except:
         log.error('Failed to update user password')
         log.error(traceback.format_exc())
 
     return True
 
 
 @task(ignore_result=True)
+@dbsession
 def send_email(recipients, subject, body, html_body=''):
     """
     Sends an email with defined parameters from the .ini files.
 
     :param recipients: list of recipients, it this is empty the defined email
         address from field 'email_to' is used instead
     :param subject: subject of the mail
     :param body: body of the mail
     :param html_body: html version of body
     """
     log = get_logger(send_email)
-    sa = get_session()
+    DBS = get_session()
+    
     email_config = config
     subject = "%s %s" % (email_config.get('email_prefix'), subject)
     if not recipients:
         # if recipients are not defined we send to email_config + all admins
         admins = [u.email for u in User.query()
                   .filter(User.admin == True).all()]
         recipients = [email_config.get('email_to')] + admins
 
     mail_from = email_config.get('app_email_from', 'RhodeCode')
     user = email_config.get('smtp_username')
     passwd = email_config.get('smtp_password')
     mail_server = email_config.get('smtp_server')
     mail_port = email_config.get('smtp_port')
     tls = str2bool(email_config.get('smtp_use_tls'))
     ssl = str2bool(email_config.get('smtp_use_ssl'))
     debug = str2bool(config.get('debug'))
     smtp_auth = email_config.get('smtp_auth')
 
     try:
         m = SmtpMailer(mail_from, user, passwd, mail_server, smtp_auth,
                        mail_port, ssl, tls, debug=debug)
         m.send(recipients, subject, body, html_body)
     except:
         log.error('Mail sending failed')
         log.error(traceback.format_exc())
         return False
     return True
 
 
 @task(ignore_result=True)
+@dbsession
 def create_repo_fork(form_data, cur_user):
     """
     Creates a fork of repository using interval VCS methods
 
     :param form_data:
     :param cur_user:
     """
     from rhodecode.model.repo import RepoModel
 
     log = get_logger(create_repo_fork)
+    DBS = create_repo_fork.DBS
 
-    Session = get_session()
     base_path = Repository.base_path()
 
-    RepoModel(Session).create(form_data, cur_user, just_db=True, fork=True)
+    RepoModel(DBS).create(form_data, cur_user, just_db=True, fork=True)
 
     alias = form_data['repo_type']
     org_repo_name = form_data['org_path']
     fork_name = form_data['repo_name_full']
     update_after_clone = form_data['update_after_clone']
     source_repo_path = os.path.join(base_path, org_repo_name)
     destination_fork_path = os.path.join(base_path, fork_name)
 
     log.info('creating fork of %s as %s', source_repo_path,
              destination_fork_path)
     backend = get_backend(alias)
     backend(safe_str(destination_fork_path), create=True,
             src_url=safe_str(source_repo_path),
             update_after_clone=update_after_clone)
     action_logger(cur_user, 'user_forked_repo:%s' % fork_name,
-                   org_repo_name, '', Session)
+                   org_repo_name, '', DBS)
 
     action_logger(cur_user, 'user_created_fork:%s' % fork_name,
-                   fork_name, '', Session)
+                   fork_name, '', DBS)
     # finally commit at latest possible stage
-    Session.commit()
+    DBS.commit()
 
 def __get_codes_stats(repo_name):
     repo = Repository.get_by_repo_name(repo_name).scm_instance
 
     tip = repo.get_changeset()
     code_stats = {}
 
     def aggregate(cs):
         for f in cs[2]:
             ext = lower(f.extension)
             if ext in LANGUAGES_EXTENSIONS_MAP.keys() and not f.is_binary:
                 if ext in code_stats:
                     code_stats[ext] += 1
                 else:
                     code_stats[ext] = 1
 
     map(aggregate, tip.walk('/'))
 
     return code_stats or {}