Changeset - 3a007d806f0f
[Not reviewed]
beta
0 2 0
Marcin Kuzminski - 13 years ago 2012-09-05 22:09:25
marcin@python-works.com
Fixing issues of cache invalidation for multiple instances running in rhodecode.
- during server start always register all cache keys for all repositories
- Don't clear cache markers on start since multiple instances can clear each other out
2 files changed with 11 insertions and 5 deletions:
0 comments (0 inline, 0 general)
rhodecode/lib/utils.py
Show inline comments
 
@@ -320,238 +320,243 @@ def make_ui(read_from='file', path=None,
 
            if ui_.ui_active:
 
                log.debug('settings ui from db[%s]%s:%s', ui_.ui_section,
 
                          ui_.ui_key, ui_.ui_value)
 
                baseui.setconfig(ui_.ui_section, ui_.ui_key, ui_.ui_value)
 
            if ui_.ui_key == 'push_ssl':
 
                # force set push_ssl requirement to False, rhodecode
 
                # handles that
 
                baseui.setconfig(ui_.ui_section, ui_.ui_key, False)
 
        if clear_session:
 
            meta.Session.remove()
 
    return baseui
 

	
 

	
 
def set_rhodecode_config(config):
 
    """
 
    Updates pylons config with new settings from database
 

	
 
    :param config:
 
    """
 
    hgsettings = RhodeCodeSetting.get_app_settings()
 

	
 
    for k, v in hgsettings.items():
 
        config[k] = v
 

	
 

	
 
def invalidate_cache(cache_key, *args):
 
    """
 
    Puts cache invalidation task into db for
 
    further global cache invalidation
 
    """
 

	
 
    from rhodecode.model.scm import ScmModel
 

	
 
    if cache_key.startswith('get_repo_cached_'):
 
        name = cache_key.split('get_repo_cached_')[-1]
 
        ScmModel().mark_for_invalidation(name)
 

	
 

	
 
def map_groups(path):
 
    """
 
    Given a full path to a repository, create all nested groups that this
 
    repo is inside. This function creates parent-child relationships between
 
    groups and creates default perms for all new groups.
 

	
 
    :param paths: full path to repository
 
    """
 
    sa = meta.Session()
 
    groups = path.split(Repository.url_sep())
 
    parent = None
 
    group = None
 

	
 
    # last element is repo in nested groups structure
 
    groups = groups[:-1]
 
    rgm = ReposGroupModel(sa)
 
    for lvl, group_name in enumerate(groups):
 
        group_name = '/'.join(groups[:lvl] + [group_name])
 
        group = RepoGroup.get_by_group_name(group_name)
 
        desc = '%s group' % group_name
 

	
 
        # skip folders that are now removed repos
 
        if REMOVED_REPO_PAT.match(group_name):
 
            break
 

	
 
        if group is None:
 
            log.debug('creating group level: %s group_name: %s' % (lvl,
 
                                                                   group_name))
 
            group = RepoGroup(group_name, parent)
 
            group.group_description = desc
 
            sa.add(group)
 
            rgm._create_default_perms(group)
 
            sa.flush()
 
        parent = group
 
    return group
 

	
 

	
 
def repo2db_mapper(initial_repo_list, remove_obsolete=False,
 
                   install_git_hook=False):
 
    """
 
    maps all repos given in initial_repo_list, non existing repositories
 
    are created, if remove_obsolete is True it also check for db entries
 
    that are not in initial_repo_list and removes them.
 

	
 
    :param initial_repo_list: list of repositories found by scanning methods
 
    :param remove_obsolete: check for obsolete entries in database
 
    :param install_git_hook: if this is True, also check and install githook
 
        for a repo if missing
 
    """
 
    from rhodecode.model.repo import RepoModel
 
    from rhodecode.model.scm import ScmModel
 
    sa = meta.Session()
 
    rm = RepoModel()
 
    user = sa.query(User).filter(User.admin == True).first()
 
    if user is None:
 
        raise Exception('Missing administrative account !')
 
    added = []
 

	
 
#    # clear cache keys
 
#    log.debug("Clearing cache keys now...")
 
#    CacheInvalidation.clear_cache()
 
#    sa.commit()
 

	
 
    for name, repo in initial_repo_list.items():
 
        group = map_groups(name)
 
        db_repo = rm.get_by_repo_name(name)
 
        # found repo that is on filesystem not in RhodeCode database
 
        if not db_repo:
 
            log.info('repository %s not found creating now' % name)
 
            added.append(name)
 
            desc = (repo.description
 
                    if repo.description != 'unknown'
 
                    else '%s repository' % name)
 
            new_repo = rm.create_repo(
 
                repo_name=name,
 
                repo_type=repo.alias,
 
                description=desc,
 
                repos_group=getattr(group, 'group_id', None),
 
                owner=user,
 
                just_db=True
 
            )
 
            # we added that repo just now, and make sure it has githook
 
            # installed
 
            if new_repo.repo_type == 'git':
 
                ScmModel().install_git_hook(new_repo.scm_instance)
 
        elif install_git_hook:
 
            if db_repo.repo_type == 'git':
 
                ScmModel().install_git_hook(db_repo.scm_instance)
 
        # during starting install all cache keys for all repositories in the
 
        # system, this will register all repos and multiple instances
 
        key, _prefix, _org_key = CacheInvalidation._get_key(name)
 
        CacheInvalidation._get_or_create_key(key, _prefix, _org_key, commit=False)
 
    sa.commit()
 
    removed = []
 
    if remove_obsolete:
 
        # remove from database those repositories that are not in the filesystem
 
        for repo in sa.query(Repository).all():
 
            if repo.repo_name not in initial_repo_list.keys():
 
                log.debug("Removing non existing repository found in db `%s`" %
 
                          repo.repo_name)
 
                try:
 
                    sa.delete(repo)
 
                    sa.commit()
 
                    removed.append(repo.repo_name)
 
                except:
 
                    #don't hold further removals on error
 
                    log.error(traceback.format_exc())
 
                    sa.rollback()
 

	
 
    # clear cache keys
 
    log.debug("Clearing cache keys now...")
 
    CacheInvalidation.clear_cache()
 
    sa.commit()
 
    return added, removed
 

	
 

	
 
# set cache regions for beaker so celery can utilise it
 
def add_cache(settings):
 
    cache_settings = {'regions': None}
 
    for key in settings.keys():
 
        for prefix in ['beaker.cache.', 'cache.']:
 
            if key.startswith(prefix):
 
                name = key.split(prefix)[1].strip()
 
                cache_settings[name] = settings[key].strip()
 
    if cache_settings['regions']:
 
        for region in cache_settings['regions'].split(','):
 
            region = region.strip()
 
            region_settings = {}
 
            for key, value in cache_settings.items():
 
                if key.startswith(region):
 
                    region_settings[key.split('.')[1]] = value
 
            region_settings['expire'] = int(region_settings.get('expire',
 
                                                                60))
 
            region_settings.setdefault('lock_dir',
 
                                       cache_settings.get('lock_dir'))
 
            region_settings.setdefault('data_dir',
 
                                       cache_settings.get('data_dir'))
 

	
 
            if 'type' not in region_settings:
 
                region_settings['type'] = cache_settings.get('type',
 
                                                             'memory')
 
            beaker.cache.cache_regions[region] = region_settings
 

	
 

	
 
def load_rcextensions(root_path):
 
    import rhodecode
 
    from rhodecode.config import conf
 

	
 
    path = os.path.join(root_path, 'rcextensions', '__init__.py')
 
    if os.path.isfile(path):
 
        rcext = create_module('rc', path)
 
        EXT = rhodecode.EXTENSIONS = rcext
 
        log.debug('Found rcextensions now loading %s...' % rcext)
 

	
 
        # Additional mappings that are not present in the pygments lexers
 
        conf.LANGUAGES_EXTENSIONS_MAP.update(getattr(EXT, 'EXTRA_MAPPINGS', {}))
 

	
 
        #OVERRIDE OUR EXTENSIONS FROM RC-EXTENSIONS (if present)
 

	
 
        if getattr(EXT, 'INDEX_EXTENSIONS', []) != []:
 
            log.debug('settings custom INDEX_EXTENSIONS')
 
            conf.INDEX_EXTENSIONS = getattr(EXT, 'INDEX_EXTENSIONS', [])
 

	
 
        #ADDITIONAL MAPPINGS
 
        log.debug('adding extra into INDEX_EXTENSIONS')
 
        conf.INDEX_EXTENSIONS.extend(getattr(EXT, 'EXTRA_INDEX_EXTENSIONS', []))
 

	
 

	
 
#==============================================================================
 
# TEST FUNCTIONS AND CREATORS
 
#==============================================================================
 
def create_test_index(repo_location, config, full_index):
 
    """
 
    Makes default test index
 

	
 
    :param config: test config
 
    :param full_index:
 
    """
 

	
 
    from rhodecode.lib.indexers.daemon import WhooshIndexingDaemon
 
    from rhodecode.lib.pidlock import DaemonLock, LockHeld
 

	
 
    repo_location = repo_location
 

	
 
    index_location = os.path.join(config['app_conf']['index_dir'])
 
    if not os.path.exists(index_location):
 
        os.makedirs(index_location)
 

	
 
    try:
 
        l = DaemonLock(file_=jn(dn(index_location), 'make_index.lock'))
 
        WhooshIndexingDaemon(index_location=index_location,
 
                             repo_location=repo_location)\
 
            .run(full_index=full_index)
 
        l.release()
 
    except LockHeld:
 
        pass
 

	
 

	
 
def create_test_env(repos_test_path, config):
 
    """
 
    Makes a fresh database and
 
    install test repository into tmp dir
 
    """
 
    from rhodecode.lib.db_manage import DbManage
 
    from rhodecode.tests import HG_REPO, GIT_REPO, TESTS_TMP_PATH
 

	
 
    # PART ONE create db
 
    dbconf = config['sqlalchemy.db1.url']
 
    log.debug('making test db %s' % dbconf)
rhodecode/model/db.py
Show inline comments
 
@@ -1332,198 +1332,199 @@ class UsersGroupRepoGroupToPerm(Base, Ba
 
    users_group_id = Column("users_group_id", Integer(), ForeignKey('users_groups.users_group_id'), nullable=False, unique=None, default=None)
 
    group_id = Column("group_id", Integer(), ForeignKey('groups.group_id'), nullable=False, unique=None, default=None)
 
    permission_id = Column("permission_id", Integer(), ForeignKey('permissions.permission_id'), nullable=False, unique=None, default=None)
 

	
 
    users_group = relationship('UsersGroup')
 
    permission = relationship('Permission')
 
    group = relationship('RepoGroup')
 

	
 

	
 
class Statistics(Base, BaseModel):
 
    __tablename__ = 'statistics'
 
    __table_args__ = (
 
         UniqueConstraint('repository_id'),
 
         {'extend_existing': True, 'mysql_engine': 'InnoDB',
 
          'mysql_charset': 'utf8'}
 
    )
 
    stat_id = Column("stat_id", Integer(), nullable=False, unique=True, default=None, primary_key=True)
 
    repository_id = Column("repository_id", Integer(), ForeignKey('repositories.repo_id'), nullable=False, unique=True, default=None)
 
    stat_on_revision = Column("stat_on_revision", Integer(), nullable=False)
 
    commit_activity = Column("commit_activity", LargeBinary(1000000), nullable=False)#JSON data
 
    commit_activity_combined = Column("commit_activity_combined", LargeBinary(), nullable=False)#JSON data
 
    languages = Column("languages", LargeBinary(1000000), nullable=False)#JSON data
 

	
 
    repository = relationship('Repository', single_parent=True)
 

	
 

	
 
class UserFollowing(Base, BaseModel):
 
    __tablename__ = 'user_followings'
 
    __table_args__ = (
 
        UniqueConstraint('user_id', 'follows_repository_id'),
 
        UniqueConstraint('user_id', 'follows_user_id'),
 
        {'extend_existing': True, 'mysql_engine': 'InnoDB',
 
         'mysql_charset': 'utf8'}
 
    )
 

	
 
    user_following_id = Column("user_following_id", Integer(), nullable=False, unique=True, default=None, primary_key=True)
 
    user_id = Column("user_id", Integer(), ForeignKey('users.user_id'), nullable=False, unique=None, default=None)
 
    follows_repo_id = Column("follows_repository_id", Integer(), ForeignKey('repositories.repo_id'), nullable=True, unique=None, default=None)
 
    follows_user_id = Column("follows_user_id", Integer(), ForeignKey('users.user_id'), nullable=True, unique=None, default=None)
 
    follows_from = Column('follows_from', DateTime(timezone=False), nullable=True, unique=None, default=datetime.datetime.now)
 

	
 
    user = relationship('User', primaryjoin='User.user_id==UserFollowing.user_id')
 

	
 
    follows_user = relationship('User', primaryjoin='User.user_id==UserFollowing.follows_user_id')
 
    follows_repository = relationship('Repository', order_by='Repository.repo_name')
 

	
 
    @classmethod
 
    def get_repo_followers(cls, repo_id):
 
        return cls.query().filter(cls.follows_repo_id == repo_id)
 

	
 

	
 
class CacheInvalidation(Base, BaseModel):
 
    __tablename__ = 'cache_invalidation'
 
    __table_args__ = (
 
        UniqueConstraint('cache_key'),
 
        Index('key_idx', 'cache_key'),
 
        {'extend_existing': True, 'mysql_engine': 'InnoDB',
 
         'mysql_charset': 'utf8'},
 
    )
 
    cache_id = Column("cache_id", Integer(), nullable=False, unique=True, default=None, primary_key=True)
 
    cache_key = Column("cache_key", String(255, convert_unicode=False, assert_unicode=None), nullable=True, unique=None, default=None)
 
    cache_args = Column("cache_args", String(255, convert_unicode=False, assert_unicode=None), nullable=True, unique=None, default=None)
 
    cache_active = Column("cache_active", Boolean(), nullable=True, unique=None, default=False)
 

	
 
    def __init__(self, cache_key, cache_args=''):
 
        self.cache_key = cache_key
 
        self.cache_args = cache_args
 
        self.cache_active = False
 

	
 
    def __unicode__(self):
 
        return u"<%s('%s:%s')>" % (self.__class__.__name__,
 
                                  self.cache_id, self.cache_key)
 

	
 
    @classmethod
 
    def clear_cache(cls):
 
        cls.query().delete()
 

	
 
    @classmethod
 
    def _get_key(cls, key):
 
        """
 
        Wrapper for generating a key, together with a prefix
 

	
 
        :param key:
 
        """
 
        import rhodecode
 
        prefix = ''
 
        iid = rhodecode.CONFIG.get('instance_id')
 
        if iid:
 
            prefix = iid
 
        return "%s%s" % (prefix, key), prefix, key.rstrip('_README')
 

	
 
    @classmethod
 
    def get_by_key(cls, key):
 
        return cls.query().filter(cls.cache_key == key).scalar()
 

	
 
    @classmethod
 
    def _get_or_create_key(cls, key, prefix, org_key):
 
    def _get_or_create_key(cls, key, prefix, org_key, commit=True):
 
        inv_obj = Session().query(cls).filter(cls.cache_key == key).scalar()
 
        if not inv_obj:
 
            try:
 
                inv_obj = CacheInvalidation(key, org_key)
 
                Session().add(inv_obj)
 
                if commit:
 
                Session().commit()
 
            except Exception:
 
                log.error(traceback.format_exc())
 
                Session().rollback()
 
        return inv_obj
 

	
 
    @classmethod
 
    def invalidate(cls, key):
 
        """
 
        Returns Invalidation object if this given key should be invalidated
 
        None otherwise. `cache_active = False` means that this cache
 
        state is not valid and needs to be invalidated
 

	
 
        :param key:
 
        """
 

	
 
        key, _prefix, _org_key = cls._get_key(key)
 
        inv = cls._get_or_create_key(key, _prefix, _org_key)
 

	
 
        if inv and inv.cache_active is False:
 
            return inv
 

	
 
    @classmethod
 
    def set_invalidate(cls, key):
 
        """
 
        Mark this Cache key for invalidation
 

	
 
        :param key:
 
        """
 

	
 
        key, _prefix, _org_key = cls._get_key(key)
 
        inv_objs = Session().query(cls).filter(cls.cache_args == _org_key).all()
 
        log.debug('marking %s key[s] %s for invalidation' % (len(inv_objs),
 
                                                             _org_key))
 
        try:
 
            for inv_obj in inv_objs:
 
                if inv_obj:
 
                    inv_obj.cache_active = False
 

	
 
                Session().add(inv_obj)
 
            Session().commit()
 
        except Exception:
 
            log.error(traceback.format_exc())
 
            Session().rollback()
 

	
 
    @classmethod
 
    def set_valid(cls, key):
 
        """
 
        Mark this cache key as active and currently cached
 

	
 
        :param key:
 
        """
 
        inv_obj = cls.get_by_key(key)
 
        inv_obj.cache_active = True
 
        Session().add(inv_obj)
 
        Session().commit()
 

	
 
    @classmethod
 
    def get_cache_map(cls):
 

	
 
        class cachemapdict(dict):
 

	
 
            def __init__(self, *args, **kwargs):
 
                fixkey = kwargs.get('fixkey')
 
                if fixkey:
 
                    del kwargs['fixkey']
 
                self.fixkey = fixkey
 
                super(cachemapdict, self).__init__(*args, **kwargs)
 

	
 
            def __getattr__(self, name):
 
                key = name
 
                if self.fixkey:
 
                    key, _prefix, _org_key = cls._get_key(key)
 
                if key in self.__dict__:
 
                    return self.__dict__[key]
 
                else:
 
                    return self[key]
 

	
 
            def __getitem__(self, key):
 
                if self.fixkey:
 
                    key, _prefix, _org_key = cls._get_key(key)
 
                try:
 
                    return super(cachemapdict, self).__getitem__(key)
 
                except KeyError:
 
                    return
 

	
 
        cache_map = cachemapdict(fixkey=True)
 
        for obj in cls.query().all():
 
            cache_map[obj.cache_key] = cachemapdict(obj.get_dict())
 
        return cache_map
 

	
 

	
 
class ChangesetComment(Base, BaseModel):
 
    __tablename__ = 'changeset_comments'
 
    __table_args__ = (
 
        Index('cc_revision_idx', 'revision'),
0 comments (0 inline, 0 general)