From de8f74aec889e4d6398a228dc14d58c2b475cd86 Mon Sep 17 00:00:00 2001 From: Terry Wilson Date: Thu, 26 Jan 2023 08:37:24 -0600 Subject: [PATCH] Never raise an exception in notify() notify() is called from python-ovs code which is not built to recover from an exception in this user-overriden code. If there is an exception (e.g. the DB server is down when we process the hash ring), this exception can cause an unrecoverable error in processing OVSDB messages, rendering the neutron worker useless. Change-Id: I5f703d82175d71a222c76df37a82b5ccad890d14 (cherry picked from commit 67e616b2380d6549308a15077b2043721dbea5d0) (cherry picked from commit 848787785eb1140ee7d0eac72f3967b39345e625) Conflicts: neutron/plugins/ml2/drivers/ovn/mech_driver/ovsdb/ovsdb_monitor.py (cherry picked from commit 3566cc065eb7e811822a472bd40e37ecf7668971) --- .../ovn/mech_driver/ovsdb/ovsdb_monitor.py | 62 ++++++++++--------- 1 file changed, 33 insertions(+), 29 deletions(-) diff --git a/neutron/plugins/ml2/drivers/ovn/mech_driver/ovsdb/ovsdb_monitor.py b/neutron/plugins/ml2/drivers/ovn/mech_driver/ovsdb/ovsdb_monitor.py index 9962be8973a..50323bc4d8c 100644 --- a/neutron/plugins/ml2/drivers/ovn/mech_driver/ovsdb/ovsdb_monitor.py +++ b/neutron/plugins/ml2/drivers/ovn/mech_driver/ovsdb/ovsdb_monitor.py @@ -567,38 +567,42 @@ class OvnIdlDistributedLock(BaseOvnIdl): self._last_touch = None def notify(self, event, row, updates=None): - self.notify_handler.notify(event, row, updates, global_=True) try: - target_node = self._hash_ring.get_node(str(row.uuid)) - except exceptions.HashRingIsEmpty as e: - LOG.error('HashRing is empty, error: %s', e) - return - if target_node != self._node_uuid: - return - - # If the worker hasn't been health checked by the maintenance - # thread (see bug #1834498), indicate that it's alive here - time_now = timeutils.utcnow() - touch_timeout = time_now - datetime.timedelta( - seconds=ovn_const.HASH_RING_TOUCH_INTERVAL) - if not self._last_touch or touch_timeout >= self._last_touch: - # NOTE(lucasagomes): Guard the db operation with an exception - # handler. If heartbeating fails for whatever reason, log - # the error and continue with processing the event + self.notify_handler.notify(event, row, updates, global_=True) try: - ctx = neutron_context.get_admin_context() - ovn_hash_ring_db.touch_node(ctx, self._node_uuid) - self._last_touch = time_now - except Exception: - LOG.exception('Hash Ring node %s failed to heartbeat', - self._node_uuid) + target_node = self._hash_ring.get_node(str(row.uuid)) + except exceptions.HashRingIsEmpty as e: + LOG.error('HashRing is empty, error: %s', e) + return + if target_node != self._node_uuid: + return - LOG.debug('Hash Ring: Node %(node)s (host: %(hostname)s) ' - 'handling event "%(event)s" for row %(row)s ' - '(table: %(table)s)', - {'node': self._node_uuid, 'hostname': CONF.host, - 'event': event, 'row': row.uuid, 'table': row._table.name}) - self.notify_handler.notify(event, row, updates) + # If the worker hasn't been health checked by the maintenance + # thread (see bug #1834498), indicate that it's alive here + time_now = timeutils.utcnow() + touch_timeout = time_now - datetime.timedelta( + seconds=ovn_const.HASH_RING_TOUCH_INTERVAL) + if not self._last_touch or touch_timeout >= self._last_touch: + # NOTE(lucasagomes): Guard the db operation with an exception + # handler. If heartbeating fails for whatever reason, log + # the error and continue with processing the event + try: + ctx = neutron_context.get_admin_context() + ovn_hash_ring_db.touch_node(ctx, self._node_uuid) + self._last_touch = time_now + except Exception: + LOG.exception('Hash Ring node %s failed to heartbeat', + self._node_uuid) + + LOG.debug('Hash Ring: Node %(node)s (host: %(hostname)s) ' + 'handling event "%(event)s" for row %(row)s ' + '(table: %(table)s)', + {'node': self._node_uuid, 'hostname': CONF.host, + 'event': event, 'row': row.uuid, + 'table': row._table.name}) + self.notify_handler.notify(event, row, updates) + except Exception as e: + LOG.exception(e) @abc.abstractmethod def post_connect(self):