Max retries exceeded with StaleDataError in standardattributes during live migration

[Ryan Tidwell]

I am observing StaleDataError on stable/pike during live migration, causing live migration to fail. It occurs when attempting to live-migrate a handful of VM's (5-6 VM's is all it takes) in rapid succession from the same source to the same target. This quick and dirty script is able to make the issue appear reliably:

for i in `openstack server list --all-projects --host <origin> -c ID -f value`; do openstack server migrate $i --live <target>; done

From the neutron server logs:

DB exceeded retry limit.: StaleDataError: UPDATE statement on table 'standardattributes' expected to update 1 row(s); 0 were matched.
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api Traceback (most recent call last):
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api File "/opt/stack/venv/neutron-20181030T130300Z/lib/python2.7/site-packages/oslo_db/api.py", line 138, in wrapper
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api return f(*args, **kwargs)
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api File "/opt/stack/venv/neutron-20181030T130300Z/lib/python2.7/site-packages/neutron/db/api.py", line 128, in wrapped
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api LOG.debug("Retry wrapper got retriable exception: %s", e)
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api File "/opt/stack/venv/neutron-20181030T130300Z/lib/python2.7/site-packages/oslo_utils/excutils.py", line 220, in __exit__
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api self.force_reraise()
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api File "/opt/stack/venv/neutron-20181030T130300Z/lib/python2.7/site-packages/oslo_utils/excutils.py", line 196, in force_reraise
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api six.reraise(self.type_, self.value, self.tb)
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api File "/opt/stack/venv/neutron-20181030T130300Z/lib/python2.7/site-packages/neutron/db/api.py", line 124, in wrapped
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api return f(*dup_args, **dup_kwargs)
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api File "/opt/stack/venv/neutron-20181030T130300Z/lib/python2.7/site-packages/neutron/plugins/ml2/plugin.py", line 1346, in update_port
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api mech_context, attrs)
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api File "/opt/stack/venv/neutron-20181030T130300Z/lib/python2.7/site-packages/neutron/plugins/ml2/plugin.py", line 354, in _process_port_binding
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api db.clear_binding_levels(plugin_context, port_id, original_host)
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api File "/opt/stack/venv/neutron-20181030T130300Z/lib/python2.7/site-packages/oslo_db/sqlalchemy/enginefacade.py", line 979, in wrapper
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api return fn(*args, **kwargs)
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api File "/usr/lib64/python2.7/contextlib.py", line 24, in __exit__
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api self.gen.next()
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api File "/opt/stack/venv/neutron-20181030T130300Z/lib/python2.7/site-packages/oslo_db/sqlalchemy/enginefacade.py", line 1029, in _transaction_scope
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api yield resource
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api File "/usr/lib64/python2.7/contextlib.py", line 24, in __exit__
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api self.gen.next()
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api File "/opt/stack/venv/neutron-20181030T130300Z/lib/python2.7/site-packages/oslo_db/sqlalchemy/enginefacade.py", line 655, in _session
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api self.session.flush()
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api File "/opt/stack/venv/neutron-20181030T130300Z/lib/python2.7/site-packages/sqlalchemy/orm/session.py", line 2171, in flush
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api self._flush(objects)
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api File "/opt/stack/venv/neutron-20181030T130300Z/lib/python2.7/site-packages/sqlalchemy/orm/session.py", line 2291, in _flush
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api transaction.rollback(_capture_exception=True)
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api File "/opt/stack/venv/neutron-20181030T130300Z/lib/python2.7/site-packages/sqlalchemy/util/langhelpers.py", line 66, in __exit__
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api compat.reraise(exc_type, exc_value, exc_tb)
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api File "/opt/stack/venv/neutron-20181030T130300Z/lib/python2.7/site-packages/sqlalchemy/orm/session.py", line 2255, in _flush
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api flush_context.execute()
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api File "/opt/stack/venv/neutron-20181030T130300Z/lib/python2.7/site-packages/sqlalchemy/orm/unitofwork.py", line 389, in execute
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api rec.execute(self)
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api File "/opt/stack/venv/neutron-20181030T130300Z/lib/python2.7/site-packages/sqlalchemy/orm/unitofwork.py", line 548, in execute
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api uow
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api File "/opt/stack/venv/neutron-20181030T130300Z/lib/python2.7/site-packages/sqlalchemy/orm/persistence.py", line 177, in save_obj
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api mapper, table, update)
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api File "/opt/stack/venv/neutron-20181030T130300Z/lib/python2.7/site-packages/sqlalchemy/orm/persistence.py", line 760, in _emit_update_statements
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api (table.description, len(records), rows))
2019-01-24 09:57:34.959 255478 ERROR oslo_db.api StaleDataError: UPDATE statement on table 'standardattributes' expected to update 1 row(s); 0 were matched.

It appears that somewhere in clear_binding_levels() there is some contention on the update to the standardattributes record for the port. I can confirm this only on stable/pike, I have not had a chance to confirm on master.

[Swaminathan Vasudevan (swaminathan-vasudevan)]

You mentioned that it is trying to migrate from the same source to the same target. So we are just removing and adding it to the same node?

[Ryan Tidwell (ryan-tidwell)]

Sorry, that should have been more clear. The source and target are 2 different hosts, but all VM's are being moved to the same place.

[Ryan Tidwell (ryan-tidwell)]

These could be related as well:

https://bugs.launchpad.net/neutron/+bug/1716321
https://bugs.launchpad.net/neutron/+bug/1738337