openstack
/
neutron
Code Issues Proposed changes

Fixes DBDeadlock race condition during driver initialization. 

When multiple controllers are initially built, the ml2_vlan_allocations
will be empty and the neutron servers under different controllers will
try to populate the entry at the same time. This will cause this
DBDeadlock error as they try to access the db at the same time.

This patch set is to add db_api.retry_db_error decorator in the
_sync_vlan_allocations method to avoid this race condition.

Change-Id: I7c0e3ae515f592a5852e6decf6820b103f146761
Closes-bug: 1617499
(cherry picked from commit 55b6c6a63a)
This commit is contained in:
Kahou Lei 2016-08-26 17:43:38 -07:00 committed by Jakub Libosvar
parent 608eee3814
commit 2da513f856
1 changed files with 1 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
neutron/plugins/ml2/drivers/type_vlan.py
View File

@ -96,6 +96,7 @@ class VlanTypeDriver(helpers.SegmentTypeDriver):
sys.exit(1)
LOG.info(_LI("Network VLAN ranges: %s"), self.network_vlan_ranges)
@db_api.retry_db_errors
def _sync_vlan_allocations(self):
session = db_api.get_session()
with session.begin(subtransactions=True):