x/vmware-nsx
Code Issues Proposed changes

Get rid of additional db contention on fetching VIP

Browse Source 
This fix gets rid of fetching related record from ports table
that could potentially be blocked by concurrent operation
This is to prevent it from updating the
vip table while a port SQL lock is in place, which
vip entries have a reference to.

Partial-Bug: 1283522
Change-Id: Ic64cc268ff4817a31971734a44671c744047231b
This commit is contained in:
Kevin Benton 2014-03-20 00:24:50 -07:00 committed by Eugene Nikanorov
parent 28cfb4f15d
commit 05f6e3a428
1 changed files with 10 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
neutron/db/loadbalancer/loadbalancer_db.py
View File

@ -192,7 +192,16 @@ class LoadBalancerPluginDb(LoadBalancerPluginBase,
def update_status(self, context, model, id, status,
status_description=None):
with context.session.begin(subtransactions=True):
v_db = self._get_resource(context, model, id)
if issubclass(model, Vip):
try:
v_db = (self._model_query(context, model).
filter(model.id == id).
options(orm.noload('port')).
one())
except exc.NoResultFound:
raise loadbalancer.VipNotFound(vip_id=id)
else:
v_db = self._get_resource(context, model, id)
if v_db.status != status:
v_db.status = status
# update status_description in two cases: