diff options
| author | Clark Boylan <clark.boylan@gmail.com> | 2012-12-06 17:07:55 -0800 |
|---|---|---|
| committer | Clark Boylan <clark.boylan@gmail.com> | 2012-12-06 17:07:55 -0800 |
| commit | fa5e490e4687349685d50c9ce93bc289be6a61bf (patch) | |
| tree | 4ae1dd7dc439ab06ff7ff0ef95245784eba5101b | |
| parent | f495b95d0ee28141bd067ca64a0176189f5d8293 (diff) | |
| download | nova-fa5e490e4687349685d50c9ce93bc289be6a61bf.tar.gz nova-fa5e490e4687349685d50c9ce93bc289be6a61bf.tar.xz nova-fa5e490e4687349685d50c9ce93bc289be6a61bf.zip | |
Order instance faults by created_at and id.
It is possible that instance faults have the same created_at values so
ordering on that field alone is not deterministic. Add an additional
order by clause to order descending on id as well.
I am not sure if the order by exists to make tests pass; however,
without the additional "id" the tests do occasionally fail. See
http://logs.openstack.org/15078/18/check/gate-nova-python26/10709/console.html
for an example failure. If order does not actually matter we may want to
make the test smarter instead.
Part of blueprint grizzly-testtools
Change-Id: I1c7a07669846c832270631c1734dae509dba7586
| -rw-r--r-- | nova/db/sqlalchemy/api.py | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/nova/db/sqlalchemy/api.py b/nova/db/sqlalchemy/api.py index 4155fdc3d..537ebaa9e 100644 --- a/nova/db/sqlalchemy/api.py +++ b/nova/db/sqlalchemy/api.py @@ -4536,7 +4536,7 @@ def instance_fault_get_by_instance_uuids(context, instance_uuids): rows = model_query(context, models.InstanceFault, read_deleted='no').\ filter(models.InstanceFault.instance_uuid.in_( instance_uuids)).\ - order_by(desc("created_at")).\ + order_by(desc("created_at"), desc("id")).\ all() output = {} |
