Zane Bitter e4d2452f2f Always use string ID for WaitConditionHandle signals ... Using an integer as the ID - which is what we currently do for OS::Heat::WaitConditionHandles without an ID explicitly supplied by the signaller - means that we can implicitly create an entry that will collide with an explicitly specified string ID when we serialise to and then deserialise from JSON to store the metadata in the database. These types of collisions are silent and avoid being logged (despite the fact that we attempt to catch and log collisions), which makes any issues very, very difficult to debug. Change-Id: Ia613e79dc9757c4594b84ed2105bfc1357d1e037 Related-Bug: #1738653		2018-02-09 20:20:22 -05:00
..
aws	Merge "Refactor CooldownMixin"	2018-01-12 21:19:24 +00:00
openstack	Merge "Get ResourceGroup/Chain attributes from nested stack outputs"	2018-02-09 07:28:48 +00:00
__init__.py	Log missing clients as WARNING instead of ERROR	2016-08-16 16:27:55 -05:00
alarm_base.py	Update incorrect timezone description	2017-09-19 13:46:32 +08:00
scheduler_hints.py	Fix path_in_stack for scheduler hints	2016-08-05 10:21:10 +00:00
server_base.py	Merge "Mark Server os_collect_config attribute as CACHE_NONE"	2017-06-27 16:51:35 +00:00
signal_responder.py	Allow for keeping prefix in heat_url	2017-06-07 20:09:39 +05:30
stack_resource.py	Avoid always loading nested stack on update	2018-01-08 18:19:10 -05:00
stack_user.py	Remove log translations	2017-03-25 17:11:50 +08:00
template_resource.py	Cache the TemplateResource reference ID like an attribute	2018-01-11 22:14:20 -05:00
volume_base.py	Merge "Deal with cinder-backup service absent"	2016-08-18 17:32:03 +00:00
wait_condition.py	Always use string ID for WaitConditionHandle signals	2018-02-09 20:20:22 -05:00