unit tests fail on mysql due to timestamp dropping microseconds
Bug #1178041 reported by
Adam Young
This bug affects 1 person
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
OpenStack Identity (keystone) |
Fix Released
|
Medium
|
Adam Young |
Bug Description
FAIL: test_null_
-------
Traceback (most recent call last):
File "/opt/stack/
self.assertEqua
AssertionError: {'a': 'b', 'user_id': 'testuserid', 'expires': datetime.
- {'a': 'b',
+ {u'a': u'b',
Changed in keystone: | |
assignee: | nobody → Jamie Lennox (jamielennox) |
Changed in keystone: | |
status: | New → In Progress |
Changed in keystone: | |
assignee: | Jamie Lennox (jamielennox) → nobody |
Changed in keystone: | |
assignee: | nobody → Adam Young (ayoung) |
Changed in keystone: | |
milestone: | none → havana-1 |
status: | Fix Committed → Fix Released |
Changed in keystone: | |
importance: | Undecided → Medium |
Changed in keystone: | |
milestone: | havana-1 → 2013.2 |
To post a comment you must log in.
Hi, we had a long chat on IRC about this.
tl;dr: I think there are valid cases for large numbers of multiple tokens for the same user particularly when deploying large cluster systems - one token per node allows for granular revocation when systems fail. So I'd like to be able to deploy a 5K node cluster where each node has a keystone token - in less than 90 minutes : it is an embarassingly distributed problem
From that conversation it sounds like going back to microseconds would work well, or you could use a salt : one way to generate unique salts in a distributed environment would be to use uuidgen, or something like snowflake.