timeutils is not leap second safe

[Ihar Hrachyshka]

We are having a leap second in June: http://www.usatoday.com/story/tech/2015/01/08/computer-chaos-feares/21433363/

It will mean we'll have 61 seconds at that moment. datetime module is not leap aware and will fail with "ValueError: second must be in 0..59" if seconds=60 is passed. The proper response to it would be to avoid using datetime and switch to time module that operates with struct_time objects.

The problem is that timeutils API is bound to datetime. Specifically, the following functions are affected:

- parse_strtime
- utcnow
- unmarshall_time

This may mean that we can't fix it as a bug, but may need to introduce new proper time API.

An easy way to check that we're affected is e.g.:

+ def test_unmarshall_time_leap_second(self):
+ leap_time = dict(day=30, month=6, year=2015,
+ hour=23, minute=59, second=60, microsecond=0)
+ leap_datetime = timeutils.unmarshall_time(leap_time)
+

[ChangBo Guo(gcb) (glongwave)]

Fix c50d8161ac251ee0c9cad6012bc7ceab4f899d6f was merged , so we can close it now.

[Davanum Srinivas (DIMS) (dims-v)]

This issue was fixed in the openstack/oslo.utils 1.4.2 release.

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/oslo.utils 1.4.2 release.