Comment 7 for bug 1215676

Nova notifications have timestamps like: '_context_timestamp': '2013-08-24T14:31:06.154444'. This is similar to the unit test data setup.

After some poking around, there are a couple of places where this function is called without the higher precision flag (sub-second): https://github.com/openstack/ceilometer/blob/master/ceilometer/openstack/common/timeutils.py#L35-L44. Like here: https://github.com/openstack/ceilometer/blob/master/ceilometer/compute/pollsters/net.py#L59.

Also, it looks like up until the latter versions of MySQL (5.6.4+), sub-second values in temporal fields were truncated (http://dev.mysql.com/doc/refman/5.6/en/fractional-seconds.html). That's probably why Sandy just went with a Decimal representation in StackTach.

So, do you think we ought to fix this up to preserve the precision provided to CM from the source? From my perspective: it's destructive in that it removes a level of precision that already existed in the notification. I think it's important to preserve this for auditing and replay purposes.