Comment 11 for bug 1548909

note to self (and others). this error is caused by the following policies:

+---------+-------------+-----------------------------------------------------------------------------+------------------------------------------------+
| name | back_window | definition | aggregation_methods |
+---------+-------------+-----------------------------------------------------------------------------+------------------------------------------------+
| default | 0 | - points: 288, granularity: 0:05:00, timespan: 1 day, 0:00:00 | std, count, 95pct, min, max, sum, median, mean |
| | | - points: 420, granularity: 0:24:00, timespan: 7 days, 0:00:00 | |
| | | - points: 240, granularity: 2:48:00, timespan: 28 days, 0:00:00 | |
| | | - points: 720, granularity: 12:00:00, timespan: 360 days, 0:00:00 | |
| | | - points: 720, granularity: 1:00:00, timespan: 30 days, 0:00:00 | |
| | | - points: 360, granularity: 1 day, 0:00:00, timespan: 360 days, 0:00:00 | |
| | | - points: 102, granularity: 7 days, 0:00:00, timespan: 714 days, 0:00:00 | |
| | | - points: 60, granularity: 30 days, 0:00:00, timespan: 1800 days, 0:00:00 | |
| | | - points: 60, granularity: 90 days, 0:00:00, timespan: 5400 days, 0:00:00 | |
| | | - points: 60, granularity: 180 days, 0:00:00, timespan: 10800 days, 0:00:00 | |
| | | - points: 60, granularity: 360 days, 0:00:00, timespan: 21600 days, 0:00:00 | |
+---------+-------------+-----------------------------------------------------------------------------+------------------------------------------------+

it's probably the 60 year timestamp+---------+-------------+-----------------------------------------------------------------------------+------------------------------------------------+
| name | back_window | definition | aggregation_methods |
+---------+-------------+-----------------------------------------------------------------------------+------------------------------------------------+
| default | 0 | - points: 288, granularity: 0:05:00, timespan: 1 day, 0:00:00 | std, count, 95pct, min, max, sum, median, mean |
| | | - points: 420, granularity: 0:24:00, timespan: 7 days, 0:00:00 | |
| | | - points: 240, granularity: 2:48:00, timespan: 28 days, 0:00:00 | |
| | | - points: 720, granularity: 12:00:00, timespan: 360 days, 0:00:00 | |
| | | - points: 720, granularity: 1:00:00, timespan: 30 days, 0:00:00 | |
| | | - points: 360, granularity: 1 day, 0:00:00, timespan: 360 days, 0:00:00 | |
| | | - points: 102, granularity: 7 days, 0:00:00, timespan: 714 days, 0:00:00 | |
| | | - points: 60, granularity: 30 days, 0:00:00, timespan: 1800 days, 0:00:00 | |
| | | - points: 60, granularity: 90 days, 0:00:00, timespan: 5400 days, 0:00:00 | |
| | | - points: 60, granularity: 180 days, 0:00:00, timespan: 10800 days, 0:00:00 | |
| | | - points: 60, granularity: 360 days, 0:00:00, timespan: 21600 days, 0:00:00 | |
+---------+-------------+-----------------------------------------------------------------------------+------------------------------------------------+

basically the overflow is caused by two reasons:

1. having a ridiculously high sampling value so that when freq is passed to round_timestamp[1], it rounds the value to something way too old. sampling * POINTS_PER_SPLIT * 10e8 ends up being ~= 4.478976e+20
2. the ts passed in to round_timestamp is very old (1956-03-15 00:00:00).

a combination of 1+2 == OverflowError. we should have a check to ensure policies are valid.

[1] https://github.com/openstack/gnocchi/blob/6ca313dec3f4980defa3af9bd58921fe331849d0/gnocchi/carbonara.py#L303