statistics shouldn't try to query samples before time of earliest sample

[Liusheng]

In mysql scenario, when I use:

ceilometer statistics -m instance -p 36000 -q "start=2012-03-20T04:07:57;end=2014-03-10T06:17:57"

the get_meter_statistics() method will try to query sample before time of earliest sample.

I have used pdb to verify this:

        for period_start, period_end in base.iter_period(
                sample_filter.start or res.tsmin,
                sample_filter.end or res.tsmax,
                period):
            q = query.filter(models.Sample.timestamp >= period_start)
            q = q.filter(models.Sample.timestamp < period_end)
            import pdb
            pdb.set_trace()
            for r in q.all():

-----------------------------------------------------------------------------------------------------------------
> /opt/stack/ceilometer/ceilometer/storage/impl_sqlalchemy.py(820)get_meter_statistics()
-> for r in q.all():
(Pdb) c
> /opt/stack/ceilometer/ceilometer/storage/impl_sqlalchemy.py(819)get_meter_statistics()
-> pdb.set_trace()
(Pdb) p period_start
datetime.datetime(2012, 3, 20, 14, 7, 57)

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/82048