Actually, I see another problem, the final date in the date range is not generated correctly.
I used this schema:
I did a date range of 09/01/2016-09/22/2016 100,000 rows and there was no data for 09/22/2016.
Then I did a date range of 09/23/2016 - 09/28/2016 100,000 rows and there was only one entry for 09/28/2016
Here is the distribution by date and the corresponding count for each date:
2016-09-01 4806
2016-09-02 4714
2016-09-03 4688
2016-09-04 4738
2016-09-05 4709
2016-09-06 4700
2016-09-07 4821
2016-09-08 4850
2016-09-09 4738
2016-09-10 4855
2016-09-11 4710
2016-09-12 4809
2016-09-13 4789
2016-09-14 4716
2016-09-15 4695
2016-09-16 4722
2016-09-17 4843
2016-09-18 4884
2016-09-19 4648
2016-09-20 4803
2016-09-21 4762
2016-09-23 19790
2016-09-24 20010
2016-09-25 20013
2016-09-26 20105
2016-09-27 20081
2016-09-28 1
The single row of data for 09/28 was this:
‘2016-09-28 00:00:00’