Duplicate buckets in event_statements_histogram_by_digest

[Matt Nolf]

Hi,
I am hoping to use the event_statements_histogram_by_digest table to extract histogram data about my queries - so that I can calculate quantiles (above what is available in event_statements_summary).

However when I look at the buckets, for a few digests there are actually 900 buckets. Two sets of 0->449. One set contains data, while the other has 0 for COUNT and COUNT_BUCKET_AND_LOWER.

Furthermore, I get inconsistent results when querying the table. When I search for a specific schema_name AND digest AND bucket_number I get 1 row, however if I use a range on the bucket_number >= x AND range < x+1 I get 2 rows - one with data, the other 0 values.

For example:

mysql> SELECT * FROM events_statements_histogram_by_digest WHERE SCHEMA_NAME = 'X' AND DIGEST = 'Y' AND BUCKET_NUMBER >= 449 AND BUCKET_NUMBER < 500;
+------------------+------------------------------------------------------------------+---------------+------------------+----------------------+--------------+------------------------+-----------------+
| SCHEMA_NAME | DIGEST | BUCKET_NUMBER | BUCKET_TIMER_LOW | BUCKET_TIMER_HIGH | COUNT_BUCKET | COUNT_BUCKET_AND_LOWER | BUCKET_QUANTILE |
+------------------+------------------------------------------------------------------+---------------+------------------+----------------------+--------------+------------------------+-----------------+
| X | Y | 449 | 9120108393559408 | 18446744073709551615 | 0 | 18930820 | 1.000000 |
| X | Y | 449 | 9120108393559408 | 18446744073709551615 | 0 | 0 | 0.000000 |
+------------------+------------------------------------------------------------------+---------------+------------------+----------------------+--------------+------------------------+-----------------+
2 rows in set (0.26 sec)

mysql> SELECT * FROM events_statements_histogram_by_digest WHERE SCHEMA_NAME = 'X' AND DIGEST = 'Y' AND BUCKET_NUMBER = 449;
+------------------+------------------------------------------------------------------+---------------+------------------+----------------------+--------------+------------------------+-----------------+
| SCHEMA_NAME | DIGEST | BUCKET_NUMBER | BUCKET_TIMER_LOW | BUCKET_TIMER_HIGH | COUNT_BUCKET | COUNT_BUCKET_AND_LOWER | BUCKET_QUANTILE |
+------------------+------------------------------------------------------------------+---------------+------------------+----------------------+--------------+------------------------+-----------------+
| X | Y | 449 | 9120108393559408 | 18446744073709551615 | 0 | 18930820 | 1.000000 |
+------------------+------------------------------------------------------------------+---------------+------------------+----------------------+--------------+------------------------+-----------------+
1 row in set (0.12 sec)


Why is this? And how can I reliably query this table to collect all buckets for a given query.