Splunk not rolling hot buckets, maxHotSpanSecs not respected

[sylim_splunk]

I have a configuration, maxHotSpanSecs = 86399 for an index namded board, expecting the buckets keep a day amount of data and rolls as per the config.

[board]
homePath= D:\Program Files\Splunk\var\lib\splunk\board\db
maxHotSpanSecs = 86399
maxHotBuckets = 1

But the buckets are not rotating even the gap between the earliest and latest timestamps are over 86399 - find the below search output;

Search query : "|dbinspect index=<INDEXNAM> |eval SpanSec = endEpoch - startEpoch"
Output :
||SpanSec | id | path||
|14457 | 0 | E:\Program Files\Splunk\var\lib\splunk\board\colddb\db_1445912605_1445898148_0 |
|147 | 1 | E:\Program Files\Splunk\var\lib\splunk\board\colddb\db_1445912745_1445912598_1 |
|1001 | 2 | E:\Program Files\Splunk\var\lib\splunk\board\colddb\db_1445913743_1445912742_2 |
|42371533 | 3 | E:\Program Files\Splunk\var\lib\splunk\board\colddb\db_1445914570_1403543037_3 |

Please advise;

[sylim_splunk]

When maxHotBuckets is set to one, it is a special case, likely so we don't have an explosion of very small buckets created for a data set that jumps all over the place with respect to time.
In this case we go out of our way to accept the event into the single hot bucket regardless of the time (especially so for events earlier than the bucket's earliest event) because it is our only option and we want to avoid creating another hot bucket if possible.

[jeffland]

This should be noted here.