On Friday 17 October 2014 09:11:54 Josh Fisher wrote:
>Generally, schedules are starting jobs at the same time of day. It follows that the time period between successive runs will tend to be very nearly an integer number of days. If the use duration is also an integer number of days, then it can be expected that a volume's use duration will expire at very nearly the exact time that a job using a volume from that pool is scheduled to run. This is alleviated by configuring a use duration that is not an integer multiple of days. For example, instead of using a use duration of "7 days", set use duration to "156 hours" (6 days 12 hours). The hysteresis makes it far less likely that a job will be launched at exactly the same time that a volume's use duration is expiring.
>
Beside understanding the reason why the storage daemon recycled
a non empty volume, you advice seems to go in the direction to
minimime the probability this accident, use duration expiring
at the beginning of a job, would repeat.
I'll try it, but what about "9367 minutes" = "6 days 12h 7m"?
Wouldn't be better to achieve this goal?
G. Vitillaro. |