On 2021/04/02 18:41, 蔡梦娟(玊于) wrote:
>
> Hi, all
>
> I want to know why call pgstat_reset_all function during recovery
> process, under what circumstances the data will be invalid after recovery?
If my understanding is right, PITR is the case. Now, the stats files are
generated as a one-time snapshot. This means that the stats counters
saved at last may not be valid for the specific point in time.
FWIW, there was a related discussion([1]) although the behavior is not
changed yet.
[1] https://www.postgresql.org/message-id/1416.1479760254%40sss.pgh.pa.us
Regards,
--
Masahiro Ikeda
NTT DATA CORPORATION