Hi! I thought about this problem again and I think I have a solution.

On 02.12.2024 23:12, Alena Rybakina wrote:

* I've read the previous discussion on how important to keep all these
fields regarding vacuum statistics including points by Andrei and Jim.
It still worrying me that statistics volume is going to burst in about
3 times, but I don't have a particular proposal on how to make more
granular approach.  I wonder if you could propose something.

We can collect statistics on databases at all times - there are less compared to vacuum statistics of relations, but they can give enough information that can hint that something is going wrong.
With the track_vacuum_statistics guc we can cover cases of collecting extended and complete information: when it is enabled, we will collect vacuum statistics on relations both: heaps and indexes.
This will not lead to a synchronicity between constant database statistics and temporary statistics of relations, since our vacuum statistics are cumulative and it is assumed that we will look at changes in statistics over a certain period.
What do you think?

-- 
Regards,
Alena Rybakina
Postgres Professional