On 16.03.2026 11:45, Andrei Lepikhov wrote:
On 15/3/26 18:18, Andrey Borodin wrote:
On 13 Mar 2026, at 18:04, Alena Rybakina <[email protected]>
wrote:
I've decided to take a look into v31.
Overall idea of tracking VM dynamics seems good to me.
But the column naming for rev_all_visible_pages and rev_all_frozen_pages
seems strange to me. I've skimmed the thread but could not figure out
what
"rev_" stands for. Revisions? Revolutions? Reviews?
I suppose 'revert' is the exact term here. Someone decided to set the
flag, and we reverted his decision. Does this make sense to you?
Anyway, I always leave it in the natives' (and committers') hands.
I think renaming them to 'cleared' helps avoid the confusion.
I have adopted the names proposed by A. Zubkov in v34.
Some nits about the code.
I doubt if we need a test for these parameters - they reflect the
physical structure of the storage and might be unstable. But anyway,
it should be better to live in isolation tests, as similar statistics.
I moved the tests there. Regression tests are unfortunately not an
option because the statistics are not stable.
If the isolation test turns out to be unstable again, I'll move them
back to the TAP tests as I initially implemented,
following A. Borodin's suggestion.
See the version in the
https://www.postgresql.org/message-id/767d28c9-2ae8-43df-9f2e-3e8785075115%40yandex.ru
--
-----------
Best regards,
Alena Rybakina
