Hi,
On February 10, 2022 8:37:10 PM PST, Soni M <diptatapa@gmail.com> wrote:
>Sorry, accidentally send the message before finishing it.
>
>Ah yes, it is the dead tuple cannot be removed, because long running
>transactions still see them.
>Something that lead to my confusion above is pg_stat_progress_vacuum, each
>time autovacuum/manual vacuum launch during long running transaction, it
>reports that num_dead_tuples is 0, while the
>pg_stat_user_tables.n_dead_tuple still have the number of dead tuples.
>
>It is false reporting from pg_stat_progress_vacuum that lead to my
>confusion.
That just reports the number of dead tuples that it is tracking for removal. Not the number of tuples that could not
yetbe removed.
--
Sent from my Android device with K-9 Mail. Please excuse my brevity.