Hello! We solved it. The VACUUM full finished fast on pg_largeobject, because we deleted 98% of big largeobject (ours) before. And it worked as zip deletion - it created a new file and copied only living records, which was fast (3 GB vs. 80 GB).
Thanks dd Luca Ferrari <[email protected]> ezt írta (időpont: 2019. okt. 17., Cs, 17:43): > On Thu, Oct 17, 2019 at 5:10 PM Durumdara <[email protected]> wrote: > > Please help me, how the PGSQL full vacuum works internally? (1., 2. > case, or something else) > > > > As far as I know a VACUUM FULL will rewrite the whole table without > inserting (and therefore deleting later) not visible tuples. I would > be quite surprised of it duplicating the table and removing after the > tuples. > Surely it is a very invasive command that locks the table and requires I/O. > > > How we (and the clients) prepare to this operation? > > We must know it to avoid disk out problems, and too much off-time. > > The best guess is that you are going to need almost the double of the > table size. Since you said that autovacuum is preventing the table to > grow, it could mean all the reclaimed space has been consumed by other > tuples, so I'm not sure vacuum full can provide you space. > Have you looked at pg_stat_user_tables to see the dead tuples fraction? > > Luca >
