On Thu, Jun 12, 2025 at 10:28 PM Fujii Masao <masao.fu...@oss.nttdata.com> wrote: > On 2025/06/07 0:13, Robert Treat wrote: > > On Fri, Jun 6, 2025 at 9:57 AM David G. Johnston > > <david.g.johns...@gmail.com> wrote: > >> On Friday, June 6, 2025, Fujii Masao <masao.fu...@oss.nttdata.com> wrote: > >>> > >>> Hi, > >>> > >>> Since last_vacuum and vacuum_count in pg_stat_all_tables explicitly > >>> mention > >>> that they don't include VACUUM FULL ("not counting VACUUM FULL"), I think > >>> we should add the same clarification to the description of > >>> total_vacuum_time. > >>> This field also excludes VACUUM FULL, and without this note, users might > >>> mistakenly think the time spent on VACUUM FULL is included. Thought? > >>> > >>> <structfield>total_vacuum_time</structfield> <type>double > >>> precision</type> > >>> </para> > >>> <para> > >>> - Total time this table has been manually vacuumed, in milliseconds. > >>> + Total time this table has been manually vacuumed, in milliseconds > >>> + (not counting <command>VACUUM FULL</command>). > >>> (This includes the time spent sleeping due to cost-based delays.) > >>> </para></entry> > >>> </row> > >> > >> > >> Makes sense. Our naming this table rewrite vacuum full does confuse > >> people into thinking it is related to vacuum. > >> > > > > +1 for this change, > > Thanks both for the review! > > > > but I think we should also update > > n_ins_since_vacuum as well, no? > > I didn't update n_ins_since_vacuum since it's mainly used by autovacuum rather > than end users, and there haven't been any complaints about the current > description so far. That said, I don't have a strong opinion either way, > so I'm fine with making the change if others think it's worthwhile. >
Well, I admit I mostly mentioned it because when I noticed this one wasn't documented the same way the other ones were, I second-guessed myself about if I knew how it really behaved and did a quick test to confirm :-) I suspect others might have similar confusion. Robert Treat https://xzilla.net