Re: improve DEBUG1 logging of parallel workers for CREATE INDEX?

Le jeu. 9 janv. 2025 à 04:24, Sami Imseih <samimseih@gmail.com> a écrit :

> > Of course, a patch for that would be a few orders of magnitude
> > larger than what you've got here :-(.  But if you're looking
> > for a framework for reporting these sorts of details, I'd
> > much rather go in that direction than follow the model of
> > VACUUM VERBOSE.  VACUUM VERBOSE is a kluge with little to
> > recommend it other than having been easy to implement.
>
> To my surprise, REINDEX does have a VERBOSE option.
> should have check this earlier :)
>
> postgres=# reindex (verbose) index t_idx1;
> INFO:  index "t_idx1" was reindexed
> DETAIL:  CPU: user: 5.33 s, system: 0.48 s, elapsed: 6.26 s
> REINDEX
>
> Is there a reason not to do the same for CREATE INDEX?
>
>
Sounds a good idea to me.


> Also, we can improve the REINDEX verbose message by
> also providing the parallel usage.
>
>
+1


-- 
Guillaume.