Tom Lane wrote:
> Neil Conway <nconway@klamath.dyndns.org> writes:
> > Also, could we enhance REINDEX to not require an exclusive table lock?
> > Without looking at the code, I would think that we could at least allow
> > for SELECTs (using seqscans or other indexes) while a REINDEX is running.
>
> Not unless you see a way to prevent those SELECTs from trying to use the
> index...
Well, we should recreate the index as a separate file, then switch over
to the new file use relfilenode.
> Personally I think putting development effort into REINDEX is a waste of
> manpower. Study how to make btree collapse out empty pages, instead.
Agreed, but what plans do we have to fix it? I have heard no proposals,
and it has been a known problem for years.
--
Bruce Momjian | http://candle.pha.pa.us
pgman@candle.pha.pa.us | (610) 853-3000
+ If your life is a hard drive, | 830 Blythe Avenue
+ Christ can be your backup. | Drexel Hill, Pennsylvania 19026