On Tue, 2004-12-14 at 18:43, Dann Corbit wrote:
> Would it be possible to rebuild all non-btree indexes when a recovery
> takes place?
Considering how long I've seen some large hash indexes take to build,
that might be better left as an optional setting.
> Another thing that seems it might be nice is to check the non-btree
> indexes during analyze (if that is possible and not too expensive).
>
> -----Original Message-----
> From: Neil Conway [mailto:neilc@samurai.com]
> Sent: Tuesday, December 14, 2004 4:39 PM
> To: Scott Marlowe
> Cc: Dann Corbit; pgsql-general
> Subject: Re: [GENERAL] Corrupt RTREE index
>
> On Tue, 2004-12-14 at 14:12 -0600, Scott Marlowe wrote:
> > IS this same issue true for hash or GiST indexes?
>
> Yes, it is: currently, only btree indexes are WAL safe.
>
> (I spent some time recently looking into adding page-level concurrency
> and WAL to GiST, but I haven't had a chance to finish that work -- it is
> quite a big job...)
>
> -Neil
>
>