Re: pgsql: rm_cleanup functions need to be allowed to write WAL entries.

Simon Riggs <simon@2ndQuadrant.com> writes:
> On Sun, 2009-08-09 at 11:41 -0400, Tom Lane wrote:
>> Er, what's the point of that?

> Rebuilding damaged indexes automatically, rather than barfing. I regard
> that as a long term extension of crash recovery to bring a database back
> to a usable state.

Having crash recovery auto-rebuild indexes it thinks are damaged seems
to me to be a pretty horrid idea.  Just for starters, it would overwrite
forensic evidence about the cause of the damage.  A DBA might not wish
the rebuild to happen *right then* in any case.

            regards, tom lane