On Fri, May 31, 2013 at 11:00:19AM -0700, Josh Berkus wrote:
> Bruce,
>
> > Roberts statement was:
> >
> >> Loss or corruption of a single visibility map page means possible loss
> >> of half a gigabyte of data.
>
> I fail to be alarmed at this; currently losing a single page of the clog
> causes just as widespread corruption (worse, actually, since it's not
> confined to one table). It does point to the eventual need to checksum
> these things, though.
>
> > Certainly unidentified corruption of a visibility map page could easily
> > cause incorrect results. So, technically, _adding_ bits would cause
> > corruption.
>
> Yes, that's already true. I'm pointing out that if we depend on the
> vismap for all-frozen, then losing bits *also* causes corruption, so
> that's something we need to test for. Right now, there is no possible
> corruption from losing bits; we simply end up scannning more pages than
> we have to.
Right, and it is hard to see that losing and adding are somehow
more/less likely, so it seems we already realy on the visibility map
being correct.
-- Bruce Momjian <bruce@momjian.us> http://momjian.us EnterpriseDB
http://enterprisedb.com
+ It's impossible for everything to be true. +
