--VrqPEDrXMn8OVzN4
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Mon, Feb 07, 2005 at 09:11:44AM +1000, George Michaelson wrote:
> =20
> >
> >> one corrupt bit, and you risk loosing the entire held state.=20
> >
> >Only for BDB will one corrupt screw the repository, but that's sort of
> >like saying that one corrupt bit and you risk losing an entire RCS
> >file.  It's a non-point.
>=20
> Not at all. If you loose the RCS file for bin/ls/RCS/ls.c,v .. you have
> lost one .c source.
>=20
> If you loose the fsfs/db/<xx> file, you've lost the entire repos for that
> file. All of it.
>=20
> No?

I believe you are correct, in terms of the BDB backend (as I understand=20
this). If the DB goes down, everything in it goes.

However I'm not sure how much of a decisive concern that should be. The
property that an error can loose the whole DB is a property of many
databases, and is not a new concept. People have figured out how to backup=
=20
databases, and do it daily. It's different from how you backup a file=20
system (how we would back up the cvs repository), but it is doable.

I'm not saying we should switch, but that if we do, we will need to admin=
=20
an svn repository much more like a database than we do now. I expect we=20
can do it, it will just be different.

Take care,

Bill

--VrqPEDrXMn8OVzN4
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.3 (NetBSD)

iD8DBQFCCTKhWz+3JHUci9cRAullAJ472HbwqfHcAiQYyHgPoJTb6FbxEgCfUgUT
snj97GyohKMQRzc1+dPBRXg=
=fVkl
-----END PGP SIGNATURE-----

--VrqPEDrXMn8OVzN4--