Vinum, concatenated volumes and dead drives

[Robert Watson]

On Tue, 17 Feb 2004, Greg 'groggy' Lehey wrote:

> I've never tried it, but I suspect that fsck would have a hard time
> recreating the file system if any sizeable chunk is missing.  In
> general, I'd expect that you lose your file system under these
> circumstances. 

I've run into a number of situations where fsck is unable to recover from
catastrophic loss, such as wholesale stamping of zero's over the
alternative superblock and meta-data.  Under those circumstances, newfs is
a lot easier to use than fsck.  Our kernel implementation makes fairly
strong assumptions about the correctness of on-disk contents, and is not
resilient to some of the nastier failure modes associated with zeroing
(losing) large chunks of meta-data, and I've also found fsdb doesn't like
them much either (especially when you have a broken first alternate
superblock).

Robert N M Watson             FreeBSD Core Team, TrustedBSD Projects
robert at fledge.watson.org      Senior Research Scientist, McAfee Research