[BBLISA] disk corruption recovery ideas?

[Douglas Alan]

Eric Smith <esmithphoto at gmail.com> wrote:

> Using lvm and reiserfs I gain a lot of flexibility with disk layout.
> I know others who have done this and its basically a no-maintenance
> solution for them (leveraging SMART drive monitoring to warn of
> impending disk failure.)

When I used reiserfs, I suffered a lot of file corruption on unclean
shutdowns.  I.e., every time there was a power failure, or my computer
wedged for some reason, my Gnome config files would become corrupt and
I'd have to reconfigure Gnome from scratch.  I don't know if this
problem is still the case with more recent versions of reiserfs (I think
it is because worrying about file corruption was apparently not one of
the design goals of reiserfs), but this design tradeoff was apparently
known to be "feature" of reiserfs at the time.  Since then, I'd be
rather wary of using reiserfs on anything other than a filesystem used
solely for Netnews or MH, which seems to be its target application
domain.  (I.e., lots and lots of small files.)  I can probably cope with
a few corrupted mail files, but even that would be a bit annoying, if an
important mail file were trashed.

After switching back to ext3fs, however, I noticed no performance
problems with huge MH directories containing 100,000+ files, so I think
I'd stick with ext3fs for even MH.  Maybe I'd use reiserfs for News.

[Tangentially related factette: I can't use HFS on OS X with MH because
the performance is terrible when mail folders get too large.  MH using
folders on a UFS filesystem on OS X, though, performs fine.]

> The flaw I see in doing this is that I believe I either need tape
> backup or another equivalent setup (physical second RAID5) to rsync to
> to prevent bit corruption (not hardware failure.)

I've had two drives die at the same time on a RAID.  Good thing the
RAID was only used as our backup server.  I'd never trust RAID again to
be any kind of security against disk failure.

|>oug