Re: sequence number mismatch

[Michael Stevens <mstevens@etla.org>]

On Thu, Jun 02, 2011 at 07:00:22PM -0700, Colin Percival wrote:
> On 06/02/11 14:42, Michael Stevens wrote:
> > On Thu, Jun 02, 2011 at 01:19:40PM -0700, Colin Percival wrote:
> >> On 06/02/11 13:09, Daniel Staal wrote:
> >>> One cache directory is assumed to be one box, one box is assumed to be one
> >>> key.
> >>
> >> Sort of.  Really it's just that one cache directory is assumed to be used
> >> for each key.
> >>
> > Okay, so I created a separate cache directory, fscked it, and tried the
> > delete again... and it finally works!
> > 
> > Feature request: make the caches more droolproof!
> 
> Good idea.  When I was writing this code I didn't think anyone would ever
> try to use the same cache directory for two different keys.

Possibly it's a sign of not reading enough of the manual, but I didn't
realise the cache directories were key dependent.

What's meant to happen in the "oh dear, my PC blew up, need to restore"
case?

It seems like it'd be natural to try to use the key on a machine with an
existing tarsnap install to retrieve the files, only that's actually
going to corrupt your cache.

Michael