[Box Backup] sillyrestore.bash

[Martin Ebourne]

On Tue, 2005-03-01 at 11:08 +0000, Ben Summers wrote:
> This seems to be a frequent problem. Are people still adamantly against 
> my proposed feature where if the root of the backup location is an 
> empty directory, then the location is skipped without making any 
> changes to the store? This would guard against this kind of problem.

I'm not against it. In fact, sounds like a good idea to me (with
suitable log file warning). Which reminds me, there's an oddity with
removing backup locations from the store:

bbackupd[4675]: 2 redundant locations in root directory found, will
delete from store after 172800 seconds.

Well that's 48 hours, but it's been a week so far and they're still
there. The problem seems to be that the timeout is only while bbackupd
is running, and thus gets reset each time it is restarted. Well this
machine is a laptop, so that scheme is not working out for me at the
moment. Maybe it could make a record in the local db files and timeout
whether running continuously or not?

> Didn't "restore -d" do what you want?
> 
> There is also another undocumented feature which may help in this 
> situation. Use the -w flag on bbackupquery, and then use the undelete 
> command to "undelete" the directory. This will stop it having to be 
> uploaded again later, but you may find files you deleted appearing 
> again as it doesn't distinguish between files deleted now or months in 
> the past.

I think the best solution to this is to be able to set an arbitrary
date/time in bbackupquery and the store then gives you access to an
exact snapshot of how it was at that time. I think you've mentioned
before that's on your todo list, but thought I'd make sure. It would be
so simple to use and massively powerful.

Cheers,

Martin.