[Box Backup-dev] Background backup: Sync of directory listings

Chris Wilson boxbackup-dev at fluffy.co.uk
Tue Jan 30 09:16:26 GMT 2007


Hi Gary,

> BackupClientDirectoryRecord::Serialize()

OK, I managed to miss that one <grin>, thanks :-)

>> a sync. When the laptop is resumed, the SSL connection will have 
>> failed, so bbackupd will end the sync run with an exception, and delete 
>> the StoreObjectInfoFile.
>
> Doesn't bbackupd shut down the SSL connection in between scan cycles?

Yes, it does. But I think it's being interrupted in the middle of a 
cycle.

> In that scenario, deleting the cache file is a mistake, as this is not a 
> bbackupd state-altering error condition.

But if the backup cycle is interrupted by an exception, then we should 
delete it, shouldn't we? Because in general we don't know what state the 
store is in (whether an operation in progress completed or not).

> As a side note, the cache file 
> should probably never be deleted, but rather upgraded to an ACID storage 
> system, with a commit/rollback capability to match the last logical 
> backup state before bbackupd goes down.

We don't necessarily know the whole state, but we can probably only 
invalidate a small part of it (the current directory) rather than 
everything.

> I started working on this already way back using SQLite, but held off 
> due to Ben's mention of an upcoming massive rework of the internal state 
> representation.

I don't know about that, it was probably before I joined the group. 
Directory handling will certainly change a lot in 0.20, but right now I 
don't know whether 0.20 is 6 months or 6 years away :-)

Cheers, Chris.
-- 
_____ __     _
\  __/ / ,__(_)_  | Chris Wilson <0000 at qwirx.com> - Cambs UK |
/ (_/ ,\/ _/ /_ \ | Security/C/C++/Java/Perl/SQL/HTML Developer |
\ _/_/_/_//_/___/ | We are GNU-free your mind-and your software |



More information about the Boxbackup-dev mailing list