On May 22, 2010, at 12:16 AM, Marc MERLIN wrote:
>> Why sync files that aren't useful in the first place?
> That's the point, we don't sync __db* files but some local debugging run
> on a machine might create them and then they would be a mismatch with an
> rpmdb rsync on top.
> Deleting them in this case is the best way to make sure we don't have a
Well the approach I took was to erase __db* files iff an error was encountered
and retry an rpmdb open exactly once. That patch was done like 4+ years ago, you might
almost be ready for that version of rpm ;-)
The change wasn't a lot of code, just tricky to test because
of low incidence on escalating error handling paths.
73 de Jeff
Received on Sat May 22 06:26:00 2010