[Dirvish] checksum=0 may be causing inaccurate backups!!!
Barton C Massey
bart at cs.pdx.edu
Thu Jun 21 17:52:19 UTC 2007
In message <20070621133203.GA1426 at gate.kl-ic.com> you wrote:
> On Thu, Jun 21, 2007 at 03:17:21AM -0700, Barton C Massey wrote:
> > I can think of no reason why checksumming should add 6x to
> > the backup time. Am I missing something?
> > Bart
> Because dirvish/rsync normally reads only metadata off the disk,
> whereas checksumming reads every byte. The 6X worst case is
> when both the source and backup drives are on the same machine.
> It's all read time, at 1.2GB/minute in some cases, and that can
> take a LONG time.
> BTW, I woke to nightmares about this, see the major warning in
> the next message before doing "checksum=1=".
It seems like it can be heavily optimized, is all I was
thinking. If a new file is created with the same metadata
as some old file, you just need to checksum the old and new
files to see if they're the same? In other words, the ctime
should save you here?
More information about the Dirvish