Rsync 3.0.0 report and a performance question.

[Paul Slootman]

On Wed 07 Feb 2007, Matt wrote:

> Using --whole-file doesn't help, see below.  BTW, rsync needs about the
> same time even if no file has changes at all.   It must be the
> comparison of the file metadata. A rate 325 files/sec seems somewhat low
> to me but maybe its mostly the time to read all directories and the
> "File list generation time" in the stats doesn't account for this. 

When you're using incremental file list generation and transfer,
accounting for the time specifically used for the generation only would
probably use up more cpu time than the generation itself...

I suggest that the values given for "File list generation time" and
"File list transfer time" should be given as "n/a" (not applicable) when
incremental file transfers are used.


Paul Slootman