2 Questions about rsync stats.

[David Bolen]

Karl Kopper [karl_kopper at hotmail.com] writes:

> Number of files: 74
> Number of files transferred: 5
> Total file size: 1104726340 bytes
> Total transferred file size: 115163136 bytes
> Literal data: 463500 bytes
> Matched data: 114699636 bytes
> File list size: 1570
> Total bytes written: 658318
> Total bytes read: 289890
> 
> wrote 658318 bytes  read 289890 bytes  12900.79 bytes/sec
> total size is 1104726340  speedup is 1165.07
> 
> What is the total transferred file size?

The total size of all the files actually transferred.  In the above,
there were 74 files included in the overall file list (due to your
selection of directories, wildcards, include/exclude lists, etc...),
of which 5 were actually transferred.  The 74 files totalled the
"Total file size" (1.1GB) and the 5 files needing to be transferred
(whether due to timestamp/size or checksum differences) the "Total
transferred file size" (115MB).

> Is the "Total bytes written" the amount of data (not counting overhead of 
> course) sent over the network (had this been a network transfer of data)?

Yes, and actually it includes any protocol overhead (of rsync, not of
any underlying network layer like TCP or other app overhead like SSH).
Total bytes written and read count each and every byte that rsync
writes or reads to/from the other endpoint during the course of
communication.  So it really means that in order to sync up 1.1GB
worth of potential data (of which files representing ~115MB needed to
be sent), your transfer would send 658318 bytes and read 289890 bytes
in total (from the rsync process) throughout the entire session.

Note that one thing I'm not absolutely positive on is whether or not
doing a local test like this will properly reflect file list
transmission in the bytes written and read since in theory rsync can
determine both source and destination directories through local system
calls and doesn't have to make client/server calls for that.  It looks
like it does (since that same counter is used to derive the size of
the file list which is non-zero in your example), but even if not, the
difference (unless you're into millions of files) ought to be fairly
small.

-- David

/-----------------------------------------------------------------------\
 \               David Bolen            \   E-mail: db3l at fitlinxx.com  /
  |             FitLinxx, Inc.            \  Phone: (203) 708-5192    |
 /  860 Canal Street, Stamford, CT  06902   \  Fax: (203) 316-5150     \
\-----------------------------------------------------------------------/