some files already gzipped causes rsync to not compress any?

Steven Roberts strobert-rsync at
Thu Sep 11 04:40:22 EST 2003

On Wed, Sep 10, 2003 at 02:45:17AM -0700, jw schultz wrote:
> On Wed, Sep 10, 2003 at 02:21:32AM -0700, Steven Roberts wrote:
> > On Tue, Sep 09, 2003 at 11:58:57PM -0700, jw schultz wrote:
> > > On Tue, Sep 09, 2003 at 11:33:44PM -0700, Steven Roberts wrote:
> > > > Greetings,
> > > > We have encountered an odd performance issue in rsync (running version 2.4.6).
> > > If you want any support you'll need to get something from
> > > this century.  2.5.6 at least if not CVS.
> > this seems to imply that 2.5.6 even is barely supported.  If this is the case
> > then any ideas on when 2.5.7(or whatever) will come out?
> I don't know when 2.5.7 will be released.  That is more a
> matter of one of us dedicating the time to wrap it up than
> it is of the team deciding when.  There are one or two
> things i'd like to see resolved first that i've mentioned in
> the recent past.
> I think i can say that at this point only a security issue
> would cause us to produce an official update to 2.5.6.  So
> if there is a bug we won't be working on it in 2.5.6 but in
> CVS.  My feeling is that it is hardly worth producing a
> patch for someone who won't upgrade to at least the current
> release and isn't ready to build from source.
I am willing to build from source, and in fact I do actually build from
source.  I actually have a low priority bug I patch in the source code
that I will be submitting as a patch once I get some spare time.

>From a support perspective it is far more comforting to be going off an
official version+patches.  now sometimes we have based our "official version"
on a cvs date stamp, but not high on the list of what I would like to do
for a widespread deployment.

For you are installing on a hundred machines, you need some sort of
version tracking.

If you would like to discuss this further, we can probably take this off list.
> > I can understand not getting much support for older versions (and 2.4.6
> > is 3 years old).  But other OpenSource projects do support more legacy
> > versions, I was really only seeing to see if one of the developers/users
> > on the list knew of something in particular (and if it is fixed in 2.5.6
> > great, gives me good leverage to get the new version approved for production
> > use).
> The behaviour you report is not consistent with my quick
> perusal of HEAD.  Had it been i would probably have looked
> further and done more than just recommend an upgrade.

thank you.  that is what I was looking for.  A basic sanity check of yes
this doesn't sound like it would still be a problem in the current source
so try upgrading.  I don't have a problem do a test of an upgraded version
I just wanted an idea that the week to setup a test environment parralel
to production would be worth it.

I am sorry if that is what you meant by your initial comments.  I took them
as, oh, we aren't even going to bother thinking about this issue unless
you are maybe running the latest official release of preferably whatever
is on the bleeding edge (aka raw from CVS).


More information about the rsync mailing list