[lxc-devel] [PATCH 2/2] templates: use hardlink detection in rsync

Serge Hallyn serge.hallyn at canonical.com
Tue Dec 4 17:33:55 UTC 2012


Quoting Dwight Engen (dwight.engen at oracle.com):
> On Mon, 3 Dec 2012 21:25:00 -0600
> Serge Hallyn <serge.hallyn at canonical.com> wrote:
> 
> > Quoting Michael H. Warfield (mhw at WittsEnd.com):
> > > On Mon, 2012-12-03 at 20:43 -0600, Serge Hallyn wrote:
> > > > A-ha, actually busybox cp does not support -T, at least here on
> > > > ubuntu 12.10.  So should we go with rsync -ah after all since
> > > > it's more likely they'll install rsync next to busybox?  Or is
> > > > that not actually likely?
> > > 
> > > rsync -aH
> > 
> > Heh, yeah, my patch did have it right in my defense :)
> > 
> > > Capitalization is critical here.
> > > 
> > > Option -x (do not cross devices) is not the same as -X (extendended
> > > attributes).
> > > 
> > > > -serge
> > > 
> > > I would like to know where rsync is not available.  If rsync is
> > 
> > In an ultra-minimal busybox based system?
> > 
> > not sure...
> > 
> > > available then "rsync -H" is available.
> > > 
> > > That is my concern with busybox.  They are very POSIX compliant and
> > > I'm not sure this is a POSIX feature.  I run into this all the time
> > > in Android and TinyCore (I've had to recode things to run under
> > > busybox ash shell under TinyCore).  I trust rsync better than cp in
> > > differing environments for that very reason.  I know what rsync
> > > will support (if it's there).  I don't always know what options cp
> > > will support.
> > 
> > Ok, so let's stick with my original 2 patches (top of this thread).
> > 
> > Objections?  (Acks?)
> > 
> > -serge
> 
> Ack: Dwight Engen <dwight.engen at oracle.com>
> 
> Want me to add rsync as a run time dependency in the lxc.spec.in
> file?

Thanks, that would be great.

(two patches as is, with acks added, pushed to staging)

-serge




More information about the lxc-devel mailing list