continusync issue

[Matt McCutchen]

Charles,

Are you preparing to use continusync for a real-world continuous
mirroring job?  I'm afraid this may be an exercise in frustration unless
you have a lot of time on your hands to fill in the gaps in its
implementation.  Instead, you might like to use Darryl Dixon's
continuous mirroring setup based on pyinotify, which I believe is
considerably more mature than continusync:

http://lists.samba.org/archive/rsync/2007-August/018329.html

Ask him for a copy of the code.

Regards,
Matt

On Fri, 2007-11-16 at 14:55 -0800, Charles Polisher wrote:
> Matt McCutchen wrote:
> > 1. Wait for a short time after a change to the source occurs before
> > propagating it.  If a file is created and then rapidly deleted, no
> > action is necessary.  Implementing this fully would require a data
> > structure to remember pending changes; it would have to be somewhat
> > sophisticated to handle moves optimally.
> 
> > 2. Somehow get rsync to treat disappearance of the source file as
> "file
> > has vanished" rather than an error.  This could be done inelegantly
> by
> > running rsync on the top source directory but excluding everything
> > except the changed file(s).  It might be better to use a modified
> rsync
> > with an option that causes it to treat any and all missing source
> > args/--files-from entries as "file has vanished".
> 
> > 3. Figure out why continusync is exiting altogether when rsync
> returns
> > error 23 and stop it from doing that.
> 
> Thanks for the fast response, Matt. I think we'll try to do #2 locally
> as a patch. I'll let you know how it works out.
> 
> Best regards,
> Charles
>