[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: /home replication on laptop

> Torbjorn Pettersson wrote:
> > Another way to do the home account sync is to put everything in
> > it in CVS, and have one central CVS repository to syncronize
> > with. It will also handle the problem where files get updated
> > from two different ways quite nicely.
> This works and is very elegant, but is *painfully* slow and uses a lot 
> more disk space on the server than is necessary (unless you really need 
> to be able to restore any previous state).

I believe the benefit that was being aimed for is to auto-merge changes
from both ends.  CVS represents a "neutral third party" to arbitrate the
diffs, as opposed to ending up with the extra file on each/either end, or 
squishing something out of hand.

Being able to restore to ages ago is a side effect.  Doesn't CVS have any
way to age out older changes that you don't care about anymore?  

Anyways, there are other Source Management systems too, surely any of them
could be twisted to this purpose.

> I used to do this, and switched to rsync, which is at least ten times 
> faster for an archive of any decent size.  (No, never timed them... :-)

Okay, to summarize, so far I've seen mentioned for this purpose:
 a) rsync (a lot!) http://rsync.samba.org

	best fragment appears to be..
rsync -e ssh -q --links --times --update --delete  -r --exclude 'no-rsync' ~/ mc@my-work.com:.

	..thanks MC!

	Also noted that you can tell rsync to make backups if it's conflicting 
	on a file.

 b) catchup. http://www.cse.unsw.edu.au/~neilb/oss/catchup/

 c) Coda. http://www.coda.cs.cmu.edu

	the package on their ftp server is for unstable.

 d) unison. http://www.cis.upenn.edu/~bcpierce/unison/

	noted, there's an unstable debian package :) and Uwe noted that
	the provided binary worked perfectly on his potato.

 e) CVS. http://www.cvshome.org/

I have to admit, I don't use any of these (well, for this purpose, anyway);
my workstation and my laptop only serve similar purposes, not the same.  So
my laptop carries a subset of my workstation environment, and work that I
do on my laptop is posted back up to the workstation when I'm back / linked
up again, via scp.

I have a long term plan to reconfigure my workstation environment so I have
a dummy user which represnts the contents of the laptop, so that I could 
sync up quickly (either via tar|ssh, or via rsync), but, at the moment,
the laptop's contents are organized differently than the workstation and
exact syncing is not reasonable.

* Heather Stern * star@ many places...

Reply to: