On Fri, 2006-01-20 at 14:22, Dan Pritts wrote:
> > The number of hardlinks makes rsync (or other file-oriented) copies
> > impractical if the archive is large at all,
>
> It occurs to me that you could, rather than rsyncing at the filesystem,
> rsync the raw device that holds the filesystem. That would not
> require an inordinate amount of RAM.
>
> You'd of course need to use a snapshotting technique, or unmount the
> filesystems, first. It also would be prudent to fill the empty space
> in the filesystem with nulls the first time you transferred it (and use
> some method other than rsync to do the transfer - or send a dump image,
> or something). Or, I suppose the first time you could just mail a
> hard drive.
>
> I've done this with oracle database dump files, and it worked fine.
> These are binary data structures just like filesystems are.
I think the same idea occurred to me a while back but rsync
refused to use a device directly. So, you'd need to snapshot
the partition to a local file first, then rsync to a remote
file. That would take extra disk space but then make the
transfer time less critical. You'd probably want to fill
all the empty space with nulls every time too, so you don't
transfer all the changes where files have been deleted.
> > plus you'd end up
> > not being able to access the copy unless you have configured an
> > instance of backuppc that understands the layout.
>
> This strikes me as minimal effort, just rsync the configuration files
> too.
Yes, but it is nicer to have a single instance that can browse
through all files at once.
> > What I ended
> > up doing is a scheduled rsync at the remote offices to a linux
> > box there, then the central backuppc server archives that copy
> > over a WAN or VPN connection. This wastes some disk space since
> > the intermediate copy is not compressed and only keeps one copy
> > locally, but everything else works out pretty well.
>
> Does it not also waste WAN bandwidth? Doesn't backuppc's pooling stuff
> happen at the backuppc server, so you'd end up copying all the OS files
> for each machine over the WAN?
Actually it saved bandwidth because I could use compression on
the ssh connection with the linux box but couldn't have directly
with the windows machines. I'm only backing up data files from
servers and a few other machines, not whole workstations so
there is not a lot of duplication. Even if there were, it would
only matter on the initial copy since rsync wouldn't transfer
the data again.
--
Les Mikesell
[EMAIL PROTECTED]
-------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc. Do you grep through log files
for problems? Stop! Download the new AJAX search engine that makes
searching your log files as easy as surfing the web. DOWNLOAD SPLUNK!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=103432&bid=230486&dat=121642
_______________________________________________
BackupPC-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/backuppc-users
http://backuppc.sourceforge.net/
