[lug] Commercial rsync service?

[Rob Nagler]

Siegfried Heintze writes:
> Could you elaborate a bit on XP? (Or is this the wrong forum?)
> Yes XP has DAV built in. But what is the relationship between between DAV
> and rsync? I thought rsync used ssh. Can rsync use DAV too?

WebDAV is simply GET, PUT, DELETE, and PROPFIND (ls).  The RFC authors
originally thought that apps would be talking directly to WebDAV.
It's much more convenient, however, to simply bury the complexity of
WebDAV underneath the file system.  Think of WebDAV as another form of
NFS.

With davfs (Unix driver), you can mount a WebDAV "volume" just like
NFS or loopback or a CD.  Once that happens, you can then do:

    rsync -alCS --delete $HOME /mnt/some-dav-volume

The -a is similar to cp's -a.  -CSl are useful for backups, but may
not be appropriate for the purpose of making an exact image.  Read the
manpage.  rsync will make a mirror of your $HOME on /mnt/some-dav-volume.

Now back to XP (briefly).  MS in their infinite wisdom thinks that
"Web Folders" are "unsafe" or something.  A Web Folder doesn't
actually exist with a volume letter or a UNC name.  You can't even
double-click to open (or open with) most files, e.g. *.csv files.
And even if you double click, say, on a word doc, MS Word decides that
it has to reconnect to web folder so you have to reauthenticate inside
Word.  It's absolutely horrible.

Novell has a driver called netdrive that maps a WebDAV volume to an
actual volume with a drive letter.  The problem with Novell's software
is that it makes WAY too many assumptions about caching to the point
that sometimes you have to reboot to clear the cache.

And while we're at it, Mac OS has a nice interface to WebDAV, but it
too makes horrible caching assumptions.  For example, it assumes that
if you do a PROPFIND (readdir) followed by a GET that the value
returned by GET has to be the same size as what PROPFIND returns or
there's some type of error.

WebDAV is built on top of HTTP, which is a stateless protocol.
Moreover, WebDAV implements a distributed file system.  Therefore, you
can't assume anything from one request to the next.   Surprisingly,
Web Folders does quite a good job at this.  Unfortunately, there are
other bugs in Web Folders, such as, assuming PROPFIND returns an
alphabetically ordered list.

The reason I bring up WebDAV is that it's a standard Apache module
(mod_dav) that ISPs are starting to use.  It doesn't require a Unix
login or a special server like rsync does.  Therefore, you may find it
an easier to find a WebDAV service with 10GB more easily than one with
rsync.  That's just a guess on my part, however.

Rob