The Dirvish/Windows rsync+ssh problem is still mostly unsolved.
There are no great solutions but there are some OK workarounds,
and some things I would like to see some Windows users try (I
don't run Windows natively, so I can't).

The problem is that rsync + ssh together on a windows machine locks
up.  rsync+rsh does not, but that is insecure.  Setting up a Windows
machine to do insecure rsync+rsh through an external machine requires
that the Windows machine is never directly exposed to the internet.
What to do?

1)  Steve Ramage has a good writeup on the dirvish website.  I don't
know if his methods work with the latest versions of the tools, but
give them a try, and contribute fixes or clarifications if needed.

2)  About a year ago,  Tevfik Karagüll was talking about a product,
hardBackup, that does dirvish style windows backups.

3)  The rsync+ssh problem needs study and fixing.  A solution might
happen quicker if someone learns about rsync and ssh protocol, and
uses traces and sniffers to find out exactly what one of these
lockups looks like.  That would help a programmer a lot.

4)  VMware might offer an alternative.  A VMware Linux guest running
on the Windows host in bridging mode might have a secure ssh port
facing the outside world, and communicate on a host-only network
"inside" the Windows machine to the Windows host.  Windows rsync 
could talk to the Linux guest unencrypted, or else the Linux guest
could run rsync against the windows files mounted with Samba. 

5)  Alternately, the problem could be turned inside out, and the
Windows functionality could be a VMware windows guest on a linux
host.  That is how I run Windows on the rare occasions I need to; 
VMware has limitations (and is proprietary :-(, bleh ) but it is
zero-cost and fairly easy to work with.

The larger problem is restore - however you get the files out
of a Windows system, putting them back can be troublesome, and
a bare metal restore is probably out of the question.

However this works out, it is up to our Windows users to find 
better solutions.  Steve, Terfik, Eric, Richard, and others have
made some good contributions, but this is open source, and it
works best if everyone contributes what skills they have.  The
contribution can be as simple as running tests and accurately
describing the failures, or writing up your experiences and
posting them on the wiki.  Open source is not "zero-cost"
software;  you pay with your participation.

Keith

-- 
Keith Lofstrom          [EMAIL PROTECTED]         Voice (503)-520-1993
KLIC --- Keith Lofstrom Integrated Circuits --- "Your Ideas in Silicon"
Design Contracting in Bipolar and CMOS - Analog, Digital, and Scan ICs
_______________________________________________
Dirvish mailing list
[email protected]
http://www.dirvish.org/mailman/listinfo/dirvish

Reply via email to