Problems with 0.10.0 on Windows

Ben Escoto bescoto@stanford.edu
Tue, 17 Sep 2002 09:45:12 -0700


--==_Exmh_-491270502P
Content-Type: text/plain; charset=us-ascii

>>>>> "PE" == paul-erik torronen <iso-8859-1>
>>>>> wrote the following on Tue, 17 Sep 2002 12:20:40 +0300

  PE> Background: The backup-server is a Windows 2000 Server with
  PE> latest updates as well as the newest Cygwin tools (as of
  PE> 20020913). We have compiled the librsync 0.9.5 as well as
  PE> rdiff-backup under Cygwin.

Do you mean 0.9.5.1?  I think 0.9.5 will leak a lot of memory on you,
but it should be OK if you haven't noticed that.

    ...
  PE> Question: Is it possible to either at compilation-time configure
  PE> rdiff-backup to use sockets (I assume that fifo is used to
  PE> communicate between threads) instead of fifo (perhaps combined
  PE> as a windows-target option which also excludes the mknod and
  PE> hardlinks), or at runtime with perhaps --use-sockets, to
  PE> circumvent the shortcomings of the OS?

Actually in this case it looks like the fifo isn't used by
rdiff-backup, but is created just because a fifo is detected in the
source directory.  If you don't try to back up any fifos, rdiff-backup
will not mkfifo().  There is no --exclude-fifos option, but it
wouldn't be hard to add.  If you wanted to check to see if this is the
problem, try, for instance, applying this patch:

--- selection.py        Fri Aug  9 17:43:04 2002
+++ selection.py.new    Tue Sep 17 09:34:41 2002
@@ -452,7 +452,7 @@
                        Log("Warning: exclude-device-files is not the first "
                                "selector.\nThis may not be what you intended", 3)
                def sel_func(dsrp):
-                       if dsrp.isdev(): return 0
+                       if dsrp.isdev() or dsrp.isfifo(): return 0
                        else: return None
                sel_func.exclude = 1
                sel_func.name = "Exclude device files"

Then when you use --exclude-device-files it will also exclude fifos.
Hopefully that will solve your "Function not implemented" error.

    However, I am more concerned with these messages:

  PE> Warning: Found unexpected
  PE> destination file
  PE> /cygdrive/e/host.domain.net/lib/modules/2.4.18-3/kernel/net/ipv4/netfilter/i
  PE> pt_mark.o, not processing it.  Warning: Found unexpected
  PE> destination file
  PE> /cygdrive/e/host.domain.net/lib/modules/2.4.18-3/kernel/net/ipv4/netfilter/i
  PE> pt_tcpmss.o, not processing it.  Warning: Found unexpected
  PE> destination file
    ...

These warnings are supposed to happen when a file is created while
rdiff-backup is running, but given the number of these messages that
probably isn't happening here.

    Just had a thought: is it possible that your destination file
system isn't case-sensitive?  Then things are bound to get screwed
up - suppose one directory on the source side has the files "File" and
"file".  Then they can't both be stored on the destination side.  Some
other poster reported that "unexpected .. file" message, but I didn't
think of the case sensitivity issue at the time.


-- 
Ben Escoto

--==_Exmh_-491270502P
Content-Type: application/pgp-signature

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.7 (GNU/Linux)
Comment: Exmh version 2.5 01/15/2001

iD8DBQE9h1wV+owuOvknOnURAtHrAKCGgbdx640zzi+HEs/8wUCVAZYBCACfa1Wl
4j8voab5j8DGjqIjlrsREwk=
=TFi5
-----END PGP SIGNATURE-----

--==_Exmh_-491270502P--