[rdiff-backup-users] include-filelist works to local partition but is ignored to remote target via ssh?

[Ståle Dagfinrud]

Hi

I'm having some problem when running rdiff-backup over ssh to a remote system.

In short the command works perfectly locally on to another partition, but 
fails to exclude files to a remote destination over ssh.

First a little about my setup:

* client system, a laptop (t61p)
Gentoo linux, 2.6.25-hardened kernel, amd64
rdiff-backup 1.2.1
source for files, /mnt/backups/source/
(lvm snapshot of root (/) partition)
contents of /mnt/backups/target/list-root.txt:
- source/usr/portage
- source/var/tmp
- source/var/cache
- source/tmp
- source/var/lib/vmware
EOF
all commands executed from /mnt/backups directory

* server system, file server (cerebrum)
Gentoo linux, 2.6.25-hardened kernel, i686
rdiff-backup 1.2.1
target for files, raid1 disk at /mnt/backups/t61p/root/
rdiff-backup is running as root via sudo/command directives in authorized keys

So I've been running rdiff-backup for quite some time now locally on the 
laptop, dumping the backups to a standalone lvm partition (the target dir on 
t61p)

# rdiff-backup --include-filelist target/list-root.txt --print-statistics 
source/ target/root

This works perfectly every time, and the excluded dirs are not backed up as 
expected.

Now last week I set up a remote backup target via ssh, assuming this would be 
sufficient to work.
(target "cerebrum" is an alias for a backup destination in .ssh/config)

# rdiff-backup --include-filelist 
target/list-root.txt --print-statistics --remote-schema 'ssh %s 
sudo /usr/bin/rdiff-backup --server' source/ cerebrum::/mnt/backups/t61p/root

rdiff-backup completes with 4 errors which is due to exceeded max path lengt 
for some sockets on running apps at the time the snapshot was taken, no 
problem really.
The big problem however is that all the dirs in the include file (with a 
preceeding - to exclude them from the backup as in the local version) are 
included in the backup.
Since these are pretty large directories with constantly changing temporary 
files I'd rather not have them backed up.

output from client running with -v 9:
<snip>
Sun Oct 12 13:00:17 2008  Client sending (0): 'Reading filelist 
target/list-root.txt'
Sun Oct 12 11:00:17 2008  Server received (0): ConnectionRequest: 
log.Log.log_to_file with 1 arguments
Sun Oct 12 11:00:17 2008  Server received (0): 'Reading filelist 
target/list-root.txt'
Sun Oct 12 11:00:17 2008  Server sending (0): None
Sun Oct 12 13:00:17 2008  Client received (0): None
Sun Oct 12 13:00:17 2008  Reading filelist target/list-root.txt
Sun Oct 12 13:00:17 2008  Client sending (0): ConnectionRequest: 
log.Log.log_to_file with 1 arguments
Sun Oct 12 13:00:17 2008  Client sending (0): 'Sorting filelist 
target/list-root.txt'
Sun Oct 12 11:00:17 2008  Server received (0): ConnectionRequest: 
log.Log.log_to_file with 1 arguments
Sun Oct 12 11:00:17 2008  Server received (0): 'Sorting filelist 
target/list-root.txt'
Sun Oct 12 11:00:17 2008  Server sending (0): None
Sun Oct 12 13:00:17 2008  Client received (0): None
Sun Oct 12 13:00:17 2008  Sorting filelist target/list-root.txt
</snip>
EOF

I speculated on wether the command directive in authorized_keys would need to 
be fixed somehow, but seeing as the server seems to be getting the include 
files I doubt this is the case...

So any ideas? What am I doing wrong, or is this a bug?

-- 

Best regards,

Ståle Dagfinrud
Hosting/Integrations Consultant
Fronter International

Web: http://support.fronter.com

signature.asc
Description: This is a digitally signed message part.