Package: nbd-client Version: 1:2.8.7-1 Severity: normal In postinst, nbd-client offers to set up some nbd connections. As far as I can tell this postinst script is problematic: - If udev is not active, the device is created with somewhat default permissions but they should be root:disk 660 (you might want to refer to the ctte decision in #316883 regarding lvm2 and the situation is the same here). - If udev is active the device node is either not created if it's already there, or it _is_ created but lost in next reboot.
My suggestion is to use /sbin/MAKEDEV instead of mknod which will take
care of all that. This will however restrict the device names to
/dev/nbd ... *eeek* Now I see that udev and static-dev (i.e. MAKEDEV)
disagree on the name of the network block device. MAKEDEV wants
/dev/nb[0-7] which is also suggested by the kernel sources while udev
uses /dev/nbd[0-7]. And I find /dev/nd[0-7] in the wild, too. Weird.
OK, put this one on hold while this naming confusion has to be sorted
out.
Christoph
PS: Thanks a lot for your report regaring nbd and kernel 2.6.18 in
http://www.grep.be/blog/2006/10/17/#linux-2.6.18
Without it I'd have lost hours in debugging the nbd problems introduced
by that.
-- System Information:
Debian Release: testing/unstable
APT prefers testing
APT policy: (500, 'testing')
Architecture: i386 (i686)
Shell: /bin/sh linked to /bin/bash
Kernel: Linux 2.6.18.1
Locale: [EMAIL PROTECTED], [EMAIL PROTECTED] (charmap=UTF-8)
Versions of packages nbd-client depends on:
ii debconf [debconf-2.0] 1.5.6 Debian configuration management sy
ii libc6 2.3.6.ds1-4 GNU C Library: Shared libraries
nbd-client recommends no packages.
-- debconf information excluded
signature.asc
Description: Digital signature
