On Thu, 2016-01-28 at 02:10 +0100, Michael Biebl wrote: > Am 28.01.2016 um 02:02 schrieb Grant McLean: > > Possibly, but there's no point calling update-rc.d (which calls > > insserv?) on a system where udev is disabled. The udev.postinst script > > already omits some things when udev is disabled. I'm suggesting that it > > could omit calling update-rc.d too. > > Well, there is. You might want to debootstrap a system and later deploy > the image to systems where udev actually needs to run. > How exactly did you "disable udev"?
Prior to this problem occurring, we ran a systemctl command to mask out the udev service. The command created symlinks to /dev/null from: /etc/systemd/system/udev.service /etc/systemd/system/systemd-udevd.service After looking through the udev.postinst script we learned that the script considers that udev is 'disabled' on the host if this file exists: /etc/udev/disabled This was the reason for my original suggested addition to the postinst script: [ -e /etc/udev/disabled ] && exit 0 I'm not sure if that file has any significance outside of the udev.postinst script. > If you run "insserv udev", what's the output? # insserv udev insserv: Service mountkernfs has to be enabled to start service udev insserv: exiting now! Which seems reasonable - the udev init script does declare a dependency on mountkernfs and the mountkernfs service is also masked in the container since that function is handled by the LXC startup. Regards Grant _______________________________________________ Pkg-systemd-maintainers mailing list [email protected] http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/pkg-systemd-maintainers
