On Tue, 2015-02-03 at 10:45 -0600, Corey Wright wrote: > > [...] > > sorry for the delay, but i had to work out the initrd error about no > permission to write to /sys/kernel/uevent_helper (ie wheezy's udev still > requires CONFIG_UEVENT_HELPER=y, though turned-off in debian's 3.18 config in > experimental). My hosts also failed booting because of this bug. The wheezy-backports version fixes the script in the initramfs (debian bug #752742). udev does not need uevent_helper anymore, but the script in the initramfs unconditionally write to that file and dies (because of set -e). Unfortunately the wheezy-backports version does not fix a similar bug in the init script for udev (debian bug 756312). But this is not so critical, the server at least can boot with this issue. > corey Alex > -- > undefined@pobox.com > > > that patch moved the vshelper array entry out of the CONFIG_COMPACTION > > preprocessor conditional but not out of the vm_table array and into > > kern_table array (to it appears under /sys/kernel and not /sys/vm). > > > > new patch coming up (unless herbert beats me to it). > > > > corey > > -- > > undefined@pobox.com > > > > > thanks to AlexanderS on irc for his comment that tipped me off to where the > > > problem was (which was probably a bit faster than grepping for vshelper). > > > > > > haven't run-time tested it (been busy debugging why lxc-start mounts > > > filesystems in a container on a non-vserver kernel but not on a vserver > > > kernel, though root can do it without problem within the container), but it > > > compiles. > > > > > > corey > > > -- > > > undefined@pobox.com > > > > > > > There is I good chance that I missed something during setting up the > > > > test system, any ideas? If that matters - I used 3.18.5 instead of > > > > 3.18.4 as base. > > > > > > > > Oli > > > > -- > > > > Protect your environment - close windows and adopt a penguin! > > >