desktop/live cd netboot won't boot X

[Carl Karsten]

using https://wiki.ubuntu.com/LiveCDNetboot
I am somewhat able to boot the desktop live cd over pxe, but it never gets into
X or a prompt. but... if i boot into single, drop to a root propt, then run
startx, it boots into X, and then hangs.

This happens on 4 different boxes, including qemu, all of which run intrepid
just fine.

When I do it in qemu, after a min or 2 of blank display, it goes back to text
mode and I see:

m Starting bluetooth [ OK ]
Starting Network connection manager daemon: NetworkManager.
Starting Network connection manager daemon: NetworkManager
Starting GNOME Display Manager...
[ 675.386467] INFO: task loopO:2556 blocked for more than 120 seconds.
[ 675.387060] "echo O > /proc/sys/kernel/hung_task_timeout_secs" disables this
message.
[ 675.390675] INFO: task NetworkManager:6380 blocked for more than 120 seconds.
[ 675.39O798] "echo O > /proc/sys/kernel/hung_task_timeout_secs" disables this
message.
[ 675.393575] INFO: task gdm:6406 blocked for more than 120 seconds.
[ 675.393671] "echo O > /proc/sys/kernel/hung_task_timeout_secs" disables this
message.

The 6 lines then repeat every 120 seconds.

Steps to repo this problem:

cd /var/lib/tftpboot/ubuntu/intrepid/
wget -N http://cdimage.ubuntu.com/daily-live/current/intrepid-desktop-i386.iso

  mount
/var/lib/tftpboot/ubuntu/intrepid/intrepid-desktop-i386.iso on
/var/lib/tftpboot/ubuntu/intrepid/desktop type iso9660 (rw,loop=/dev/loop2)

# /etc/exports
/var/lib/tftpboot/ubuntu/intrepid/desktop
192.168.1.0/255.255.255.0(async,no_root_squash,no_subtree_check,ro)

# /var/lib/tftpboot/pxelinux.cfg/default
label ubuntu-intrepid-livecd
         kernel ubuntu/intrepid/desktop/casper/vmlinuz
         append initrd=ubuntu/intrepid/desktop/casper/initrd.gz boot=casper
netboot=nfs nfsroot=192.168.1.7:/var/lib/tftpboot/ubuntu/intrepid/desktop/
root=/dev/nfs single --

The details of my qemu setup: https://wiki.ubuntu.com/QemuPxe

You can see the behavior in qemu without having to setup the pxe environment by
just loading the casper kernel into qemu:

carl@dell29:~/temp$ sudo qemu -m 400 -net tap -net nic,model=pcnet hdb.qcow2
-kernel /mnt/nfs/shaz/intrepid/casper/vmlinuz \
-initrd /mnt/nfs/shaz/intrepid/casper/initrd.gz \
-append "boot=casper netboot=nfs
nfsroot=192.168.1.7:/var/lib/tftpboot/ubuntu/intrepid/desktop/ root=/dev/nfs
single -- "

and as a control, this works fine:
$ qemu hda.qcow2 -cdrom intrepid-desktop-i386.iso -boot d

[Colin Watson (cjwatson)]

I've reproduced this; indeed, from that point I can switch to a shell prompt but can't start any processes at all, so something is badly wrong. I have no idea what's causing it, though, and the aforementioned symptom makes it rather hard to investigate ...

[Marc D. (koshy)]

If you start a shell early on in the boot process and then use tools such as netstat, ip and ifconfig so that they are cached, you can analyse the situation after the system supposedly “hangs”. You will then see that eth0 is down, although eth0 is required for the nfs-mounted / to work.

I assume the problem is in NetworkManager: It starts, then shuts down eth0 so that it can later initialise eth0 by itself, but before doing so requires a read from /, which then hangs (indefinitely) because eth0 is down.

I think this problem was not present in intrepid alpha2 but appeared in intrepid alpha3.

[Thomas Hood (jdthood)]

@Marc: Is this still a problem?

[Launchpad Janitor (janitor)]

[Expired for network-manager (Ubuntu) because there has been no activity for 60 days.]

[Stephen Parry (sgparry)]

This bug is affecting me too. Not sure how this relates to this bug:
https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/1096943