I've set up a wiki page which also covers live-media-path:

  http://wiki.grml.org/doku.php?id=rescueboot

The current release notes at http://grml.org/changelogs/README-grml-2012.05-rc1/
already link to this document, I'll also add according notes to the previous
release notes as well as the upcoming ones (as reminder).

Thanks for your feedback,
-mika-

For the curious, the symptom looks like this

Begin: Running /scripts/live-premount ... 
Waiting for ethernet card(s) up... If this fails, maybe the ethernet card is not
supported by the kernel 3.1.0-3-grml-amd64?
Looking for a connected Ethernet interface ... eth0 ? eth1 ? eth2 ? eth3 ?
Connected eth0 found
done.
Using timeout of 60 seconds for network configuration.
Net: Turning on all device links...
If you want to boot from a specific device use bootoption ethdevice=...
Executing ipconfig -t 60 eth0
 IP-Config: eth0 hardware address de:ad:be:ef mtu 1500 DHCP RARP
 IP-Config: eth0 guessed broadcast address 1.2.3.255
 IP-Config: eth0 complete (from 1.2.3.10):
  address 1.2.3.12  broadcast 1.2.3.255 netmask 255.255.255.0
  ..etc..
Executing ipconfig -t 60 eth1
IP-Config: eth1 hardware address de:ad:be:ee mtu 1500 DHCP RARP
IP-Config: no response after 60 secs - giving up
  (repeats for eth2, eth3)
Creating /etc/resolv.conf
Begin: Trying netboot from 1.2.3.4:/path/to/iso/img/2011.12
Begin: Trying nfsmount -o nolock -o ro  1.2.3.4:/path/to/iso/img/2011.12
/live/image ...
done.

It then tries looking in /live/image for *.squashfs, *.ext2, *.ext3, *.ext4,
*.jffs2, *.dir and calls 'panic'.
  panic No supported filesystem images found at /live.

At this point the screen shows a debian swirl with the message
  "BOOT FAILED!"
  This image failed to boot

and several paragraphs of instructions.
A shell is started so you can go look at the live-image.log.

If you try with debug=1 boot option you get much more useful info in the
live-image.log, basically "set -x" output of what is happening behind the scenes.

The problem turned out to be that I did not have the 
 live-media-path=/live/${FLAVOUR}
boot option in the APPEND line of the PXE configuration file.

It was not needed before, I was able to boot 32-bit and 64-bit
flavours of 2011.05 without it.

This does not appear to be mentioned in the release notes,
so the only bug here is in the documentation.

Thanks for grml.

On Sat, Apr 07, 2012 at 01:41:03AM +1000, Ulrich Dangel wrote:
> 
> Ulrich Dangel <mru@grml.org> added the comment:
> 
> On 04/03/12 05:46, Vince wrote:
> 
> > I tried 2011.12 and it fails to mount the nfs share.
> > The boot line args are
> >   boot=live lang=us nomce quiet apm=power-off nodhcp noprompt noeject
> 
> As mentioned in the changelog (though quite hidden) you should not use 
> the nodhcp boot option as it does exactly what it describes, it prevents 
> the use of dhcp see http://grml.org/changelogs/README-grml-2011.12/
> 
> Just remove the parameter and it should work.
> 

Thanks for the pointer. I tried that and it did not work.
I'll try to explain what that means in a way you can understand
when I return to working on this issue.

On 04/03/12 05:46, Vince wrote:

> I tried 2011.12 and it fails to mount the nfs share.
> The boot line args are
>   boot=live lang=us nomce quiet apm=power-off nodhcp noprompt noeject

As mentioned in the changelog (though quite hidden) you should not use 
the nodhcp boot option as it does exactly what it describes, it prevents 
the use of dhcp see http://grml.org/changelogs/README-grml-2011.12/

Just remove the parameter and it should work.

cheers,
Ulrich

I have a working setup for 2011.05 amd64 and i386.
It does a pxe boot of the grml kernel and then mounts the iso over NFS.
We use dhcp with 'next-server' to tell the machine where to find the PXE server.

I tried 2011.12 and it fails to mount the nfs share.
The boot line args are
 boot=live lang=us nomce quiet apm=power-off nodhcp noprompt noeject

live-boot.log shows:
  Connected eth0 found
  Net: turning on all device links
  ignoring net device eth0 due to nodhcp
  Creating /etc/resolv.conf
  connect: Network is unreachable.
  connect: Network is unreachable.
  NFS over TCP not available from 11.22.33.44 (address of NFS server)

The connect messages go on for a while and then I get dropped into a shell.
At this point -
 - ifconfig eth0 does not show an ipv4 address on the interface
 - /etc/resolv.conf does not exist
 - df shows a /dev (fstype udev) and a /run (tnmpfs)

I am able to boot the machine into its usual OS and mount the NFS share
containing the 2011.12 iso image; it's not an issue there. It is something
to do with the initial networking setup.


On 2011.05 we get different behaviour using the same boot options.
There, it does this
 found eth0
 IP-Config: eth0 hardware address de:ad:be:ef MTU 1500 DHCP RARP
 IP-Config: eth0 guessed broadcast address 11.22.33.255
 IP-Config: eth0 complete from 11.22.33.55  (dhcp server addr)  
 Then it prints the dhcp config it has received and creates /etc/resolv.conf.

Then it mounts the nfs share and continues happily.

When I try to boot 2011.12 without the nodhcp option it fails,
possibly because it seems it is trying to boot into /pxelinux.0.

I also tried ethdevice-timeout=60 and there was a message indicating the
timeout had been changed but that had no obvious effect on the issue; the
failure occurred just as quickly as before.