client hangs after eth0 becomes ready

[Thomas Lange]

>>>>> On Wed, 16 Nov 2011 14:00:45 +0100, Carsten Aulbert <carsten at welcomes-you.com> said:

    > Hi all
    > On Wednesday 16 November 2011 08:21:26 Carsten Aulbert wrote:
    >> Specifying eth0 will not help as the kernel "arbitrarily" changes the names
    >> upon reboot. The machine I'm currently struggling with is choosing eth0 for
    >> the first on-board NIC for ~50-60% of the time (no hard numbers on that
    >> one) and the first add-on NIC otherwise (currently only these two are
    >> connected).
    >> 
    >> I think it highly depends on timing when the kernel is scanning for
    >> available devices. As long as you only have NICs handled by a single
    >> driver/module I think you are fine, but add different cards into the same
    >> machine and you will run into trouble...

    > I'm just about to dive into this issue now.

    > Is there some "working" code already, incorporating the BOOTIF/ip information 
    > into scripts/live-premount/select_eth_device?
I use initramfs-tools 0.98.8 in my nfsroot.
There's a subroutinte configure_networking() in script
usr/share/initramfs-tools/scripts/functions inside the nfsroot which
should use the BOOTIF information.

    > And more importantly, is there an easy way to debug initrds without continuous 
    > reboots and set-x / echo?
http://wiki.debian.org/InitramfsDebug. The break parameter will help.

-- 
regards Thomas