[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Building headless network install the proper way



I need help doing properly what I did before by modifying an initrd.

A while back, I used the "twin" package and modified the sarge network install ramdisk to support a fully interactive headless network install.

By controlling the reset switch with a serial port controlled relay module, and a dhcp/tftp server, this allowed me to bring up machines from scratch and do a full install if needed.

I built this the hard way by adding files and libraries to the initrd till it all worked, running twin continuously till it stayed up (network had come up) and pre-supplying some of the installation answers and running at high-priority so not all questions were asked.

This was enough for the installer to bring the network up and then pop up the installer under twin, on my X server (specified in kernel boot string).

I now have the kit installed to build by own debian-installer setup and I guess to this I am going to have to add udebs that contain twin, the extra init scripts and so forth.

er... help!

I'm looking to know which config files I need to mod, what to do to get twin compiled so it doesn't need a full glibc, help help
etc

I'm not dumb... I think.... so specific questions:

1) I want to add a package to the initrd as built using the debian-installer builder. How do I go about having the package build statically? Is this what I should be doing?

2) Is there a better way to have the network brought straight up by DHCP or with kernel boot config strings?

I'll be testing the ramdisk using colinux initially and when I think its right I'll try on some real hardware.

Sam



Reply to: