Hi, We have custom boot-cd, with a totaly non redhat kernel and / those boot-cd give me network+disk drivers on any machine we have. >From those boot CD, i try to start an ananconda redhat install, and sofar results are not 100% satisfactory : Typically, 1- i boot from my custom CD, 2- NFS mount the content of stage2.img into a /KS directory 3- do some mount -o bind to build a suitable root directory for anaconda (/proc, /dev, /tmp...) 4- NFS mount a copy of the 3 install disks under /KS/DISKS 5- ask the enduser some questions, guess some answer, and put a suitable kickstart.cfg in /KS 6- start anaconda in this chrooted env by : chroot /KS /usr/bin/anaconda --kickstart /tmp/anaconda.cfg --text -m nfs://DISKS Step 3 is not easy, because i don't know what anaconda will use/expect in it's chroot filesystem... For example, it needs a valid /etc/mtab ( which i made link to /KS/proc/mounts, KS/proc beeing a bind-mount from the real /proc). Currently everything is working on an ide-based Pc. However, on a scsi-based Pc, the /boot/initrdXXXYYYZZZ is not corectley generated by the postinstall script of the kernel rpm... Debugging this seems to be a nightmare and i'd like to know how to find the root filesystem on which stage2.img is loaded during the standard CDROM install.. is there a stage1.img somewhre ? What are hdstg2.img and netstg2.img ? Any thoughts on our approach ? -- Nicolas Turro <Nicolas.Turro@xxxxxxxxxxxxxxx> INRIA
