Hello Thomas, I've found a workaround, in the process identifying a likely cause of the problem.
On a hunch, I created a bootable Etherboot USB stick, as described here: http://etherboot.org/wiki/removable Usually, this is done for systems with a network adapter that doesn't have pxe capability, but in my case it also worked. This indicates that the onboard implementation of pxe/ftp is somehow broken on my computer. Now I get my root prompt and can continue forward. I will of course have to use that USB stick for installs, but for my handfull of machines, this is not really a problem. As for what gave me the hunch that something wasn't right with the onboard software: The computers I use were former office PCs of a big insurance company. Even before booting, they had the name of that insurance company plastered all over them, indicating that they had meddled with the bios/firmware... Best regards, Daniel On 07/16/2014 05:53 PM, "Dr. Daniel Löb" wrote: > On 07/16/2014 03:53 PM, Thomas Lange wrote: >>>>>>> On Wed, 16 Jul 2014 12:20:40 +0200, "Dr. Daniel Löb" >>>>>>> <dan...@sardisverlag.de> said: >> > When I boot the target machine after that, it _should_ start fai but >> then drop into a root@none prompt. What it instead does is to show these two >> messages: >> > Received 001a Kbyte, sorted 001a Kbyte, invoked primary bootstrap >> > C6e: can't start from RAM disk image >> Mmm. I'm not sure when it drops into the prompt. Do you see the red >> ascii text? Or does it drop into the rescue shell of the initrd? >> I never saw this error message. > Oh, it doesn't drop to the prompt. The Wiki said that it should drop > to the prompt if FAI is working (without having the target system > config properly set up). > > Instead, the error message appears and then the system immediately > boots from the next boot option (Hard Disk). >> You can enable debugging the initrd by adding rd.debug=1 to the kernel >> command line. >> > I tried it again with >> fai-chboot -FIv -k rd.debug=1 joffrey > unfortunately, the result is the same. System shows the error message, > then proceeds to boot from disk. > > I've also activated the --verbose option of tftp on the server and it > gives this output in the logs >> Jul 16 17:39:13 varys in.tftpd[8111]: RRQ from 192.168.28.3 filename >> pxelinux.0 > So it seems that something weird is happening on the target machine, > before/while the kernel is loaded... > > > My Google searches showed that this problem has appeared every now and > then in the last 10 years for people trying to boot over pxe. Somewhat > dishearteningly, not a single one of the people asking questions about > it on forums managed to get it fixed... > > I'm grateful for any further suggestions. > > Best regards, > Daniel > > -- -- *Dr. Daniel Löb*: Geschäftsführer Sardis Verlag UG (haftungsbeschränkt). <http://www.sardisverlag.de> Tel:: +4916091236780 <callto:+4916091236780> Web: www.sardisverlag.de <http://www.sardisverlag.de> Sardis Verlag UG (haftungsbeschränkt) Alexanderstraße 9 64732 Bad König Telefon: +4916091236780 Registergericht: Amtsgericht Darmstadt Registernummer: HRB 93023 Ust-ID: DE293817909 Verantwortlicher i.S.d. § 55 Abs. 2 RStV: Dr. Daniel Löb, Alexanderstraße 9 64732 Bad König
