Hi guys, I am trying to set up a server from where clients on a network can do a diskless boot of fedora 16 through pxe. I have set up the server with dhcp,tftp,nfs correctly and all are confirmed to be working. According to the guide mentioned here I installed dracut-network on the server, generated initramfs from the appropiate kernel, and placed it under tftp home and copied the filesystem to the NFS export. When I try to do the pxe boot at the client side, the client boots up to the pxe config and loads the kernel and initramfs. Then during the initialization I get the following error-

and then it drops to the debug shell.

I think this is happening because dracut is trying to get new IP address on top of the pxe assigned one. someone please help me.

I think something should have shown below that last line.

See this maybe.

"http://serverfault.com/questions/399990/pxe-boot-fedora-17-error"

your "clients" are aware that in TWO MONTHS fedora 16 hits it's END OF LIFE
and that you will be forced to install fedora 17 or 18
and if 17 then in 6 months after that ( July 2013) install fedora 18 ( or 19) when fedora 17 hits EOL

in 60+ ,or so, DAYS there will NEVER be any more security updates to 16 .
It will be dead for all practical purposes .

I can install the os at client side over pxe with the "repo=" option, but I want the rootfs to be directly served from the server.

By "clients" i meant client "computers". This is kind of a pet endeavor of mine, It's not my responsibility or final objective to set up the system, I'm just doing it out of curiosity. As for your advise, I tried to install fedora17 on one of the machines, but it repeatedly failed to boot in graphical mode throwing some kind of "render error" and fedora16 worked, so I went with that. I will try to do the same with fedora 18 though and I'll update the outcomes, but I just wanted to know what I am doing wrong with the current setup.