Node boot problems
Ethan
snoz at ksu.edu
Fri Sep 20 10:01:25 PDT 2002
Hello all,
I'm new to Beowulf, and having a little trouble with my first cluster.
Currently, the cluster is composed of two (dual Athlon 1900 MP) nodes
with an Intel Pro 1000 (82544E1) in each. I installed SCYLD 27bz-8 on
the controlling node without much trouble. I did have to get the e1000
module from Intel to get the gigabit interface up on the controlling node.
The problem I'm having is that I have been unable to get the
subbordinate node to finish booting phase 1. It fails when trying to
insert the e1000 module with the folowing message.
module install failed
failed command insmod /modules/e1000
no usable network interfaces found
I verified that /etc/config.boot containd the proper hardware id string
and module name. I made sure that all of the copies of e1000.o, on the
controlling node, were the correct version (same as the one that the
controlling node uses to start the gig card). I created new phase 1 and
phase 2 boot images. I mounted the phase 1 nodeboot.img as a loopback
and again made sure that config.boot contained the correct information
and that e1000.o was the correct version. Finally, I made a phase 2
boot floppy and booted the subbordinate node from it. It booted started
the gig nic up and started rarping the controlling node.
I would really like the subbordinate node to boot from the phase 1
floppy and get phase 2 from the controlling node if possible. I thank
you for any help that you can offer.
--
Ethan Bisceglia
Kansas State University
More information about the Beowulf
mailing list