problem with nfs install

Robert G. Brown rgb at
Tue Jul 2 10:35:27 PDT 2002

On Tue, 2 Jul 2002, Greg MATTHEWS wrote:

> > 
> > I tend to use url (e.g.)
> > 
> > url --url
> not sure our webmaster would like all that on the server! i may have to set up 
> my own.
> > (which won't work outside of duke, but you get the idea).  Note that the
> > i386 path above points to the ENTIRE i386 image, not just the Redhat
> > directory.  I think that you should have used something like:
> > 
> >   cp -a /mnt/cdrom /export/rh-7.3
> i think i had the syntax right before... this will create a directory called 
> cdrom in /export/rh-7.3

If rh-7.3 already exists -- I was assuming that it doesn't so the copy
is ONTO.  If it already exists, try

  cp -a /mnt/cdrom/* /export/rh-7.3

to copy the contents of the CD INTO rh-73.

> no the RedHat directory was definitely there in fact i simply followed the 
> instructions in the readme on the cd for copying the relevant files for a 
> network installation this creates the RedHat directory and all its 
> subdirectories. the structure of the two exported versions looks identical, 
> its just that one works and the other doesnt!

Well, then you've got me stumped.  When it starts, do you get any
message about it getting e.g. netstg1.img or stage2.img?  I still think
that the problem is that the raw / image(s) aren't getting loaded into
the loopback-mounted ramdisk.  I also assume that you're using the
correct kernel image (e.g. bootnet.img) so that you CAN access an NFS

Can you do a hand-install of the node via the NFS image?  That is, boot
bootnet.img into text mode, and at least get it through the point where
the NFS image is mounted, the netstg1.img is retrieved, and it is
formatting the filesystems?  The alternate consoles in text mode provide
you access to both running logs of the install (including error messages
about e.g. paths or permissions) and even gives you a shell where you
can do a few simple things by hand (like attempt to mount the install

There are other questions I might ask; is dhcp giving the new node its
IP number so that it knows who it is at boot time?  How are you
providing it with the ks file path (via dhcp and a server, I presume)?
Booting it into text mode and watching the install proceed is a decent
way to debug the problem.

> thanks for trying.. i'm wondering if i've mounted the partition that i'm using 
> for the 7.3 stuff in a way that is messing with the nfs exporting, altho the 
> nfs server doesnt complain in any way and i can mount them both from a node 
> that is already installed.

In that case, I doubt that this is your problem although if your
installed node isn't getting its IP number correctly it might be.

Good luck


Robert G. Brown	             
Duke University Dept. of Physics, Box 90305
Durham, N.C. 27708-0305
Phone: 1-919-660-2567  Fax: 919-660-2525     email:rgb at

More information about the Beowulf mailing list