On Thu Feb 7 19:00:11 UTC 2008, Rek Jed rekjed at gmail.com wrote:

Subject was: pxeboot, TFTP only, NFS MOUNT RPC error: 60, timeout

Hey,

I've been building FreeBSD jumpstart infrastructure and it mostly
works.  I'm using tftp  to boot off the network in to scripted
sysinstall. I compiled the boot loader with tftp support but every time I boot it will first try nfs, then timeout after around two minutes (it
cannot find nfs) and finally boot from tftp.  Is there any way that I
can make it boot from tftp straight away rather than wait for nfs to
timeout?


Hi. I'm new to this list, just came across this
thread as I had the same error message - but the
messages posted so far didn't help me. I found a
solution for my case, and I'm posting it here so
maybe it can help others too.

In some cases like mine, the symptoms indicate
that the problem might be caused by differing ways
of interpreting bootp/dhcp information by the
PXE hardware, rather than than in pxeboot itself.

In my case, I had a dhcp "filename" entry which
contained slashes (/), and this was causing trouble
_after_ the referenced pxeboot was correctly loaded
and executed.

Long story:

I wanted the loader to mount root and access
/ and /boot from NFS, not from tftp.

I did get so far that the pxeboot was found
and started, but the NFS root could not be mounted
correctly, therefore /boot and the kernel were not
found. it stopped stating NFS MOUNT RPC error: 60,
timeout after a while, throwing me back into the
loader prompt.

The interesting thing is, a different machine
booted just fine with the same setup. In my case:
a DELL PowerEdge 1750 did boot, a PowerEdge 1950
did not.

My setup:

One server (7.0-STABLE) was serving TFTP and NFS
to the local net, and both nfs root and tftp root were the same:

/pxebootroot -alldirs -rw -maproot=root -network 192.168.1 -mask 255.255.255.0
in /etc/exports as well as

tftp dgram udp wait root /usr/libexec/tftpd tftpd -s /pxebootroot
in /etc/inetd.conf

In my dhcpd.conf on the dhcp server I had the
following entries
(192.168.1.61 being the tftp and nfs server):

filename "/boot/pxeboot";
next-server 192.168.1.61;
option root-path "192.168.1.61:/pxebootroot";

After a few attempts with the above error message
I had a look into lsdev in the loader. It listed dell's
virtual disk as well as, as a last entry, a
192.168.1.61:/boot/pxeboot

this made me think that something strange was going
on: why should that be listed as a device? maybe
it got nfsmounted already? I tried "load"ing
the kernel from there, but I didn't get far.

Whatever was wrong or not with it, it gave me an
idea, ant it helped, in my case
(on the tftp- and nfs server machine):

cp -p /pxebootroot/boot/pxeboot /pxebootroot/

change the filename in dhcpd.conf of the dhcp server:
-> from filename "/boot/pxeboot";
-> to filename "pxeboot";

and it boots fine with NFS root. I think somehow the
slashes (/) within the filename in dhcp were
(mis)interpreted by the pxe hw of the 1950.

I'm still finishing a setup with different pxe boot
environments, so if I come across any other details
for this I'll be glad to share them.


Regards,


Lorenzo




_______________________________________________
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "[EMAIL PROTECTED]"

Reply via email to