Subject: Can't boot 3100 over network to begin install.
To: None <port-pmax@netbsd.org>
From: Malcolm MacArthur <malcolmm@rustic-place.demon.co.uk>
List: port-pmax
Date: 05/29/1999 16:19:58
Hello all,

Now, I'm afraid I don't really have a great background in Unix/
networking, so I may have missed something obvious here.

At a scrap metal auction (!) recently, some friends and I picked
up a bunch of Decstation 3100's for about five pounds each :-). However,

they were previously used diskless, and despite having RZ24s in them,
don't have Ultrix loaded ("can't find ultrixboot"). So, I spent
four-and-a-half hours(!) downloading all the binary install sets for
NetBSD/pmax, two weeks learning how to set up tftpd/dhcpd, and a
fruitless
afternoon trying to boot the beast.

There are various problems with it. Firstly, the instructions in
INSTALL.txt
are wrong; "boot -f tftp()" sends a request for a file called ""
(nothing)
to the tftp server, despite the fact dhcpd has been set up to send the
file name "install" to the pmax. By setting debugLevel in dhcpd, I got
the
following (this is after I'd sorted out a whole host of other problems,
such as mis-specified access lists, dhcpd not working because there was
no broadcast support for "address" 255.255.255.255 on eth0, etc.)


rustic-place:/tftpboot # tail /var/log/messages
May 29 15:15:47 rustic-place tftpd[785]: request read 'install' from
192.68.23.106; cwd='/tftpboot'
May 29 15:15:47 rustic-place tftpd[785]: access to 'install' for
192.68.23.106 is readonly in list 1
May 29 15:15:47 rustic-place tftpd[785]: sent 'install' to
192.68.23.106: failed

So, it seems the send is failing. It would be nice if it told me why.
Hmmm.
traceroute gives the following output:

rustic-place:/tftpboot # traceroute 192.68.23.106
traceroute to 192.68.23.106 (192.68.23.106), 30 hops max, 40 byte
packets
 1  rustic-place.demon.co.uk (192.68.23.3)  0.180 ms !H  0.134 ms !H
0.108 ms !H

So, host unreachable! But, I can see nothing wrong with the routing:

rustic-place:/tftpboot # route
Kernel IP routing table
Destination     Gateway         Genmask         Flags Metric Ref    Use
Iface
255.255.255.255 *               255.255.255.255 UH    0      0        2
eth0
192.68.23.0     *               255.255.255.0   U     0      0        8
eth0
loopback        *               255.0.0.0       U     0      0        5
lo

(the 255.255.255.255 route added to get dhcpd to work).

And, all I get on the PMAX is:
KN01 V7.02
08-00-2b-1a-5b-55
0x01000000
>> boot -f tftp()install
?477 ordr: tftp()install
?470 bt err: tftp()install
>>

If I reissue the command (without the install), I get:
>> boot -f tftp()
?429 tftp err: 1 (File not found)
(rep. 4 times)
?470 bt err: tftp()
>>

and a search in /var/adm/messages reveals that the tftpd is looking for
a
null-named file.

What am I doing wrong??
The DHCP and tftp servers are running on my Linux (sorry ;) 2.0.35
system running yale-tftpd ("rpm -i" says version 3.0-1; another version
I picked up refused to compile :) and the ISC dhcpd.

I am guessing (but I don't really have a clue!) that the problem is at
my (the server) end.

I have enclosed the relevant parts of tftp.conf and dhcpd.conf below as
well.

Can anyone help? Or are these machines destined to become doorstops :(?
Thanks,
Malcolm.

--
Malcolm MacArthur                 malcolmm@rustic-place.demon.co.uk
"Hardware: The parts of a computer that can be kicked" - Jeff Pesis

--- dhcpd.conf (From the logs, it picks up the IP address OK.)
server-identifier 192.68.23.3;
subnet 192.68.23.0 netmask 255.255.255.0 {
  range 192.68.23.100 192.68.23.150;
}

host dstn06 {
  hardware ethernet 08:00:2b:1a:5b:55;
  fixed-address 192.68.23.106;
  option broadcast-address 255.255.255.0;
  filename "/install";
}

--- tftpd.conf

defaultDirectory   "/tftpboot"
rootDirectory         "/tftpboot"
debugLevel 255
accessList         1  permit     192.68.23.0      0.0.0.255
defaultAccessList  1
restrict pmaxinst 1