hallo folks,

I have got a second hp300 "without everything" (hp next door to me throws them
away and replaces them with pentiums ... I can't understand them), therefore
I want to use it diskless as a client for my hp300-netbsd1.2 server.
I did everything mentioned in "man diskless", but it doesn't work:
the kernel is loaded and starts to initialize the hardware, then suddenly
the server refuses to answer to the client's requests. tcpdump on the server
says:
<apollo is the client and utopia is the server>
:
:
15:39:24.937928 apollo.stars.de.1022 > utopia.stars.de.704: udp 88
15:39:29.967560 apollo.stars.de.1022 > utopia.stars.de.704: udp 88
15:39:34.997172 apollo.stars.de.1022 > utopia.stars.de.704: udp 88
:
:
<this goes infinitly, no not really: after 1/2 hour the client reboots
and the same procedure begins>
:
:

the client on other side prints on the screen:
:
:
le0 at scode21 ipl5: adress 08:09:03:f8:7d
le0 8 receive buffers, 2 transmit buffers
root on le0
nfs_boot: using network interface: le0      < correct>
nfs_boot: client addr: 0x c0 a8 00 07	    < correct = 192.168.0.7, apollo >
nfs_boot: server addr: 0x c0 a8 00 03	    < correct = 192.168.0.3, utopia >
nfs_boot: hostname=apollo.stars.de	    < the client >
root on utopia:/usr/apollo.stars.de/root    < here is the kernel, etc >

RPC timeout for server 0x c0 a8 00 03	    < this correspondents to    >
RPC timeout for server 0x c0 a8 00 03       <  the lines above (server) >
RPC timeout for server 0x c0 a8 00 03
:
< ad infinitum >
:
:

has somebody an idea what can go wrong at this stage in the process?


some further information:

0. I tried it with different kernels: the generic and the diskless one.

1. the client has mounted the appropriate directory:
     utopia# showmount  -a
     All mount points on localhost:
     apollo.stars.de:/usr/apollo.stars.de/root (here is the kernel etc)

2. the kernel is loaded, that implies that the 2 machines can communicate and
   that the following files are correct
   /etc/ethers
   /etc/hosts
   /etc/rbootd.conf (and /usr/mdec/rbootd/SYS_NBOOT)
   /etc/exports is correct (because of 1.)

3. furthermore I did the stuff for the client mentioned in "man diskless",
   except that I have no fstab.nfs (what's that?). I did a "MAKEDEV all" in
   /usr/apollo.stars.de/root/dev (/usr/apollo.stars.de/root is the client's
   new root directory), copied everyting from /bin, /sbin, /lib to the new
   directories under /usr/apollo.stars.de/root. then I did the stuff in etc
   mentioned in "man diskless".

4. a ping to the client is answered, but (of course) telnet, etc. fails.


thanks
      Stefan Schaeckeler



ps. put a little skrewdriver on your hp-monitor (my: 98753A) an turn the
    monitor on: the skrewdriver then makes some turns ... oh oh, this looks
    not very healthy ...