Original Post:
>Last weekend, we upgraded a Sparc 1+ running SunOS 4.1.1B to a Sparc-10 running
>Solaris 1.1.1. After the upgrade we used add_services to add support for sun4c
>architecture clients. However, diskless sun4c clients no longer boot: they
>simply hang at "booting from le()" and eventually - after 5 or more minutes -
>time out with "tftp receive failed". There are no other error messages.
>
>Etherfind can find the clients. Bootparamd is running and bootparms is
>correct. /etc/ethers has no blank lines or other mistakes. Rarpd is running
>and works correctly. The file systems are exported correctly. Other
>peripherals such as network printers succesfully boot from the (new) Sparc-10
>server. Only diskless sun clients have the problem. (P.S. sun4m clients also
>hang in the same way -- we tried to add_client one as a test).
>
>Sun sent us a Solaris 1.1.1 problems document and we looked through it for
>something related to our situation without luck ... so any help from this list
>will be greatly appreciated.
The solution (at least for us):
We never discovered precisely why the clients failed to boot, but reinstalling
from CD made the problem go away. The part of this procedure that is not well
documented is "How do I get a Sparc 10 running 4.1.3m to support 4.1.3c clients
as well as 4.1.3m clients?"
The answer is: After you install your new system from CD, leave the CD on line
and reboot using the newly-installed kernel. Go to /usr/etc/install and run
add_services. Choose the "Add New Release" option on the form. You'll be
presented with a seris of questions that will eventually allow you to install
support for other architectures.
For those who asked, the document we got from Sun (via FAX) is:
Infodoc ID: 1269
Synopsis: Diskless Boot Troubleshooting Guide
It contains some interesting info, but nothing that helped us get to a solution
to our particular problem.
Thanks for responses to:
mulara88@matrix.newpaltz.edu (james mularadelis)
thimm@idiap.ch (Georg Thimm)
poffen@San-Jose.ate.slb.com (Russ Poffenberger)
Dan Stromberg - OAC-DCS <strombrg@hydra.acs.uci.edu>
mike@trdlnk.chi.il.us (Michael Sullivan)
david@srv.PacBell.COM (David St. Pierre)
adamfox@super.org (Adam Fox)
Caspar Arquint <arquint@inf.ethz.ch>
stlee@patch.tandem.com
----------------------------------------------------------------------------
Several respondents provided useful advice that might be interesting to others.
Details follow:
--From: poffen@San-Jose.ate.slb.com (Russ Poffenberger)
I assume you did "add_client" for the clients? Even if you moved the stuff over before, I suggest you re-run it. You may have to delete the clients first.
--
From: Dan Stromberg - OAC-DCS <strombrg@hydra.acs.uci.edu>
Be sure you have the appropriate symlinks in /tftpboot. Be sure you have tftpd enabled in /etc/inetd.conf.
Try killing rarpd and rpc.bootparamd, and starting (probably in two different xterm windows) with the "-d" option.
If you have a solaris 2 box, use snoop to watch the packet activity. Snoop often tells you exactly what file was being hunted for, when there's a file that can't be found. There is a snoop available for solaris 1, though it's unsupported there.
--
From: adamfox@super.org (Adam Fox)
You didn't say anything about if tftpd was set up. I suggest you check the following:
1) That tftp is NOT commented out of the /etc/inetd.conf file of your boot server. You normally don't want tftpd, except when you are booting diskless.
2) That tftp is no commented out of your /etc/services file.
3) That you have a /tftpboot directory with entries for the Suns you are trying to boot. You'll need the IP address in hex and an <ip_address_in_hex>.`arch -k`. These should be symlinks to the proper boot block in the /tftpboot directory.
For example: If I'm booting a sun4c at 192.31.192.5, then I would need:
C01FC005 and C01FC005.SUN4C
as symlinks pointing to the boot block for sun4c.sunos.4.1.x
Use bc to find the hex values
quine: bc obase = 16 192 C0 < -- 192 = C0 31 < -- 31 = 1F 1F 192 C0 5 < -- 5 = 05 (use 2 digits each) 5
Use etherfind on a 3rd host to see if you are actually tftp'ing.
--
From: Caspar Arquint <arquint@inf.ethz.ch>
I had asimilar problem. While upgrading a sparcstation 1+ running 'sunupgrade -v' always produced 'Unable to ... tftpboot ...'. Don't remember exactly. The fact was that in /tftpboot (where our boot images for the clients laying) the symlinks where pointing to boot.sun4c.4.1.3 (something like that). But there was only a file called boot.sun4c.4.1.3_U1 ! All I did was copy the boot*U1 to boot*4.1.3! After that the clients booted as usual.
-- alfred Internet: unisql!alfred@cs.utexas.edu UUCP: cs.utexas.edu!unisql!alfred
Ears are what keep your feet from getting in your mouth. -- Anonymous
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:08:58 CDT