SUMMARY: Booting Problems on a sparc-2

From: vispi@lgc.com
Date: Thu Sep 19 1991 - 14:34:26 CDT


Hi all
I received several responses to my earlier question. Unfortunately since
it was a user down, I didn't have the time to try all the suggestions
that were mail to me. I found it much quicker to reload the OS and
save the important files. (fstab, passwd etc...)
Thanks to everyone who replied. I'm convinced of this group's excellent
functionality.
The summary follows my original posting.
-------

ORIGINAL POSTING.
==>
==>
==>Hi all
==>
==> I am a regular reader, but a first time poster so I may have forgotten
==> some of the obvious information to include. However here goes..
==>
==> We have a very ill sparc-2 here.
==>
==> Info: Sparc 2 running sunOS 4.1.1
==> 2 graphic card (2 heads) cgsix and cgthree
==> Sun 1.0G disk in external bootbox
==>
==>
==> Case:
==>
==> booting single user works just fine.
==> booting multi user goes through the entire boot sequence
==> displays all message, but hangs right before the time when the
==> "console login" prompt should be displayed.
==>
==> The /etc/rc rc.local rc.boot .... files are identical to the
==> files on other sparc 2's
==>
==> the /etc/ttytab file seems to be in order too.
==>
==> a "set -x" entry in each of the rc files showed no anomalies
==> either, and an ls -tl shows that none of the files in
==> /etc and /bin have been modified recently.
==>
==> a read analyze run from format yields a clean / and /usr
==> filesystem
==>
==> Another symptom (if it would be on any help) is that a
==> telnet into the machine yields:
==>
==> Connected to machine
==> Escape character is '^]'.
==>
==> and then hangs.
==>
==> Any suggestions or hints would be greatly appreciated.
==> I will summarize naturally :-)
==>
==> Thanks
==>
==> -Vispi Dumasia
==> vispi@lgc.com
==> (713) 579 4871
==>
==>
==>

-------

SUMMARY:
========

>From (cdr@kpc.com) Carl Rigney:
Look in /dev. Do you have a /dev/console and /dev/pty* and tty* and all that?
If not try a ./MAKEDEV

        => The above files were there.

>From (gene.saunders@West.Sun.COM) Gene Saunders:
Gotta be something with /etc/ttytab

        => I replaced it with a ttytab from an identically configured machine.

>From (parens@dazixco.ingr.com) Paul Arens:
You SS2 sounds just like one we had here. Our problem was cleared by
replacing the /etc/ttytab file and then removing and rebuilding the
/dev/console device. Even if there is no *visible* problem with the ttytab
file, putting a new one in can't hurt.

You may want to check the functionality of the login program also.

If this fails to fix the problem start at the botton of the rc.local file
and start commenting lines out until your system boots. Something is not
starting properly.!

        => checksumming /usr/etc/getty and /usr/bin/login matched with a
        healthy machine. I didn't remove /dev/console

>From (nicky@davinci.concordia.ca) Nick Ayoub:
        Hello,

        I can't say that I am an expert, but in a case like this I would get a
        copy of login and getty from a working sparc 2 and then do a cmp(1) or
        diff(1). This would indicate a problem with the binaries. ftp'ing from
        single user to another working sparc should be ok.

        Another thing you might want to look at are the eeprom settings for the
        console. ( eeprom -i )

        The eeprom suggestion was also sent by Peter (pag@scg.boulder.co.us)
        Gross.

                => This was very informative. I wouldn't have thought of
                this, but this proved to match with a healthy machine too.

        You might try to remove the graphics console and hang a teminal
        off of port A. It should automatically become the new console. This might
        tell you if the problem lies the graphics console ( which is doubtfull
        since it works in single-user.)

                => Did not do this.

        I also like to put a lot of echo statements in the rc scripts. You can get a
        better feel for the flow of the boot-to-multi-user process.

        I hope these suggestions help. I'm sorry that I couldn't be more specific.

                => Did this and the bootsequence ran just fine.

>From (datri@concave.convex.com) Anthony A. Datri:
Check the permissions on /dev/console

        => Checked out okay.

Jim ( napier@ames.ucsd.edu) Napier suggested to do a binary compare of the
login program.

Michael (mikem@juliet.ll.mit.edu) Maciolek suggested to check /etc/rc.single
and check if / and /usr were mounted read-write. Or that portmap died and to
check that by a ps -auwwx at the end of rc.local.

Fabrice (fabrice@sj.ate.slb.com) Le Metayer writes:
Verify that you haven't lost the /usr/bin/login binary, that you are
running /usr/etc/getty on the console (see entries in /etc/ttytab),
and that the /usr/etc/getty hasn't been lost either.

        =>That was verified.

Lee (thx1138@nyu.edu) writes
 I was under the impression that you needed homogenous frame buffers on
 a SS2 for proper functionality (e.g. two CG3s or two CG6s).

        => We have several systems here with one cgsix and one cgthree card
        and they work fine.

Steve (steve@pacacc.com) Riley writes:
This sounds like what happened to me when I accidently deleted my
/etc/ttytab file. No ttytab, no logins.

Mike (miker@sbcoc.com) Raffety writes:
Is your getty binary OK? How about /etc/gettytab?
        
        => both summed fine.

One thing worth mentioning was what the sun support person said.
I should have checked /dev/null and /dev/zero. I'd checked /dev/zero but
not /dev/null.
=====
Once again thanks to all who took the time to reply.

        -Vispi



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:06:19 CDT