Re: sparc 10 hanging
- From: "DoN. Nichols" <dnichols@xxxxxxxxxxx>
- Date: 8 Jul 2009 23:31:47 GMT
On 2009-07-08, merrittr <merrittr@xxxxxxxxx> wrote:
On Jul 7, 11:02 pm, v_borch...@xxxxxxxxxxxxx (Volker Borchert) wrote:
DoN. Nichols wrote:
Hmm ... does the image drive have a bootblock installed?
and the last line I see before the hang is
dump on sdb0 fstype spec size 66628k
Looking at the dmesg of this here diskless 10 running 4.1.4, this
seems to be output by the kernel after probing the devices, so the
bootblock should be OK.
O.K.
[ ... ]
If I can get this booted of of a cd or external disk is there anywah I
could mount the disk and edit a file to reset the time
or possible even chroot and rerun date
It is not a file -- it is information stored in the superblock
of the disk
On a Solaris 10 system in the file /usr/include/sys/fs/ufs_fs.h,
I find (among many other things) the following line:
======================================================================
time32_t fs_time; /* last time written */
======================================================================
Which is apparently updated every time anything is written to the boot
disk.
So -- if you get as far through the boot system as needed to
write *anything* to the disk, the date in the superblock will be updated
to match what is in the system's clock counter, which was normally
loaded from the TOD clock during the boot process.
In an openbsd system the information is found in
"/usr/include/ufs/ffs/fs.h".
======================================================================
time_t fs_time; /* last time written */
======================================================================
So, since they are so similar between SysV (Solaris) and BSD, I
would expect the entry in the equivalent file for SunOs 4.1.3 (a BSD
flavor) to be quite similar. The only visible difference is the
typedef'd name for the fs_time variable.
so what is the correct syntax for date on a 4.1.3 machine
date yymmddhhmm ?
I don't have a 4.1.4 system running any more, so I don't know
for sure. For Solaris 2.6 it is:
/usr/bin/date [-u] [[mmdd]HHMM | mmddHHMM[cc]yy][.SS]
with the "-u" option saying to use GMT instead of the local timezone.
For Solaris 10:
/usr/bin/date [-u] [ [mmdd] HHMM | mmddHHMM [cc] yy] [.SS]
similar, but spaces help delimit the century, year, and seconds if
entered.
For OpenBSD (3.9 and 4.2 at least)
date [-ajnu] [-d dst] [-r seconds] [-t minutes_west] [+format]
[[[[[[cc]yy]mm]dd]HH]MM[.SS]]
rather than giving a different synopsis line for setting instead of
reading, it is all presented as a single folded line.
But -- this problem in reading the date should not stop the boot
process. Remember that it is hit every time you boot Sunos 4.1.x from a
tape or a cdrom, and it continues through this.
If you are really encountering the system halting at this, it
suggests that there is something wrong with the disk which halts the
boot process. Get it to boot and the date problem should fix itself.
So -- first try your duplicate disk -- but if you did not run
"installboot" when you made it, it is probably not bootable.
If you have a CD-ROM of SunOs 4.1.3 or SunOs 4.1.4, it should be
possible to boot from that and run installboot to make the disk
bootable. I am not sure how different the boot block is Between SunOs
4.1.x (BSD flavor) and Solaris 2.x through Solaris 10 (SysV flavor), but
I do know that the later releases of Solaris 10 include options in
installboot for zfs mirrored disks and for hsfs (CD-ROM boot installs)
in addition to the standard ufs bootblock -- and each accesses a
different bootblock example.
Do you have -- or can you get a CD-ROM for SunOs 4.1.3 or SunOs
4.1.4 (assuming that your damaged system was running SunOs 4.1.3)?
Without that, I would consider you to have problems. Hmm ... does Sun
even care about giving away copies of the SunOs 4.1.4 (Solaris 1.1.2)
any more?
Is the company who wrote the software for the robot still in
business? Is the ethernet connection actually to the robot arm? If so,
then you might be able to use a much newer system without problems, and
continue to run the old program. I know that at least up to Solaris 2.6
there were shared libraries to run old SunOs programs on Solaris 2.x
systems.
If the software is hostid keyed then you will need to either fix
your SS-10 or to pay the authors for a newer version keyed to your later
hardware.
And if you don't have a couple of spare SS-10 systems to swap in
if there are hardware problems, I would suggest that you move up to
something newer -- with the CD-ROM or DVD-ROM to re-install the OS at
need. Actually -- I would build a second system which could be swapped
into service while trying to troubleshoot the original one. Old Sun
machines are cheap on eBay. You can get a couple of spare real Sun
SS-10 systems, and get Sun to provide new NVRAM chips since all of them
will be getting older and more problematic.
Oh yes -- another question. When the other person ran the date
command and then tried to reboot the system -- did he power the system
down at all? The system can run forever with a UPS keeping the power
clean even when the battery in the NVRAM chip is dead, because it is
also powered from the system's power supply. Only when the power supply
is switched off for a short time do you actually lose the information in
it.
IIRC -- it came out that you did not have a genuine Sun, but
rather one of the clones -- "Axil" IIRC -- so the implementation of the
NVRAM may be different there. I know that in the Solbourne semi-clones
of the SS-2 machines, the HOSTID and the MAC address were in a bipolar
ROM chip, and everything which changes regularly (the clock chip and the
normal EEPROM settings) are in another soldered-in chip powered by a
coin cell in a holder on the system board. This made it easy to deal
with battery failures. You might check for something like this on the
Axil. Have you looked for a coin cell on the system board? Do you have
hardware documentation for the hardware? Sun finally move the EEPROM
setup to a SEEPROM (8-pin chip which does not need a cell to maintain
it), and the clock ship powered by a coin cell on the system board.
Good Luck,
DoN.
--
Email: <dnichols@xxxxxxxxxxx> | Voice (all times): (703) 938-4564
(too) near Washington D.C. | http://www.d-and-d.com/dnichols/DoN.html
--- Black Holes are where God is dividing by zero ---
.
- References:
- sparc 10 hanging
- From: merrittr
- Re: sparc 10 hanging
- From: DoN. Nichols
- Re: sparc 10 hanging
- From: merrittr
- Re: sparc 10 hanging
- From: DoN. Nichols
- Re: sparc 10 hanging
- From: Volker Borchert
- Re: sparc 10 hanging
- From: merrittr
- sparc 10 hanging
- Prev by Date: Re: sparc 10 hanging
- Next by Date: Leakage current in SF 280R
- Previous by thread: Re: sparc 10 hanging
- Next by thread: Leakage current in SF 280R
- Index(es):
Relevant Pages
|
Loading
