SUMMARY: what is UBC write error(28)

From: Claudia Burg <cac_at_lambda.la.asu.edu>
Date: Mon, 20 Jan 97 10:48:21 MST

thanks all for the help!


From: Kurt Knochner <Kurt_Knochner_at_Physik.TU-Muenchen.DE>
From: Nick Hill <N.M.Hill_at_rl.ac.uk>
From: Thomas Eisele <eisele_at_pfa.research.philips.com>
From: "WHITTAKER, Bruce" <bjw_at_ansto.gov.au>
From: Daniel S. Riley <dsr_at_mail.lns.cornell.edu>
From: Thomas Leitner <tom_at_finwds01.tu-graz.ac.at>
From: Alan Rollow <alan_at_nabeth.cxo.dec.com>

original message:

         hi
         
         i am getting the following error scrolling continuously
         in my dxconsole:
         
                 UBC write error(28) on device(0,0) at block 16 size 8192
                 UBC write error(28) on device(0,0) at block 32 size 8192
                 UBC write error(28) on device(0,0) at block 48 size 8192
         
         i am on a dec alpha 200 4/166 running digital unix v3.2.
         the monitor is a 19" digital vrc21-ha
         
         the device(0,0) seems to be the monitor - according to
         an ls -l in /dev. i have only noticed one wierd display
         problem when the display did not refresh a text file properly.
         i am actually sending this from the computer with the error
         message.

end original message

my summary:
        it seems that the most likely trouble was a combination of
        the terminal device being a writeable device (as mentioned
        by kurt - and later confirmed with ls -al in /dev) and
        way too many graphics and netscape windows being opened
        at the same time. after an already needed shutdown of the
        computer the error went away. as there where no disks or
        other devices that are full on this computer i will attempt
        to follow kurts suggestion and change the terminal settings.

        it seems the best way to learn what the error may be is with:
                grep # /usr/include/errno.h
        where the '#' is replaced by the error number in question

        thanks again for the help!

        claudia-angelica teresa chiarenza burg


responses:

***************************************************************************

UBC is the unified buffer cache. This is the in memory filesystem buffer cache.
Most filesystem writes go into memory and are then flushed to disk at some
later date. I would interpret this as a disk problem in that it is failing to
write to disk device 0 for some reason. I would look in the error logs for
more information. Use uerf to look for disk or SCSI related errors. uerf -R
will list them in reverse order so you see the latest errors first.



Nick Hill

+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
DCI, Rutherford Appleton Laboratory, Tel: +44 (0)1235-445598
Chilton, Didcot, Oxon, OX11 0QX, England. Fax: +44 (0)1235-446626

N.M.Hill_at_rl.ac.uk http://www.cis.rl.ac.uk/people/nmh1/contact.html
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

****************************************************************************
         

It looks like you created the console device as a block special device
whereas it should be a character special device. The output of ls -al
should give you something like:

        crw--w---- 1 root terminal 0, 0 Jan 16 10:34 console
        ^
        |--- character device !!!

To recreate the device:

        boot to single user mode (not sure if necessary, but you will be
                                  on the safe side ;-)
        cd /dev
        mv console console.not-ok
        ./MAKEDEV console

Regards
Kurt

**************************************************************************

Dear Claudia,

the error number refers to the numbers in errno.h. In Your case:

        grep 28 /usr/include/errno.h

delivers:

        #define ENOSPC 28 /* No space left on device */

Device(0,0) normally refers to NFS connections.

So someone on Your machine writes via NFS onto a disk on a different
machine which is full.

Hope it helps, Thomas.

-- 
                                                              ...   __o
                                                           ....   _-\<,
                                                            .... (_)/(_)
+----------------------------------------------------------------------+
| Thomas Eisele                 Email: eisele_at_pfa.research.philips.com |
| Philips Research Labs Aachen (Germany)        Tel.: +49(241)6003-565 |
| Weisshausstrasse 2, D-52066 Aachen            Fax.: +49(241)6003-518 |
+----------------------------------------------------------------------+
***************************************************************************
As the error is fomr the UBC I would have expected the error to mean 
that it is having problems accessing a disk. Probably one of the swap 
areas.
but that's just a guess.
Bruce Whittaker
ANSTO  - Physics Division
email: bruce.whittaker_at_ansto.gov.au
Phone - +61 (02) 9717 3662
Fax      - +61 (02) 9717 3257
and now for the standard thing,,,,
DISCLAIMER: The views expressed in this E-mail message do not =
necessarily
represent the official views of ANSTO from which this message was =
conveyed.
*******************************************************************
Claudia Burg <cac_at_lambda.la.asu.edu> writes:
> 	UBC write error(28) on device(0,0) at block 16 size 8192
> 	UBC write error(28) on device(0,0) at block 32 size 8192
> 	UBC write error(28) on device(0,0) at block 48 size 8192
dsr_lns100> grep 28 /usr/include/errno.h 
#define ENOSPC		28		/* No space left on device */
error 28 is no space on device, UBC is the file system unified buffer
cache, and device(0,0) is what it unfortunately reports for all NFS
mounted file systems.  So something on your system is trying to write
to a full NFS mounted file systems--'df' ought to identify the file
system, but you may need something like lsof to find the process
responsible.
-- 
Dan Riley                                         dsr_at_mail.lns.cornell.edu
Wilson Lab, Cornell University       HEPNET/SPAN: lns598::dsr (44630::dsr)
http://www.lns.cornell.edu/~dsr/ "Distance means nothing/To me" -Kate Bush
**************************************************************************
Claudia,
On Fri, 17 Jan 1997, Claudia Burg wrote:
> i am getting the following error scrolling continuously 
> in my dxconsole:
> 
> 	UBC write error(28) on device(0,0) at block 16 size 8192
> 	UBC write error(28) on device(0,0) at block 32 size 8192
> 	UBC write error(28) on device(0,0) at block 48 size 8192
UCB is the "unified buffer cache" which caches write data in memory
for writing to disk at a later time. The above message seems to
indicate that this disk write failed.
I think you just have some bad blocks on your harddisk. You could
use SCU to fix them. I have not tried this yet myself so I cannot
give you detailed instructions how to do it. Check the "media scan"
comman in SCU. Beware that some of the commands are destructive,
though. Additionally if this is youe system disk, you would have to boot
from the CD to access it with SCU, I think.
I think the error is totally unrelated to your graphics card or
console device.
Hope this helps a bit -- Tom
--------------------------------------------------------------------------
T o m   L e i t n e r                       Dept. of Communications
                                            Graz University of Technology, 
e-mail    : tom_at_finwds01.tu-graz.ac.at      Inffeldgasse 12
Phone     : +43-316-873-7455                A-8010 Graz / Austria / Europe
Fax       : +43-316-463-697
Home page : http://wiis.tu-graz.ac.at/people/tom.html
PGP public key on : ftp://wiis.tu-graz.ac.at/pgp-keys/tom.asc or send 
mail with subject "get Thomas Leitner" to pgp-public-keys_at_keys.pgp.net
--------------------------------------------------------------------------
*************************************************************************
	This was a misfeature added in V3.2something.  All the
	write errors out of the buffer cache were logged in the
	form shown.  When the file system is NFS mounted, it
	uses the device number 0,0.  The errno value associated
	with the error is the 28, which is ENOSPC (no space).
	While write errors from the are significant, NFS write
	errors are pretty common, and this logged all the NFS
	write errors.  You can see the other errno values in the
	include file /usr/include/errno.h.

Received on Mon Jan 20 1997 - 19:14:09 NZDT