Date: Wed, 17 Jan 2001 20:42:02 +0100 From: Cejka Rudolf <cejkar@dcse.fee.vutbr.cz> To: Matthew Jacob <mjacob@feral.com> Cc: freebsd-scsi@FreeBSD.ORG Subject: Re: Troubles with Mammoth2 if there is a tape error Message-ID: <20010117204202.B30484@dcse.fee.vutbr.cz> In-Reply-To: <Pine.BSF.4.21.0101171036200.24797-100000@beppo.feral.com>; from mjacob@feral.com on Wed, Jan 17, 2001 at 10:41:47AM -0800 References: <20010117190636.A28937@dcse.fee.vutbr.cz> <Pine.BSF.4.21.0101171036200.24797-100000@beppo.feral.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Matthew Jacob wrote (2001/01/17):
> Somehow I doubt this. Did you try interupting dd? Is dd getting
Yes. Ctrl-C does not work. Here is a cut from one session with Ctrl-Z
pressed after there are generated error messages in a terminal window,
where I'm logged as a root and with a kill command:
$ cat bin/tapewtest
:
DST=/dev/nrsa0
BS=32k
mt -f ${DST} rewind
mt -f ${DST} comp off
mt -f ${DST} status
time dd if=/dev/zero of=${DST} bs=${BS}
time mt -f ${DST} rewind
mt -f ${DST} comp on
mt -f ${DST} status
mt -f ${DST} offline
$ sh bin/tapewtest
Mode Density Blocksize bpi Compression
Current: 0x28:X3.224 variable 37871 disabled
---------available modes---------
0: 0x28:X3.224 variable 37871 0x4
1: 0x28:X3.224 variable 37871 0x4
2: 0x28:X3.224 variable 37871 0x4
3: 0x28:X3.224 variable 37871 0x4
---------------------------------
Current Driver State: at rest.
---------------------------------
File Number: 0 Record Number: 0
^Z
[1]+ Stopped sh bin/tapewtest
$ kill -9 %1
[1]+ Killed sh bin/tapewtest
$ ps -ax | grep dd
25627 p0 DL 0:11.41 dd if=/dev/zero of=/dev/nrsa0 bs=32k
25646 p0 S+ 0:00.01 grep dd
$ kill -9 25627
and next old lines are lost, but I know that messages were still
generated. I don't know how long I should wait before reboot, but
one of rotated "messages" file has 86 MB, contains 1069569 lines
of error messages and this one were generated exactly one hour,
what means roughly 100 threeline messages per second :-)
But I repeat this test - for assurance and in order
to show full session with kill attempts, if you want.
> an error propagated back? I would presume it would be, unless you're
No, as you can see, dd does report nothing.
> doing something interesting like selecting a 20 megabyte size,
> which will be (silently) split at a 64k sizes.
Do you think "20 megabyte size" as a block size? It is 32k.
> Are you using the drive in fixed length mode?
If I understand correctly, it is variable mode: mt status says
variable (blocksize).
> Error handling might be broken there. Please advise.
Very gladly. This server is relatively usable for experiments
and this form of troubles is very critical for me. Is it
valuable to compile kernel with any CAM debugging options?
I have one FreeBSD 5.0-CURRENT, Sun Dec 10 ... 2000 and this
box has the same problem.
> I can't speak for Novell, but I wrote the Solaris tape driver, and
Great.
> I would prefer to use the FreeBSD driver myself.
Me too, but without these forced reboots ;-)
> It has nothing to do with anything here except that the drive
> is saying that the error is for an event that ocurred in the
> past (not for this command but a previous one).
So the bug has to be in FreeBSD kernel? I thought that it means
that receiver can ignore these errors and the bug is possibly in
the new firmware. I still do not understant, why it worked with
older firmware...
Thanks.
--
Rudolf Cejka (cejkar@dcse.fee.vutbr.cz; http://www.fee.vutbr.cz/~cejkar)
Brno University of Technology, Faculty of El. Engineering and Comp. Science
Bozetechova 2, 612 66 Brno, Czech Republic
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-scsi" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20010117204202.B30484>