On Sunday 18 September 2005 02:03, Jon LaBadie wrote:
>On Sat, Sep 17, 2005 at 11:25:48PM -0400, Florian Lengyel wrote:
>> Here's the post-mortem: I've been using a hand-me-dowm
>> Spectra Logic 2K. No matter what tape I tried in drive 0
>> (the only one of two that my AMANDA configuration seems
>> to recognize) I had I/O errors. Even an innocuous command such as
>>
>> mt -f /dev/nst0 compression off
>>
>> [by the way, the argument "off" is worth a small fortune in
>> consulting fees, since it has to be guessed]
>>
>> Resulted in an I/O error. Such things are often caused by
>> hardware trouble. After checking cables, interface card seating,
>> and other things, I tried using the cleaning tape.
>>
>> Problem solved.
>
>You wouldn't think a 'compression off' request would have
>any involvement with possibly dirty heads.  Here is a guess,
>I've seen some tape drives that when an error happens flag
>the error and refuse to do anything until it is cleared.
>At one client they had a DLT drive.  Anything went wrong
>and the "clean drive" light came on.  Then you could do
>nothing unless you cleaned it or 'I think' there was some
>manual way to turn off the light.  Maybe power off/on :)

Thats been the case here Jon.  If they have the cleaning flag set,
not even a powerdown was able to restore one of those old DDS2 drives
for command response.

>BTW you mention you have other ?drives? not being seen?
>Based on replies here, most linux kernels default to only
>scanning LUN 0 on the scsi buss.  Some configuation change
>lets it scan others.  Similarly on my Solaris system, by
>default only scsi IDs 0-7 are scanned for tape drives.
>Again a config setting allows IDs 8-15 to be scanned.
>
>Maybe something similar is your problem.

Thats been a constant problem for those running Red Hat built
kernels, Red Hat doesn't want to waste time scanning for them when
only .1% of the users have the scsi stuff in the first place.

Linux, I believe, also cheats in that scanning process as even when
its turned on (its a kernel config & recompile option), the scsi scan
is completed 50x faster than the specs say, which proclaim each
device being scanned has 1 second to respond to a query.  Thats 49
seconds for all LUNs and all addresses on a scsi-2 narrow bus but
linux swings thru that in 3 or 4 seconds.  It seems to work most of
the time though.

-- 
Cheers, Gene
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
99.35% setiathome rank, not too shabby for a WV hillbilly
Yahoo.com and AOL/TW attorneys please note, additions to the above
message by Gene Heskett are:
Copyright 2005 by Maurice Eugene Heskett, all rights reserved.