Re: DLT4000 media errors. Bad drive?

[Marco Peereboom <slash@peereboom.us>]

It is probably a bad drive. The LED behavior is normal.

Did you verify all the settings on the SCSI card (reset to factory 
might help)?

On Feb 5, 2004, at 8:55 PM, Dan Brosemer wrote:

> On Thu, Feb 05, 2004 at 08:25:07PM -0600, Marco Peereboom wrote:
>> Weird. Should work normally although some tapes are sometimes moody if
>> they have been used in other systems although I have never seen this 
>> on
>> OpenBSD. Did you try a simple tar command? something like:
>> tar cvf /dev/st0 /bin
>> Oh, and try this twice in a row; I have seen that fix problems before.
>>
>> What is weird is the error code. 0x80 0x00 is vendor specific and I
>> don't have a DLT4K manual handy to look it up. It is likely an 
>> internal
>> fault. Do you have another tape to try?
>> And better, another drive?
>
> Thanks for the quick response.
>
> I've tried with three brand-new tapes.  I don't have a used tape to 
> try or
> I'd try a read test (I think that might help figure things out).
>
> Unfortunately, I don't have another drive to try.
>
> Well, that tar command appears to write (on the first try).  
> Interesting.
>
> During the write, the 20GB and Compression lights blink and the tape 
> in use
> is on solid.  I would expect the opposite, which is why I mention it.
>
> But then when I attempt to read after that:
>
> # tar xvf /dev/st0
> tar: Failed read on archive volume 1: Input/output error
> tar: Attempting to recover from an archive read failure.
> [last 2 lines repeated several times]
> tar: Archive read error limit (10) reached
> tar: Sorry, unable to determine archive format.
>
> On my old Exabyte 8505XL, I used /dev/nrst0 for everything.  Here, you
> suggest using the block device instead of the raw device, and that 
> causes
> the tar to at least think it wrote data fine.
>
> Even a dd write thinks it works with that device:
>
> root@mace:p7[/home/odin/bar]# dd if=/dev/zero of=/dev/st0 bs=10k 
> count=4
> 4+0 records in
> 4+0 records out
> 40960 bytes transferred in 0.000 secs (83421589 bytes/sec)
>
> But a read still fails:
>
> root@mace:p7[/home/odin/bar]# dd if=/dev/st0 of=blah bs=10k
> dd: /dev/st0: Input/output error
> 0+0 records in
> 0+0 records out
> 0 bytes transferred in 54.333 secs (0 bytes/sec)
>
> This time, the failure is accompanied by this from the kernel, though:
>
> st0(siop0:6:0): Check Condition on opcode 0x8
>     SENSE KEY: Media Error
>    INFO FIELD: 2048
>      ASC/ASCQ: ASC 0x80 ASCQ 0x00
>
> Does this help narrow things down at all?
>
> Thanks again.
>
> -Dan