Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 8 Dec 96 21:38:24 -0500
From:      curt@kcwc.com (Curt Welch)
To:        Jeff Aitken <jaitken@dimension.net>
Cc:        questions@freebsd.org
Subject:   Re: HP T4000s tape drive
Message-ID:  <9612090238.AA06537@mail.kcwc.com>

next in thread | raw e-mail | index | archive | help
>  I recently purchased an HP T4000s "Colorado" SCSI-2 tape drive

Me too.  I have two of them on FreeBSD boxes.  Also running 2.1.5R

>  I'm a little unsure just how I should invoke, e.g.,
>  dump(1) to have things work "optimally".  If I specify
>  no arguments at all, like this:

>	dump 0uf /dev/nrst0 /home

>  then I get nowhere near the advertised capacity (the tape
>  "fills up" after only a few hundred MB's).  So, I gather
>  I need to specify the density and length of the tape (via
>  the 'd' and 's' options, respectively).

>  Is it as simple as: 


> 	dump 0udsf 67733 740 /dev/nrst0 /home

>  or not?  I presume that the "bit density" the manual
>  mentions is the bits per inch (BPI)?  The tapes claim to
>  be 740ft (225.6m) in length.

Don't waste your time with the d and s options.  Dump uses
a complex formula for mapping those numbers into what it
thinks the tape will hold in terms of blocks.

Use the newer and better B option.  (I don't known when this
got added, but FreeBSD was the first system I saw this on.)

With B, you just specify how many blocks the tape can hold, a
much cleaner way to do things.  However, It's unclear to me
if these "blocks" are 1K or 8K, or the size specified by b.
I would assume they were the size specified by the b option,
but some testing I did made me wonder if this was correct.

>  Just for fun, I've experimented with a variety of
>  different density/length values, none of which satisfy
>  me (yet).  Using the above-mentioned values doesn't even
>  come close, so I've obviously missed something.

Yes, what dump does with the d and s numbers is a mystery (unless
you read the source).

>  I figure I must be missing something, since my drive
>  hardly "rips right along".  I'm seeing between 100 and
>  300 KBps during backups, far from the advertised 514KBps
>  (which, no doubt, only occurs under "optimal"
>  circumstances, like when using the HP-provided software
>  under an HP-supported platform :-) :-)

Yes, I too had the same problem.  A test program to fill the
entire tape up took over 8 hours to run which works out
to about 7.5 Meg/minute (or 122KBps).

The trick I found was to use block sizes of at least 15K.  I
now normally use 32K blocks.  Once you do this, the drive
can stream (write without pausing and backing up).  My test
program was then able to fill the tape in 2 hours and 21
minutes.  This works out to 27.9 meg/minute (476KBps).

BTW, my test program wrote 126042 blocks of 32768 bytes for
a total tape size of 4130144256 bytes, which is 3.9 Meg in
1024^2 terms.

For dump, I'm using options like this:

  dump 0ubBf 32 1200000 /dev/nrst0  /dev/some-disk

Since I put multiple backups on a single tape, the only thing
important about the B number is that it's large enough to keep
dump from stopping too soon.

I have some questions about these drives (and the st driver) also.

When the system boots, I get:

  ahc0:A:4: refuses syncronous negotiation.  Using asyncronous  
transfers
  (ahc0:4:0): "HP T4000s 1.06" type 1 removable SCSI 2
  st0(ahc0:4:0): Sequential-Access density code 0x0,  drive empty

Is this message about refusing sync[h]ronous negotiation normal
for SCSI tape drives?  Is it normal for this drive?

And every time you access the drive when it's rewound, I
get the kernel message:

st0(ahc0:4:0): ILLEGAL REQUEST asc:20,0 Invalid command operation code

Can anyone tell me what's causing this?  It doesn't seem to
be a problem but it bugs me and I'd like to find a way to get
rid of the message.  I assume the st driver is sending some
type of command to the drive that it's rejecting.  Can anyone
tell me what it might be trying to do or how I might go about
figuring this out?

And two or three times, I've gotten:

st0(ahc0:4:0): timed out in dataout phase, SCSISIGI == 0x0
st0(ahc0:4:0): BUS DEVICE RESET message queued.
Bus Device Reset Message Sent
st0(ahc0:4:0): Bus Device Reset delivered. 1 SCBs aborted
ahc0:A:4: refuses syncronous negotiation.  Using asyncronous transfers

When this happens, the drive stops working until I reboot.
Any advise on what might be causing this and what might be
done about it?

Curt Welch
curt@kcwc.com



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?9612090238.AA06537>