Amanda-Users

Re: tapetype questions

2003-04-16 18:27:54
Subject: Re: tapetype questions
From: Gene Heskett <gene.heskett AT verizon DOT net>
To: KEVIN ZEMBOWER <KZEMBOWE AT jhuccp DOT org>, amanda-users AT amanda DOT org
Date: Wed, 16 Apr 2003 16:27:01 -0400
On Wed April 16 2003 12:19, KEVIN ZEMBOWER wrote:
>I'm trying to run tapetype on my Seagate 20G/40G (Archive Python
> 06408) drive, in a Dell PowerEdge 2450 host, using a DDS-4 tape.
> From running tapetype before, I know that it can sometimes run a
> long time, often longer than 24 hours.
>
>I kicked it off Monday morning, and it ran almost 48 hours. I
> guessed that everything was okay, because of the flashing lights
> on the drive. However, I got suspicious after 48 hours and
> interrupted the job. The only output was a message, "Could not
> write any data in this pass. Short write." Just this morning, I
> started it again, and got this same message in the output file
> after a couple of hours. There was no job shown by ps involving
> tape, but the lights on the drive were still flashing merrily.
>
>Does this message indicate normal operation, and should I just be
> more patient and wait longer than 48 hours? Is the problem how
> I'm trying to run the job? I don't want to leave a terminal open,
> so I've tried to run it by piping the command to 'at now' and
> using 'nohup ... &'. Is there a better way?
>
>Thanks for your thoughts.
>
>-Kevin Zembower

Dat drives are a bit slower than some of the other more expensive 
drives, but 48 hours to do a tapetype to what should be a 20gig 
drive?  Nah...  Theres something else going on here I think.

Kevin, be aware that the tapetype program uses /dev/urandom as its 
data source, and that this random data will expand 10 to 20% in the 
drives compressor if its enabled,  Find the dipswitch and shut it 
off for good.  Using the compression in the drive hides the real 
tape capacity from amanda, and its NOT the 40gigs they claim in 
their sales propaganda. Amanda counts bytes going to the drive and 
if the drive is massaging the data stream, then amanda has a pretty 
poor idea of what the drive can hold.  If you tell amanda to use 
its compression, then its the compressed data that gets counted, 
and this is much more accurate, varying not more than a couple of 
percent from tape to tape.

Also, DDS drives mark the tape as to the compression status when the 
tape was written, and will override your choice as the tape is 
inserted and recognized by the drive each time.

So in order to clear that flag, one must set the compression to off, 
and then cause a buffer flush.  Thats best done by dd'ing the tapes 
label out to a scratch file using the rewinding version of the  
device, then doing the compression off thing with mt, and then 
dd'ing the label back to the tape using the rewinding device.  That 
*should* convert that tape to an uncompressed tape.  Note that this 
is seperate from setting the dipswitch.  The dipswitch just sets 
the default at powerup.

See the manpages for dd and mt for details.

-- 
Cheers, Gene
AMD K6-III@500mhz 320M
Athlon1600XP@1400mhz  512M
99.26% setiathome rank, not too shabby for a WV hillbilly
Yahoo.com attornies please note, additions to this message
by Gene Heskett are:
Copyright 2003 by Maurice Eugene Heskett, all rights reserved.


<Prev in Thread] Current Thread [Next in Thread>