Amanda-Users

Re: Onstream ADR50 Block Size

2003-04-17 15:27:32
Subject: Re: Onstream ADR50 Block Size
From: Gene Heskett <gene.heskett AT verizon DOT net>
To: Moritz Both <moritz AT daneben DOT de>, rwk AT americom DOT com
Date: Thu, 17 Apr 2003 13:47:53 -0400
On Thu April 17 2003 08:56, Moritz Both wrote:
>rwk AT americom DOT com wrote:
>> [...]
>> Gene,
>> I would just like to know if anyone else out there is using an
>> ADR50 on Linux and if so, what you did (or why it works for
>> you).
>
>See: http://groups.yahoo.com/group/amanda-users/message/32158
>
>It simply doesn't work, as far as I tried.
>
Moritz, I believe, from the gist of a discussion we had here last 
week, that the real problem seems to be a bug in the ADR drives 
firmware.  And apparently OnStream did admit it before they did the 
big backflop into bankruptcy pool.  But they never released that 
anybody knows of, a bugfixed firmware file.

It appears that it forgets where its at (mt -f /dev/ice tell) if it 
sits idle for much over a minute.  Since amdump first verifies it 
has the right tape, then goes off to get estimates and such, often 
it may be 15 to 20 minutes before the next write to the tape.  
Apparently it cheerfully rewinds and writes the data from the BOT 
position.  By-by label and any amanda related info in the first 
block would be my best guess.  Somebody with such a drive and a 
copy of dd can probably verify or deny that this is what happens.

One suggsted cure is to leave the tape out of the drive while amdump 
is running, and when its done, run amflush which will then write 
pretty close to non-stop till its done.

Another person suggested useing a dummy tape device that is rm'd at 
the top of a wrapper script that runs amdump and amflush in order, 
and when amdump is done, it re-creates the dummy link and 
immediately runs amflush, thereby makeing it a middle of the night, 
no bodies required 'cron'nable job.  The device doesn't exist when 
amdump is run so it uses the holding disk only, but it does exist 
when amflush is run.  I'd be carefull of specifying any sleep 
delays in the amanda.conf though, going without any, and adding 
maybe 5 seconds till it works without amanda claiming the device is 
still busy or some such foofurah.

The drive is a big enough drive to be usefull at least until its 
heads bite the big one, eg 5 to 15 thousand hours of runtime, and 
its a shame that so many got caught out on it.

>Greetings,
>Moritz

-- 
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.