ADSM-L

STK autoclean (was 3570/3575 autoclean)

1998-09-11 12:07:16
Subject: STK autoclean (was 3570/3575 autoclean)
From: "Prather, Wanda" <PrathW1 AT CENTRAL.SSD.JHUAPL DOT EDU>
Date: Fri, 11 Sep 1998 12:07:16 -0400
Craig,

With STK libraries I think "it depends" which library you are using.
With some ACSLS-driven libraries, cleaning is done under the control of
ACSLS.
With the 9710 library in particular, the function was moved outboard into
the microcode whether you are using ACSLS or not.

With the 9710, when it works properly, it works exactly as you describe
autoclean for the 3570/3575.  There is a small cable that runs from each
drive back to the robot; when the drive detects reduced signal strength, it
calls for a clean and the robot will automatically mount the cleaning
cartridge when the current data cartridge is dismounted.

The cleaning cartridge resides in a special slot, and neither the cartridge
or the cleaning operation is visible to ADSM.

We never had problems getting this to work with 3490 drives in the 9710.  We
did initially have problems getting it to work with DLT drives in the 9710.
They originally thought it was related to auto-load, but it turned out that
our driver for the robot (the mchangerx device) had a 5 minute timeout
limit, and it can take longer than 5 minutes to clean a DLT drive.

The DLT cleaning tape is not read from beginning to end; each clean uses
1/20th of the tape, then writes a tape mark.  The next clean starts at that
tape mark, and uses more of the tape.  So the first clean on a new cleaning
tape takes 3 minutes, and the 20th clean takes over 5 minutes.  Anyway, we
finally got a driver change from STK that increased the timeout limit and
resolved that issue.

It's still not a great design, because the DLT drives will call for a clean
when the heads get dirty, which may happen to be at the beginning of a DLT
cartridge that ADSM is trying to reclaim with 10K files on it.  So the
reclaim will sit there and get I/O errors for hours and make no progress.
The 3490 idea of cleaning after a specified number of tape feet processed
works much better, but that is really a DLT design issue.

I have no idea whether the 9714 and 9730 STK libraries autoclean the same
way or not.

This is probably more than you wanted to know already, but if you want to
know more about the 9710, feel free to contact me directly.

**********************************************************
Wanda Prather
The Johns Hopkins Applied Physics Lab
443-778-8769
wanda_prather AT jhuapl DOT edu

"Intelligence has much less practical application than you would think" -
Scott Adams/Dilbert
**********************************************************







> -----Original Message-----
> From: Craig Bell [SMTP:rcbell AT us.ibm DOT com]
> Sent: Thursday, September 10, 1998 11:12 AM
> To:   ADSM-L AT vm.marist DOT edu
> Subject:      Re: 3570/3575 autoclean
>
>
> The 3570 and 3575 autoclean feature does not interfere with ADSM.  The
> 3570 has
> it's own slot for the cleaner that is not visible to ADSM, and the 3575
> hides
> the cleaners from ADSM.  I encourage you to always use autocleaning if you
> have
> a library that supports it.
>
> I have heard that there are issues with STK libraries using  autocleaning,
> I
> believe that are related to "fast load", but have never been presented
> with
> enough information, either here or in a PMR, to see if there's a workable
> solution.  Would anyone like to comment on this?
>
> Craig Bell
<Prev in Thread] Current Thread [Next in Thread>
  • STK autoclean (was 3570/3575 autoclean), Prather, Wanda <=