ADSM-L

Re: Migration Mishap

1999-05-26 16:05:05
Subject: Re: Migration Mishap
From: "Prather, Wanda" <PrathW1 AT CENTRAL.SSD.JHUAPL DOT EDU>
Date: Wed, 26 May 1999 16:05:05 -0400
In my experience "it depends".  On one type of tape drive I get all the data
back after a WRITE I/O error, on another type I sometimes don't.

What I recommend is exactly what you did - try a MOVE DATA.  If ADSM can
read all the data back, I believe you can rest easy that nothing was lost.
I have had to deal with some very poor quality tape drives, and lots of I/O
errors, and I have never had any reason to suspect that ADSM has missed data
and not told me about it.

If ADSM can't read all the data back, you will see I/O errors on the MOVE
DATA and ADSM will show the tape as having files remaining that can't be
read.  Then yes, the data is lost.  Let ADSM remove everything it can with
the MOVE DATA, then you can list the contents of the tape to see which files
are unretrieveable.

If you are doing read-only archives, then it's just a matter of running the
archive again to make a new copy from those particular source files.

If you are doing archives with DELETE, and you do a lot of them regularly,
then you have a Hobson's Choice:

You can buy enough disk to hold all the archives in the disk pool, and run
BACKUP STGPOOL to create a COPYPOOL tape direct from the disk pool BEFORE
you migrate to TAPEARCHIVEPOOL.  That way you never have to rely on just one
tape to get your data back.

Or, you leave things as they are and migrate part of the data to
TAPEARCHIVEPOOL immediately, as happened last night, taking a calulated risk
of occasionally losing your only copy of the data to a bad tape or dirty
drive.  How great that risk is, just depends on your history with your tapes
and drives.

Good luck..


************************************************************************
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'd think" -
Scott Adams/Dilbert
************************************************************************





> -----Original Message-----
> From: Adam Slesinger [SMTP:aslesinger AT US.BNSMC DOT COM]
> Sent: Wednesday, May 26, 1999 1:25 PM
> To:   ADSM-L AT VM.MARIST DOT EDU
> Subject:      Migration Mishap
>
> Hey everyone,
>
> Test scheduled archives ran on all of our machines last night.  The data
> is sent first to disk (ARCHIVEPOOL) and then, later in the day, is
> migrated to tape (TAPEARCHIVEPOOL).  However, we don't have enough disk
> to hold all of the archived data, so usually about 7G migrates over to
> TAPEARCHIVEPOOL which is the next storage pool. All this is fine except
> for the error that we received early this morning.  Two processes for
> migration took place as the thresholds were met, once at 2:00am and the
> other at 6:55am.  The same tape was grabbed, but as the second migration
> started, an I/O error appeared, the tape was marked unavailable, another
> migration process started, and then a different tape was used instead.
>
> Here are my questions:
>
> Is there any chance that data was lost when the I/O error occurred?
> How does ADSM handle interrupted processes for migration?
> I changed the tape's access back to read/write and then successfully did
> a move data to get the data off the tape so I could eject the bad tape
> and throw it away.  What if I couldn't read tape when trying this
> process?  Is the data on that tape lost for good?
>
> Server: 3.1.1.5 on AIX 4.2.1
>
> Thanks for your help!
>
> adam
> __________________
> Adam Slesinger
> Corporate Information Systems
> Brown & Sharpe, RI
> Phone: (401) 886-2236
> Email: aslesinger AT us.bnsmc DOT com
<Prev in Thread] Current Thread [Next in Thread>