ADSM-L

Re: Spill processing please help - lowering LOWMIG

2001-10-29 10:10:48
Subject: Re: Spill processing please help - lowering LOWMIG
From: "Kauffman, Tom" <KauffmanT AT NIBCO DOT COM>
Date: Mon, 29 Oct 2001 09:55:59 -0500
Just a minor quibble here, Eric --

The backups WILL go back to the disk pool if there is space - but AFTER
processing the one file that required the media mount. TSM looks for space
for each object as it comes in, and decides for that one object which pool
to use.

Tom Kauffman
NIBCO, Inc

> -----Original Message-----
> From: Loon, E.J. van - SPLXM [mailto:Eric-van.Loon AT KLM DOT COM]
> Sent: Monday, October 29, 2001 8:45 AM
> To: ADSM-L AT VM.MARIST DOT EDU
> Subject: Re: Spill processing please help - lowering LOWMIG
>
>
> Hi Zlatko!
> This is 'normal behavior'. When backups go to a 100% full
> diskpool TSM tries
> to allocate storage in the nextstgpool defined for your
> diskpool, which is
> probably your primary tapepool. So all backups are going to wait for
> mountpoint in the tapepool. That's why the client backups
> seem to 'hang'.
> Although migration kicks in and frees the diskpool, these
> clients will never
> switch back to the diskpool.
> You should always try to prevent a 100% full diskpool. I
> agree that this is
> not always possible, unless you have a very large one. I have
> run into this
> situation numerous times, which can result in very nasty
> problems (Oracle
> databases stop working, because the archive logfiles are not
> backed up).
> Kindest regards,
> Eric van Loon
> KLM Royal Dutch Airlines
>
>
> -----Original Message-----
> From: Zlatko Krastev/ACIT [mailto:acit AT ATTGLOBAL DOT NET]
> Sent: Monday, October 29, 2001 10:55
> To: ADSM-L AT VM.MARIST DOT EDU
> Subject: Re: Spill processing please help - lowering LOWMIG
>
>
> On our test installation (TSM 4.1.4.1, AIX 4.3.3ML6) we've
> seen a strange
> behavior:
> 1. Client backups to disk pool
> 2. Migration starts after HIGHMIG limit is passed
> 3. Client fills the pool faster than migration frees it
> 4. Reaching disk pool 100% full client pauses indicating
> "waiting for mount
> of offline media"
> 5. Migration process commits a migration transaction freeing
> space in disk
> pool. Client does NOT resume.
> 6. Migration process finishes (it's a loooong process).
> Client wakes up and
> continues backup.
> 7. Administrator reduces disk pool deleting some volumes (to
> produce more
> pool fills)
> 8. Client fills the pool again and blocks
> 9. Migration finishes. Client again resumes.
> 8. ...
> 9. ...
> 8. ...
> 9. ... (many times)
> I do not know is this a bug in the release we are using or a
> feature of all
> releases but test this on your platform. If you lower
> significantly LOWMIG
> threshold and something like this happens you can see long
> waits and backup
> can go out the backup window.
>
> Zlatko Krastev
> IT Consultant
>
>
>
>
>
> Alan Davenport <Alan.Davenport AT SELECTIVE DOT COM> on 26.10.2001 18:03:38
> Please respond to "ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU>
> To:     ADSM-L AT VM.MARIST DOT EDU
> cc:
>
> Subject:        Re: Spill processing please help
>
> + -----Original Message-----
> + From: Doherty, John (ANFIS) [mailto:john.doherty AT ANFIS.CO DOT UK]
> + Sent: Friday, October 26, 2001 10:21 AM
> + To: ADSM-L AT VM.MARIST DOT EDU
> + Subject: Re: Spill processing please help
> +
> +
> + Thanks Alan, this looks good.  This is very similar to the
> + problem we have.
> + For example I run migration on our 3 disk pools to clear them
> + down at 18:00,
> + 21:00 and 22:00 respectively.  they finish at 18:32, 21:30 and 22:30
> + respectively.  Can I assume any migration after 22:30 would
> + indicate spill
> + processing is occurring ??
>
> Yes, that is exactly what I was trying to say. Any migration that you
> didn't
> trigger yourself is due to your pool hitting the HIGHMIGPCT limit.
>
> If your pool is filling up too often here is a little trick
> you may want to
> try. Set your LOW migration threshold to 10% or so but leave your high
> migration threshold at 90%. If tape contention with other
> users is not an
> issue this will nearly empty the pool if any unscheduled
> migration is to
> occur. (Setting low migration to ZERO during the backup window is not
> recommended because as SOON as a file hits the disc pool it
> will migrate
> causing lots of tape mounts!)
> +
> + Also I may utilize the 1 GB file limit, but I assume that
> + every time a file
> + is greater than 1 GB then a tape mount would be requested ???
>
> Yes, it will. But if you are using collocation like I am the
> amount of tape
> mounts will be much less for the over sized files than if triggered
> migration were to occur. Setting a max file size on the disc
> pool is one
> way
> of reducing the frequency of triggered migration.
>
> In a perfect world our disc pools would be large enough to
> hold everything
> from a backup window but we all know that's not always
> possible. I'm afraid
> the DASD Czar would come after me with sharp objects if I
> were to try to
> get
> any more packs this month. (:
>
>  Take care,
>      Al
>
> Alan Davenport
> Senior Storage Administrator
> Selective Insurance
> alan.davenport AT selective DOT com
> (973)948-1306
>
>
> **********************************************************************
> This e-mail and any attachment may contain confidential and
> privileged material intended for the addressee only. If you
> are not the addressee, you are notified that no part of the
> e-mail or any attachment may be disclosed, copied or
> distributed, and that any other action related to this e-mail
> or attachment is strictly prohibited, and may be unlawful. If
> you have received this e-mail by error, please notify the
> sender immediately by return e-mail, and delete this message.
> Koninklijke Luchtvaart Maatschappij NV (KLM), its
> subsidiaries and/or its employees shall not be liable for the
> incorrect or incomplete transmission of this e-mail or any
> attachments, nor responsible for any delay in receipt.
> **********************************************************************
>
<Prev in Thread] Current Thread [Next in Thread>