My understanding is that if TSM is busy dismounting a tape, then gets a
request to migrate more data to that node, it will call for a new scratch
tape, even if the tape it is busy dismounting is not full. The logic is
that it is faster to mount a fresh scratch tape than wait for the tape that
is being dismounted to be returned to its library slot then remounted
again. As you are running migration alongside backups I can see how this
can happen on a frequent basis.
Jim Armstrong
Storage Architect
Standard Life
|---------+---------------------------------->
| | "Maurice van 't|
| | Loo" |
| | <tsm@COMPARECOM|
| | PUTERS.NL> |
| | Sent by: |
| | "ADSM: Dist |
| | Stor Manager" |
| | <ADSM-L AT VM DOT MARI|
| | ST.EDU> |
| | |
| | |
| | 08/09/2005 |
| | 13:13 |
| | Please respond |
| | to "ADSM: Dist |
| | Stor Manager" |
| | |
|---------+---------------------------------->
>---------------------------------------------------------------------------------------------------------------|
|
|
| To: ADSM-L AT VM.MARIST DOT EDU
|
| cc:
|
| bcc:
|
| Subject: Re: Multiple fillings per node with collocate is group
|
>---------------------------------------------------------------------------------------------------------------|
From: "Andrew Raibeck" <storman AT US.IBM DOT COM>
> What is the storage pool's COLLOCATE value set to?
Collocate = Group
From: "Thorneycroft, Doug" <dthorneycroft AT LACSD DOT ORG>
>Are there also backups running during this period, If the diskpool doesn't
>have enough space, or a client has files that exceed the diskpools
>maxsize then some of your backups could be using the extra tapes.
Backups are running during the whole day, also during migration. And we
migrate (automaticly) during backup.
There is enough space, hi=50%, lo=10%, size=2400GB, we don't get above the
80% during the top backups.
There is no maxsize threshold defined.
From: "Volker Maibaum" <volker.maibaum AT EBERSPAECHER DOT COM>
> I don't think that TSM uses only 1 tape per collocgroup when you have
> set maxpr=3. What if only data of one collocgroup is on the diskpool?
> TSM would waste resources.
> If you turn collocation=none for the tape pool than TSM also uses three
> tapes at a time.
The doc's said that when a migrate starts, it selects a node or group to
migrate (with biggest FS or biggest node in case of colloc.groups) and does
the migration in 1 stream, so it "must" always use 1 tape at the time. (see
page 259 in 5.3 Admin Guide)
----------------------------------------------------------------------------
---
-----Original Message-----
From: Maurice van 't Loo
Sent: Tuesday, September 06, 2005 11:38 PM
To: ADSM-L AT VM.MARIST DOT EDU
Subject: Multiple fillings per node with collocate is group
Hi *SM'ers
After migration from diskpool to tapepool, we see every day that several
nodes or collocation groups have 2 or sometimes even 3 filling tapes.
As far as i know, migration is always 1 stream per node/collocgroup, so
there must be only 1 filling max.
Right?
Is this a bug, a feature or misunderstanding??
Tapepool: 3584 libr. with 5 drives
Migration: hi=50 lo=10 maxpr=3
Server 5.3.1.2 @ aix
Regards,
Maurice van 't Loo
For more information on Standard Life, visit our website
http://www.standardlife.co.uk/
The Standard Life Assurance Company, Standard Life House, 30 Lothian Road,
Edinburgh EH1 2DH, is registered in Scotland (No. SZ4) and is authorised
and regulated by the Financial Services Authority. Tel: 0131 225 2552 -
calls may be recorded or monitored. This confidential e-mail is for the
addressee only. If received in error, do not retain/copy/disclose it
without our consent and please return it to us. We virus scan and monitor
all e-mails but are not responsible for any damage caused by a virus or
alteration by a third party after it is sent.
|
