How does a node send data to a disk pool?


ADSM.ORG Senior Member
Oct 30, 2019
Reaction score
A fundamental question here. A node has a policy domain name listed for one of the parameters, obviously. How does TSM figure out which disk storage pool to send the backup client's data to?

We have multiple disk pools. Everything is working correctly. I'm just trying to figure out why. I set this up so far back now that I've been unable to track this down. Sheesh! Each disk storage pool has the Next Storage Pool parameter set to a primary tape pool, and that all makes sense.

I've been looking at everything (policy domains, policy sets, management classes, storage pools ... you name it), I can't seem to link it all. I thought I'd found the answer when I looked at the management classes, and I could see the parameter Migration Destination was set to the respective disk pool but alas, according to the IBM documentation, this only applies to Space Management clients and is not used for backup-archive clients. We're using backup-archive clients. Hmm ...
It's in the destination for the copygroup for that management class. Q CO * F=D
WOW! ... WOW! Thank you :)

I even went through `show config`, but that, of course, doesn't report the details. Even if it had occurred to me to just loop a dsmadmc command through each of the basic contructs (domain, policy, mgmtclass, stgpool, copygroup), I would never have thought to use f=d with copygroup, so it still would have alluded me.

By the way:

1. What does it do if the Copy Destination is not provided?

2. Any guesses why they called it copygroup? The name doesn't seem intuitive to what it achieves. Seems like retentionvalues or some such name would have made more sense. Maybe I'm missing something obvious here.
1. What does it do if the Copy Destination is not provided?
The define copygroup command fails:

Protect: SERVER1>define copygroup TEST stantard standard standard
ANR1572E DEFINE COPYGROUP: Missing copy destination.
ANS8001I Return code 3.

2. Any guesses why they called it copygroup?
Your guess is as good as mine, I have only been working with Spectrum Protect for 18 years and it started in the late 80s / early 90s. It started on the mainframe first before being expanded to distributed systems. Everything is cryptic on the mainframe, so maybe it made sense there. I can see how it's confusing when learning. Once you know the terms, it doesn't really matter as long as you know what controls what.
Okay, I appreciate the feedback. The reason I asked about the scenario wherein the destination is not provided is because in the documentation, they state:

"Specifies the primary storage pool where the server initially stores backup data. This parameter is optional. You cannot specify a copy storage pool as the destination."

It turns out, though, that I was looking at 'update copy pool', not 'define copy pool', so obviously it's optional in that case. I see that the documentation for "define" clearly states that it's required. I must have become confused at some point after numbing my brain with all the searching and combing through everything in efforts to track down the culprit. That's when I decided to head to this forum and stop wasting time. Ha.šŸ˜Š