Remove data from an active datapool

[manofmilk]

Hi all,

Say we had an active data pool which then became full so we stopped it being used by one of our primary storage pools by doing this:

update stg XXX activedata=""

Is this the correct way to do this and then if it is at what point does the data that came from that storage pool get removed from the active data pool?

Cheers.

 

[chad_small]

Not sure if TSM automaticlly invalidates the data. I would think as the data within it is no longer considered active then TSM would allow it to be overwritten.

 

[rore]

update stg XXX activedata="" is correct. And yes, as data becomes inactive it will be expired from your active data pool. Reclamation then will free space.

 

[manofmilk]

Thanks. We ran the command (update stg XXX activedata="") a few days ago and so we can see that the data is indeed expiring in line with the live data in the primary storage pool.

Problem is that this is a bit of a slow process and really we want to replace some of this data in the active pool with higher priority stuff: is there any way to delete the data from the active pool quicker than this (preferably without scrubbing the entire active pool and re-populating it)?

 

[rore]

When you say "higher priority stuff" you mean active data from other systems?
If you delete your active files from your client your data will be gone from your primary pools too. So, I don't see a way to delete faster the data other than delete the volumes (discard=yes). Also, if you wait for expiration the stgpool will never be empty (assuming that you backup some data that doesn't change, like for example some OS and app files).

 

[terlisimo]

Hi all,

Say we had an active data pool which then became full so we stopped it being used by one of our primary storage pools by doing this:

update stg XXX activedata=""

Is this the correct way to do this and then if it is at what point does the data that came from that storage pool get removed from the active data pool?

Cheers.

Yes... first update your stgpools not to use that active data stgpool anymore. Don't forget to remove it from mgmt class if you have it defined there for simultaneous write.

But finally you will have to manually delete all the volumes from the storage pool.

What I did:

1) parse output of "query volume stgpool=mypool" to get a list of volumes
2) create a shell script that does "delete volume VOLNAME discarddata=yes" for all of the volumes.

I had 12TB copy storage pool with 10GB volumes (around 1000), it took around 10 hours to delete all the volumes (because TSM needs to remove entry for all the objects stored there)


Note: your database size will probably not decrease, even if you do a backup/restore db because DB2 does not automatically reclaim unused pages for user tablespaces. Well maybe it would at some point but it really bugged me so I did a REORG on all tables by hand followed by LOWER HIGH WATER MARK and REDUCE MAX for relevant tablespaces. Not sure if the order of commands is right, the damned thing decreased in size only after i ran all of this twice.

 

[terlisimo]

Don't forget to remove it from mgmt class if you have it defined there for simultaneous write.

I meant domain, not management class.