Recovering Tape Volumes with Expired Data on an "Archive Only" Server

buster99

ADSM.ORG Member
Joined
May 14, 2008
Messages
148
Reaction score
1
Points
0
PREDATAR Control23

Hello, I'm having an issue recovering tape volumes that are expired from an "archive only" TSM server. The server stopped being used for active data in 2013. Our highest retention setting for the data on the "old" TSM is 7 years. I used a select statement to sort the volumes by "last write date" and found quite a few tapes that haven't been used in over 7 years in the onsite and offsite stgpools.

Can a select statement like this be considered the most accurate way of determining if a tape volume has data on it that is older than the expected retention setting of 7 years ? I see tape volumes with "last write dates" older than 7 years ago.

select volume_name,Pct_Utilized,last_write_date,DEVCLASS_NAME from volumes order by LAST_WRITE_DATE asc


I have included the 2 Delete Volhist and Expire Inventory scripts below.
The 2 Delete scripts are not active. The "todate" on each of those worries me. How would the "-180" work against our 7 year retention policy if I activate these ?


Schedule Name: DELETE_VOLHIST_DBB
Description: Expire old TSM DB backups
Command: delete volhist t=dbb todate=today-180
Priority: 5
Start Date/Time: 12/12/02 11:35:00
Duration: 1 Hour(s)
Schedule Style: Classic
Period: 1 Day(s)
Day of Week: Any
Month:
Day of Month:
Week of Month:
Expiration:
Active?: No


Schedule Name: DELETE_VOL_HIST
Description: Start volume history deletion
Command: delete volhist t=all todate=today-180
Priority: 5
Start Date/Time: 12/12/02 11:30:00
Duration: 1 Hour(s)
Schedule Style: Classic
Period: 1 Day(s)
Day of Week: Any
Month:
Day of Month:
Week of Month:
Expiration:
Active?: No


Schedule Name: EXPIRE_INVENTORY
Description: Start client file expiration
Command: expire inventory quiet=yes
Priority: 5
Start Date/Time: 02/05/08 16:50:00
Duration: 1 Hour(s)
Schedule Style: Classic
Period: 1 Day(s)
Day of Week: Any
Month:
Day of Month:
Week of Month:
Expiration:
Active?: Yes
 
PREDATAR Control23

You need a TSM DB backup and volhist file (or use a RPF from that time) from more or less the same time frame.

Restore this DB backup and you can get the data back.
 
PREDATAR Control23

Hello moon-buddy, thank you for the reply. I'm actually not looking to recover the data, just looking to reclaim the tape volumes that haven't been written to in over 7 years which is our longest retention period. I think I'm missing something to expire/reclaim the tapes.
The TSM server itself is inactive and only used for restores a this time.

Thank you
 
PREDATAR Control23

Can a select statement like this be considered the most accurate way of determining if a tape volume has data on it that is older than the expected retention setting of 7 years ?
You are missing one thing. Expiration is looking at inactive files older than the retention. If the files are active, they can't expire. Example:

You have a node that is 10 years old. 8 years ago, you backuped up /home/johndoe/file.txt
Today, if you go on that node and /home/johndoe/file.txt still exists and is 8 years old, then that version is still active on the server. So it can't expire.
 
PREDATAR Control23

Thank you for the reply. So, something is "broken" on the TSM server that has been sunset that tells expired data to turn "inactive" ? Maybe that data doesn't have a set retention period ? Thanks
 
PREDATAR Control23

So, something is "broken" on the TSM server that has been sunset that tells expired data to turn "inactive" ?
Objects are marked inactive in one of two case:
- you backup a new version of an existing file, the most current is active, the previous one becomes inactive
- after deleting a file on the machine, you run an incremental backup, the client will mark that file inactive

So the most current version is always active while it still exists on the client. If you have a really old node, you can have 10 year old data that is active if that file still exist or existed on the client at the time the last backup was taken.

Look at one of those tapes to see which nodes has files on it:
q content {volume_name}

Then look at the backups table to see what's the oldest active object for that node:
select * from backups where node_name='NODENAME' and state='ACTIVE_VERSION' order by backup_date desc
 
PREDATAR Control23

Marclant, Thank you for the reply. I did the "select" statement you provided on one of my volumes in question and it show active data from as far back as 2005. I definitely don't need to retain that data anymore. Would a "move data" on the volume or a "reclamation" on the stgpool help me to clear old data off of theses volumes and TSM history ? Our oldest retention setting is 7 years. Both of these options would be very time consuming as this TSM server itself is inactive.

Can I delete the tape volume and discard the data or is that a poor way of handling the issue within TSM ? Would the tape become a scratch tape onsite and automatically return from offsite if I did this ?

Thank you
 
PREDATAR Control23

First, is that node still active?
If not and you no longer want the data for that node, use the DECOMMISSION NODE command or delete the filespaces for that node, then delete the node. Reclamation will take care of the rest once the filespaces are deleted.

If the node is still active, browse to see if those files still exist. If they still exist and you were to delete that volume, they would get backed up again during the next backup. Active data never expire because it's active, meaning it still exist on the node.

You can use move data or reclamation to free up the tape, but the data will just get moved, it will not delete it unless you explicitly delete it.
Our oldest retention setting is 7 years.
That applies to inactive data only, active data never expires because it is active, meaning it still existed on that node at the time of the last backup.
 
PREDATAR Control23

Thanks again. Yes, the node is "active" but on a new TSM instance. No backups are done on this server instance anymore. We upgraded to a new TSM server in 2012, so this data exists for restore purposes only. So, while there are backups of a node that are within our 7 year retention period, some tape volumes haven't been written to in over 7 years. I would like to remove those volumes from onsite/offsite storage and TSM to reuse them, if possible. Is there a "quick" way to make that update to tape volumes ? If I know that data on a tape volume is not needed, would you recommend doing a "delete volume ###### discarddata=yes" command ? Would TSM reclaim the volume automatically ?

I think I will need to do a separate project to decom nodes that haven't been active for over 7 years. I have quite a few on this old TSM instance. Plus, that TSM version is pretty old...5.5.5 I believe. Not sure "decom node" command would work there.

Thank you
 
PREDATAR Control23

Yes, the node is "active" but on a new TSM instance. No backups are done on this server instance anymore.
So it really is inactive then.
If I know that data on a tape volume is not needed, would you recommend doing a "delete volume ###### discarddata=yes" command ?
You can do it, but I don't recommend it. I'd recommend to delete the filespaces for all the nodes that are moved to a new server that you no longer need, and delete the nodes first. This will free up space both in the primary and copy pools. If you delete volumes here and there, you risk missing some data for these old nodes and even greater risk of deleting good data for other active nodes.

You will likely gain more space if you delete the unneeded filespaces because they likely have data on tapes that have been written less than 7 years ago, so you are missing those from your list of old tapes. And at the same time, it will clean-up that instance.

If you do go ahead with it, at least check to see all the nodes that are on the tape first to see if you really want to discard their data:
select distinct node_name from contents where volume_name='123456L4'
Would TSM reclaim the volume automatically ?
Yes.
Plus, that TSM version is pretty old...5.5.5 I believe. Not sure "decom node" command would work there.
You are right, decom node doesn't exist, but you can delete the filespaces and then the node.
 
PREDATAR Control23

Thank you all for the replies, information and suggestions. Looks like I'm going to be doing a project on our retired TSM server to remove unnecessary nodes and hopefully recover old tape volumes...for a while.

Thank again
 
PREDATAR Control23

I'm back ! Another question about handling archived data and volumes. So now I'm pecking away at data on our archived TSM servers that we no longer back up and is now past its retention period. I'm doing "delete filespace" and "remove node" on unnecessary data. I have thousands of tapes stored onsite that are not in the tape library along with the offsite tapes. I have my Monday morning process that tells me which tapes to pull back from offsite because they are "empty" now. How do I determine which onsite tapes are now available for me to reclaim. If tape 001234L4 is sitting in a cabinet and all of its data has expired, how do I know that it is free to re-use ? I looked for something in the activity log to show me onsite tapes that are free now but I haven't had any luck like I did with tapes that are being recalled from offsite.

Thank you
 
PREDATAR Control23

If they are not listed in Q VOL and not in the volhist, they are free to use. Or if listed in Q VOL as empty.
 
PREDATAR Control23

Try running the reclamation of the primary tape pool, it may tell you that certain tapes are needed for reclamation.
 
PREDATAR Control23

I ran "q volhist"...would a volume listed as "STGDELETE" be considered empty and be able to be loaded back in the tape library to my new TSM instance ?


Volume Type: STGDELETE

STGDelete
Specifies only records that contain information about deleted sequential storage pool volumes.


SAMPLE VOLUME -

tsm: TSM18>q vol 012060L4 f=d


ANR2034E QUERY VOLUME: No match found using this criteria.

ANS8001I Return code 11.

tsm: TSM18>q drm 012060L4

ANR2034E QUERY DRMEDIA: No match found using this criteria.

ANS8001I Return code 11.

tsm: TSM18>q con 012060L4

ANR2401E QUERY CONTENT: Volume 012060L4 is not defined in a storage pool.

ANS8001I Return code 11.
 
PREDATAR Control23

Thanks again. Now...I was going to ask how do I get rid of that old volhist record for these volumes so they don't exist on our old and new TSM instances but now I see that "delete volhist" is not active on the old server. So, I was going to ctivate itt but I see that whoever created it had it looking back 180 days. We have a retention period of 7 years for our production data...is it a bad idea to run it as it is written now ? I'm thinking about needing that tape info for restores and a possible DR recovery.


Schedule Name: DELETE_VOL_HIST

Description: Start volume history deletion
Command: delete volhist t=all todate=today-180
 
Top