Does anyone else have this problem?

[droach]

I discovered that a lot of my Oracle and SQL servers had TDP backup "objects" stored on tape that were way past the company retention period. Some of the Oracle ones were due to the crosscheck not running properly...easy fix. Most of them seem to be due to special one-time backups and backups of long deleted databases. These objects stay "active" and are wasting tape space. I am trying to figure out how to auto-magically inactivate them.

To find these outdated object I use the following command:

select cast(object_id as char(15)), node_name, HL_NAME, ll_name, backup_date, state from backups where (node_name like '%%_TDP' or node_name like '%%_SQL') and backup_date < (current_timestamp - 36 days) order by node_name,backup_date​

and it returns object like this:

OBJECTID NODE_NAME HL_NAME LL_NAME BACKUP_DATE STATE
--------------- ------------------ ------------------ ------------------ ------------------ ------------------
838462739 MYSERVER9_SQL \WBS_m000\2013020- log 2013-02-02 ACTIVE_VERSION
2135702\000038F0\ 08:57:01.000000
​

This object is past my retention policy and should not be active. I know how to delete the object from the TSM database using the command 'delete object 0 838462739' command, and I know how to expose the 'Inactivate' tab in the TDP GUI and flag these objects as 'Inactive', but I am looking for a way to run a scheduled task on my TSM servers (or clients) that will mark any/all TDP objects inactive if they are older than my retention policy states.

I found this command that I can schedule to run on the clients:

tdpsqlc inactiv * * /olderthan=36
​

Before I go down this route I was curious how others are keeping their DBA's honest.

 

[tsm_dude]

Wow, deleting from the TSM DB thats something I dont see often. Most TDPs use a unique naming convention so TSM cant expire as everything is active. We get DBAs to delete in rman and use the tdposync. When I get desperate, I use the client dsmc and call the oracle node option file, then q backup /adsmorc/* -subdir=y
del backup /adsmorc/* -subdir=y -todate=04/30/2012
So SQL is also a non-expirer TDP, I didn`t know and will check mine - thx

 

[dyna]

Hi,

I've got a similar problem:

tsm: SRV>delete object 0 4892596756
Object 0.597629460 is not defined.
ANS8001I Return code 11.

TSM 5.5 ..I use the same select to get the obsolete objects, because the DBA cannot delete it via RMAN anymore.

Can you tell me what to do?

 

[droach]

Hi,

tsm: SRV>delete object 0 4892596756
Object 0.597629460 is not defined.
ANS8001I Return code 11.

This is strange. Your error message returns a different object ID than the object ID you entered. I don't know how that could happen. Does the delete command work for any of the objects you want to delete?

 

[dyna]

This is strange. Your error message returns a different object ID than the object ID you entered. I don't know how that could happen. Does the delete command work for any of the objects you want to delete?

Hi droach,

All delete commands for old oracle backups from this tsmnode give these errors.

I agreed with the DBA to delete the complete filelspace, so that would solve the problem ... for now.

Thank you for your response.

dyna

 

[tsm_dude]

If you delete via the client it will work and you wont mess up your db

 

[droach]

This has reached epedimic porportions

If you delete via the client it will work and you wont mess up your db

Ok folks...the number of objects that are not getting expired has reached epedimic porportions in my environment. TSM_DUDE, can you give me additional details of the commands you are using query and delete these backups? From a Windows server I cannot seem to get a list of the backups using the DSMC client. This is the commands I tried with no luck:

dsmc -optfile="C:\Program Files\tivoli\tsm\agentoba64\dsm.opt"
tsm> q b /adsmorc/*
ANS1107E Invalid option/value: '/ADSMORC/*'

​

I have tried several 'q b' commands and I cannot get it to return any TDP backups.

 

[krelvin]

We have the same problem for an RMAN process which for some reason loses track of files it should have deleted. they just keep stacking up until we manually remove them.

In 7.1 the command is different (Think it changed is TSM 6 though I never ran that version).

After we get the collection of objects that are past our retention periods... (I have to remove and compress data ...) I run the following Unix script to do the purge:

for i in `cat objects.txt`
do
  dsmadmc -id=ADMIN -pass=PASS -noc -dataonly=yes  delete object $i
done
echo "PROCESS COMPLETE"

Results in...

Object 40803472 and 0 dependent object(s) deleted.
Object 40803473 and 0 dependent object(s) deleted.
Object 40803474 and 0 dependent object(s) deleted.
Object 40810943 and 0 dependent object(s) deleted.
Object 40810949 and 0 dependent object(s) deleted.
Object 40810976 and 0 dependent object(s) deleted.
Object 40826920 and 0 dependent object(s) deleted.
Object 40826941 and 0 dependent object(s) deleted.
Object 40827201 and 0 dependent object(s) deleted.
Object 40827200 and 0 dependent object(s) deleted.

The difference being

delete object #####

vs the old way

delete object 0 #####

 

[TIYU]

Hi, i have de same problem...When i try to delete an object, TSM returns a diferent number of object and say that is not defined....HELP

adsm> delete object 0 4336006369
Object 0.41039073 is not defined.
ANS8001I Return code 11.