TSM database getting 100% full occasionally

[khurram]

Guys,

I need your help in resolving one issue ... One of my TSM environment is facing one problem which is that its TSM database ( which is now around 14 GB) getting filled all the time ..

Total size of backups being done is around 700 GB....with archives and oracle backups as well.

I have asked oracle dba to delete old backups , at the same time ensured that expiration should run daily... not through expire inventory but with expire inventory intreval set to 24 hours...

What else i can do to resolve the issue? already have expansion triggers defined!!

I beleive DB backups have a direct impact on DB getting fulled or not?so should i ensure daily db backups happen all the time , will it help ?

Also i have put TSM recovery log not so big , it is around 1 GB? Should i increase it as well?

Please guide!!!


many thanks

Regards

 

[moon-buddy]

Just increase both DB and LOG volume sizes.

 

[sureshg]

Hi,


My name is suresh I need help I am in middle of the project I am installed TSM 5.5 serveron windows 2003 server and BA client in linux platform and configured dsm.sys and dsm.opt

Dsm.sys:

Servername xxxxxxx
Commethod
Tcpipport
Serveraddress
Nodename
Schedmode
Errorlogname
Schedlogname
Passwordaccess

Dsm.opt:

Servername xxxxxxxx

I hope to start the client above configuration is sufficient

When I start the BA Client in command line (dsmc) I can login, I can take backup and even restoration happens fine but when I start Client GUI (dsmj) complete window opens
But, 1) when I click utilities menu or any other menu like Files, edit, help no action performed. 2) when I click on backup icon ( not menu) backup window opens completely but in the left panel Directories are not visible like:

+ home
+ opt
+ Tivoli
+ tmp
+ usr

I can only see the

+
+
+
+

If i click any of the (+) symbol I can see another(+) symbol , but when I click immediate next to (+) symbol I can see the files under that directory in left panel but directory is not visible, I tried uninstalling and reinstalled the BA client but it’s the same issue

Please take little time to read this mail it is very urgent, I am stuck

Regards
Suresh

 

[khurram]

Guys!!!

Now my issue has become worsen..

After i asked my Oracle DBA to delete his old backups through rman, now utilization of TSM DB has dropped too much.

Now out of 19 GB TSM DB , only 1.8% is used... however i can not reduce the size of TSM DB, as maximum reduction is only 500 MB!!

It seems that TSM DB is fragmented severely , but how i can be sure? Morever, how i can make sure about this my assumption. Now as TSM DB copies are in sync and everything seems fine to me , but i am worried about TSM performance and internal fragmention?

How i can solve this problem .... my ISC health monitor is also showing critical state for TSM DB but no reason


Please help!!!!!!!!!!

 

[hogmaster]

TSM database getting 100% full occasionally Reply to Thread

pls post outout from
q dbvol format=detail
and
q db format=detail

 

[Stephan]

Take a look here for some information about TSM DB defrag...

http://www.lascon.co.uk/d005106.htm#dbdefrag

I wish my TSM DB would drop too much...

 

[mikeatkc]

I need your help in resolving one issue ... One of my TSM environment is facing one problem which is that its TSM database ( which is now around 14 GB) getting filled all the time ..

The maximum size for TSM database is 530GB so using 14GB is really nothing. Simply extend your database file to larger than 14GB. Let us know what TSM version you are using. If you are using 5.5, go into ISC and select any TSM options...

-> Server Properties -> Database and Log -> Database -> Extend/Reduce

Just make certain you have the space to increase your database file or create a space trigger for your database file (again make certain you have the space/path ready).

 

[khurram]

Guys,

Thanks for help... but i am still suffering!!

It now reached up to 32 GB... Now i have decided to reduce number of days for retention archive versions of one of bigger nodes and run expiry.. Archive objects get deleted , and i am still assuming the size of DB will decrease a little bit.

At same time , now i have decided to delete some unnecessary UNIX filespaces for UNIX nodes ( like /opt and /tmp).Will it help in reducing size of TSM DB itself?

I have noticed down that one of AIX server ( for which i am running different kind of incremental backups for some critical filesystems .. i dont have sysback) , number of files through "q occupancy " is very large..

Not sure about selective backup however, how the retention policies apply to backups done through selective backup strategy? By the way i am doing slective backup of one critical directory of same AIX server ? How the expire inventory command treat backups done with selective backup command?

 

[mikeatkc]

Extend the database (increase the size) also, review to see if you have expiration process running daily, expiration will reduce the db size or maintain its current state, etc.

 

[BBB]

Oracle backups will use hardly any space in your DB at all.

Looking for the system/filesystem with the highest number of files (note: files, not gb) in "q occ" as you have done will tell you where all the growth is coming from. Most of the space used in the TSM db is related to the number of files you see listed in "q occ".

So pick the biggest couple and target those. You probably have a funny rentention policy set, or have a system doing something odd with a very large number of files. Post the output of "q copy * active".

 

[MatthewWarren]

If you are dealing with large numbers of small files then set your TXNGROUPMAX and TXNBYTELIMIT values high for the clients involved. This will kick in file aggregation within TSM and help to reduce the number of DB entries TSM creates to track the backed up files.

 

[jcyanquen]

Ask if you DBA are running "DELETE OBSOLETES"? and let TDP node "Delete backup data"

 

[picay]

About tdp for Oracle:
- the node must be configured with backdelete=y
so that rman can delete obsolete rman backups
- the rman script MUST contain the ad-hoc rman instruction to delete those obsolete backups. Why ? Like that you ensure that no unutil backups objects stay on tsm server (with rman/tdpo backups, expiration does not delete obsoletes objects...)
- about your database fragmentation, following select give an indication about fragmentation pct. It is just an indication:

SELECT CAST((100 - (CAST(MAX_REDUCTION_MB AS FLOAT) * 256 ) / (CAST(USABLE_PAGES AS FLOAT) - CAST(USED_PAGES AS FLOAT) ) * 100) AS DECIMAL(4,2)) AS PERCENT_FRAG FROM DB

If you think that the fragmentation is to high, get more information with a:
estimate dbreorgstats
Take care that it is a heavy process, resource consuming, and can be long
If you wish to reorg database, you just have to unload/load the database, with care, db backup first...
With a db of about 30GB, it will no take too much time but I can't tell you how many time