TSM Database growth - 2 GB daily?

[akilap]

Hi Wishes to all ,

I have a TSM server version running 5.3 and have around 200 clients registered to it. The present size of the tsm database is 210 GB and I can see a database growth of 1 % solidly a day (i.e) 2 to 3 GB a day alone.. We had a check over the SAP nodes exclusions , etc ,. but still not able to find which is causing this rapid growth.
IS there any way to find out the reason behind this faster growth ?
thanks and regards ..

 

[JonRoenick]

Are the clients backing up incrementally or are they set to always full backups?

 

[GregE]

Is expiration running? Even if its not, 1% a day is a pretty huge jump.

Any Windows servers bringing in millions of small files daily?

 

[akilap]

all the client tsm backups are set to incrmental , not full..

 

[akilap]

yeah . Expiration is running properly and how can i check whether any node is getting backed up a huge number of files?

 

[moon-buddy]

How many versions of files are you keeping: retain forever deleted files; set to no-limit, etc.?

I have 280 nodes of mixed apps: Windows, Novell, AIX, Linux, Oracle, Exchange, MS-SQL, etc. and my DB is only 67 GB. I keep Novell files for indefinite. Others have mixed 365, 180, and 90 days retentions.

 

[Stephan]

AKilap, do you have any archives running left and right? Archiving tons of files will definitively make your DB grow to no end...

Are you using the reporting? Check within to see your node's backups

 

[akilap]

No limit has not been set any of the nodes.for the normal baclients we have the retention set as 7,1,30,60 and for the SAP nodes we have 3,0,3,0 . and there are no archives set to run in a daily / weekly / monthly basis too.
please advice how can check this growth..

 

[akilap]

as i said there is no archives taken .. and im not able to get that reporting? what does that mean ? do u mean the daily reporting sent ?

 

[moon-buddy]

What platform is your TSM server on?

 

[akilap]

its AIX 5.3

 

[Canuck]

Try this to see which nodes & filespaces are using the most space:

select node_name, filespace_name, sum(num_files) as #Files from occupancy group by node_name,filespace_name order by #Files desc

 

[mateinone]

yeah I would be isolating your biggest nodes (by number of files as Canuck has pointed to, not by size of data).

After that I would again ask the question as to whether or not expirations is running cleanly.

ie.. It may be running cleanly, but you only run it for 5 hours and as you set a duration it.

I think you absolutely must get to the bottom of this ASAP. You are already in a bit of a no return situation unless you have a hell of a lot of physical resources handy.

What is your admin schedule (or batch file) that is running expirations?
What are your 10 largest nodes and to what backup schedule do they fit.

There are some applications out there that use literally 10's of millions of files and if you have just one of these suckers out there, it will come back to haunt you if not addressed.

run the script above look at the top 5-10 (in all honesty if it is a server issue it is likely to be just a small handful of boxes), if something stands out, then address it. Run the script the next day, run some sums and see what is happening.

 

[akilap]

hi ,
i can say the expiration is running properly .. please check the actlog output below.
>q act s='Expire Inventory' begind=-1
Date/Time Message
-------------------- ----------------------------------------------------------
02/23/09 15:00:20 ANR2750I Starting scheduled command EXPIREINV ( expire
inventory ). (SESSION: 306488)
02/23/09 15:00:20 ANR2017I Administrator IN061232 issued command: EXPIRE
INVENTORY (SESSION: 306488)
02/23/09 15:00:20 ANR0984I Process 44141 for EXPIRE INVENTORY started in the
BACKGROUND at 15:00:20. (SESSION: 306488, PROCESS: 44141)
02/23/09 19:06:57 ANR0987I Process 44141 for EXPIRE INVENTORY running in the
BACKGROUND processed 944784 items with a completion state
of SUCCESS at 19:06:57. (SESSION: 306488, PROCESS: 44141)
02/24/09 12:45:07 ANR2017I Administrator IN061232 issued command: QUERY
ACTLOG s="Expire Inventory" begind=-1 (SESSION: 308195)
02/24/09 12:59:24 ANR2017I Administrator IN061232 issued command: QUERY
ACTLOG s="Expire Inventory" begind=-1 (SESSION: 308195)

 

[akilap]

Database Size getting increased faster

Hi attached is the output of
"select node_name, filespace_name, sum(num_files) as #Files from occupancy group by node_name,filespace_name order by #Files desc"

 

[Canuck]

Also, take a look at which nodes backed up the most files in the last 24 hours:

SELECT ENTITY, SUM(AFFECTED) AS FILES_BKP FROM SUMMARY WHERE ACTIVITY='BACKUP' AND START_TIME>=CURRENT_TIMESTAMP-24 HOURS GROUP BY ENTITY ORDER BY FILES_BKP DESC

 

[mateinone]

yeah run that query without the filespace parameter and also do another query with just the sum of all files (so not grouped by node).