Expiration takes to long time?!

[Suslik]

Hi there! I have a problem with my expire inventory. My environemnt consist of 20 tsm server and on 2 server the expire inventory takes about 5 - 10 hours. We are using this server for our FS backups Linux/AIX and for our DB backups SAP (Oracle, DB2, Sybase, Maxdb). The other tsm server are runnig fine, about 60 - 90 min for the expire inventory. I check the amount of nodes, examined objects of the Servers an they are almost identitical. Do you have an advice how to find the problem?

 

[Suslik]

I tryed the following sql statement, to check the performance obj/min:


SELECT - DATE(start_time) AS date, - TIME(start_time) AS start, - TRANSLATE('ab:cd:ef', DIGITS(end_time-start_time), '________abcdefg_____',' ') AS duration, - affected, - successful, - CAST(FLOAT(affected)/(timestampdiff(2,CHAR(TIMESTAMP(end_time) - (start_time))) / 60) AS DEC(20)) AS obj_min - FROM - summary - WHERE - entity is null - AND activity='EXPIRATION' - AND start_time>current_timestamp-7 DAY

Result:
DATE START DURATION AFFECTED SUCCESSFUL OBJ_MIN
----------- --------- --------- --------------------- ----------- -----------------------
2014-01-22 10:46:06 06:06:45 452979 YES 1237
2014-01-23 10:55:11 05:59:46 310544 YES 865
2014-01-24 10:49:54 06:12:56 67585 YES 181
2014-01-25 11:16:35 08:03:51 977225 YES 2023
2014-01-26 09:29:46 07:53:38 599688 YES 1267
2014-01-27 09:56:02 07:12:44 371985 YES 861

 

[marclant]

Hi,

Got a few questions:
Is your server V5 or V6?
How is your database layed out on disk?
How many cores?
How many expiration processes running?

 

[Suslik]

Hi,

Got a few questions:
Is your server V5 or V6?
How is your database layed out on disk?
How many cores?
How many expiration processes running?

Our TSM Server is V6, Power7 with 4 cores. In the moment, we have only one dbspace location for our db, just like the other server. I know, this is nocht the best practice and we are working on this construtction. We are starting the expiration with the default, 4 resource.

 

[SniperKing]

Hi,

Do you have other admin job / client task running at the same time with the expiration (e.g. reclaim, migration, backup stg.....etc). ?
Are the two TSM instance DB sharing the same physical disk? raid level?

 

[Suslik]

Hi, there is no other job/process running in this time. Sry, I use a not-quit-correct term. No, every TSM Server has there own disk. Our raid level in our storrage sub is 5.

 

[marclant]

Expiration is very database intensive and the component that affects the database performance the most is at the disks. More disks and faster disks will always help. If you are going to use Raid, 10 is recommended over 5. Also, a minimum of 4 database containers for small systems.

I recommend you review this section of the performance tuning guide regarding the database: http://ibm.co/1aHiVZf

Also this section on solving performance problems: http://ibm.co/1aHjjH1

 

[Suslik]

Ok, this is a point. I will talk to our storrage team and give an answer.

 

[Hiten]

Hi,
I am facing bit same issue as I have 20 Windows flat backup server along with 3 SAP R3 nodes whose backup running in TSM which is on OS windows 2008 x64bit. Total nodes are like 25 but while expire inventory process running It shows only one node data expiration & also running almost for 24hrs. I could not understand the same.

 

[Suslik]

Hi,

it was a realy procedure to solve this problem, in our case. Only 2nd lvl support was able to handle this. From offline dbreorg to change TZ till many tests, I took about 1,5 months to handle our expiration time. I was realy suprised, how "expire inventory" cooperates with the OS, and how less it has to do with I/O Performance on the storage, unless your TSM db lies on an USB Device...

In your case, on which node hungs your expiration?

 

[Hiten]

Hi,
In my case expiration is not set node specific. We have set administrative schedule with expiration time to 120minutes. I am having TSM 6.4 database in which we have found major issue with DB2 permission level issue which brought in to notice by sr. Tivoli expert. He discovered that
1 TSM server service is running with ID which is having ADS authentication.
2 TSM DB2 database is running with the windows local system authentication.
He is recommended to use single id either locally created on windows server level or domain power user id which is not bind with any security restrictions. So we have done following
1 created local id at windows level & added the same in local ADMINISTRATOR,DBUSERS & DBADMINS groups.
2 Open DB2 Control center & logged in with ID authorized to login & checked the permissions. Later added the local id created with required permissions to act as db admin & security admin permissions.
3 In Services.msc changed the TSM server service authentication assigned to newly created id.
4 Restarted the server at OS level & monitored the DB2 & TSM server service which is found running.
This solution is given in below link provided by IBM.
http://www-01.ibm.com/support/docview.wss?uid=swg21445245

Now we have stopped the administration schedule now & found the automatic expiration is running with duration of 24 hours which is default mentioned in server opt file. Now it shows all 33 nodes with large amount of files expired.

 

[rajeshyeleti]

Hi all,

I need a select command for getting node names which are getting failed from past n number of days on a particular sever.