cancel
Showing results for 
Search instead for 
Did you mean: 

Performance issue BALDAT

Former Member
0 Kudos

Hi Gurus,

Last night, in our CRM system around 100 of users got stuck while accessing BALDAT table.In sm66,there was sequential reads for this table for around an hour or so and thus affecting performance.In sm13,several updates for the table was in initial status,

We manually cleared the updates,restarted work process,cleared application log,ran optimizer stats to clear the processes.

Could you please share what caused the issue and how to avoid it?In some forums its advised to deactivate the logging and could find notes for reorganizing the tables.What is recommended.Please suggest.

Thanks

Sabya

Accepted Solutions (1)

Accepted Solutions (1)

former_member307439
Discoverer
0 Kudos

Merry Xmas to all of you.

Hi Sabya,

Based on your earlier observations I have provided certain guidelines below. Please follow them to get more information to help identify the root-cause:


Last night, in our CRM system around 100 of users got stuck while accessing BALDAT table.In sm66,there was sequential reads for this table for around an hour or so and thus affecting performance.

=> 1. Since all the work processes are active in  performing read operations, check DBACOCKPIT/DB command level on what the DB sessions were waiting for? for example, db file read. If the sessions were waiting for a specific SQL access, tune the SQL statement.

=> 2. BALDAT is one of  the Basis administrative tables from which data volume could be reduced to maintain optimal performance, please refer to SAP Note 195157 'Application log: Deletion of logs' (part of 706478)

Also, after removing significant volume of data, it is suggested to perform table reorganization if you have deleted major chunk of data. Check the storage quality of the indexes and rebuild them periodically to avoid performance issues due to index fragmentation.

=> 3. Check whether all of the users that were stuck were performing same SAP transaction/program, and check on individual transaction performance from ST03/DB historical statistics

=> 4. Make sure that the application log deletion program does not run in period when there is heavy updates on BALDAT occur in the system: reschedule the reorg program to a later time accordingly

In sm13,several updates for the table was in initial status,

=> 5. Did you check if the Update server was active (SM14 or link from SM13 to Update Administration), if it was not active it would represent a core DB issue stopping all the Update activities in system. Analyze the DB logs (alert log in case of Oracle) for errors.

=> 6. If the Update server did not have any issues, this would mean a blockage of all Update work processes in certain operations -> please check what the processes were doing: if these were reading BALDAT , please work on point 1.

Regards,
Karthik

Answers (3)

Answers (3)

former_member182657
Active Contributor
0 Kudos

Hi Sabya,

Could you share SM21 logs for analysis.

Regards,

Gaurav

Former Member
0 Kudos

HI Sriram,

Sorry for late reply.Please find the attached dump,size of BALDAT table is 3 GB

Thanks

Sabya

Sriram2009
Active Contributor
0 Kudos

Hi Sabya

I think in your system your are getting more number of deadlock on the DB table BALDAT, during the time of time out dump, you can find the more number of locks in transaction code SM12. 

Could you refer the SAP Note 1839804 - Inconsistencies in application log.

BR

SS

Former Member
0 Kudos

Hi Sriram,

When this issue happened, in sm66 there were lot of sequential reads for table BALDAT and strangely in sm12 i couldnt see any locks for this table.

Could you please explain this?

Tx

Sabya

Sriram2009
Active Contributor
0 Kudos

Hi Sabya

1. When this issue happened, your system will goes to deadlock status(Lock entries table getting full), because of this you are getting more Time out dumps in ST22.

2. Is any one accessing the Database level to read/write / modification ?

BR

SS

Sriram2009
Active Contributor
0 Kudos

Hi Sabya

1.  Is this any dumps? If it is yes pls share the full dump

2. Could you check the SM21 system log any errors ?

BR

SS

Former Member
0 Kudos

Hi SS,

We do get lots of TIME OUT dumps.

Tx

Sriram2009
Active Contributor
0 Kudos

Hi

Thanks for your info

Could you share the Time out full dump details as attach or paste which appeared very beginning

BR

former_member188883
Active Contributor
0 Kudos

What is the size of BALDAT in your system ?