cancel
Showing results for 
Search instead for 
Did you mean: 

EHP5 Upgrade MAIN_NEWBAS/TABIM_UPG (Downtime) phase long runtime

Former Member
0 Kudos

Dear Experts,

We are in process of applying EHP5 for our freshly installed ECC6.0 system (Win2008R2 and Oracle 11.g - MSCS).

The upgrade is currently in the Downtime phase MAIN_NEWBAS/TABIM_UPG and is running for past 3 days. The log files are not getting updated and there is no mention of error in any files that were last updated. Besides the sever has 48 GB RAM.

Where would I get any possible lead to check for the problem.

Thank You.

Best Regards,

Prashant P.

Accepted Solutions (0)

Answers (3)

Answers (3)

nicholas_chang
Active Contributor

Hi,

Could you please check is there any ora, R3*, tp process running in task manager, and perf monitor?

Cheers,

Thanks,

Nicholas Chang

Former Member
0 Kudos

Hi Nicholas,

Yes, there is an Oracle, TP and 3 R3trans process seen running in the Task monitor.

Thanks

Best Regards,

Prashant P

nicholas_chang
Active Contributor
0 Kudos

Hi,

Try to check SLOG, ALOG and ULOG and see what packages is runing..

How many parallel processes you've configured? How many CPU cores your system has?

Cheers,

NIcholas Chang

Former Member
0 Kudos

Hello Aidan & Nicholas,

Thank You for your valuable inputs.

I had escalated this issue to SAP and the issue is rectified now.

Solution was to set oracle parameter usezero_copy_io=false according to SAP Note 1635605.

Cheers!!

Best Regards,

Prashant P.

nicholas_chang
Active Contributor
0 Kudos

Hi Prashant P.

Thanks for posting the solution here! and we've learned something!

Cheers,

Nicholas Chang

Former Member
0 Kudos

Hi,

Please Kill TP and R3Trans process using task manager and restart SUM tool and repeat the step.

Regards

MIH

former_member264034
Active Contributor
0 Kudos

Hi,

During phase TABIM_UPG(during downtime) which will import entries

into control tables, it normally requires a long runtime.

Performance in phase TABIM_UPG is mostly driven by

database performance.

Also, by the number of parallel R3trans processes configured: see note

number [1127194|http://service.sap.com/sap/support/notes/1127194].

Optimize the number of import and batch processes:

Parallel import processes (R3trans)

The parameter "parallel import processes" determines the number of

R3trans processes used during upgrade. The number of R3trans processes

mainly has an impact on the TP* phases. As a rule of thumb, set this

parameter according to the following rules:

512 MB per R3trans

Number of CPUs x 1.3

Background processes (Batch)

The parameter "background processes" determines the number of background

processes used during the upgrade. This parameter has a high impact on

the upgrade phases PARCONV_UPG, PCON* and ACT*. Set this parameter

according to the following rules:

Number of CPUs x 1.3 up to 1.5,

Bad performance in TABIM_UPG may be too a result of

- oracle patch levels

- oracle parameter settings

- database statistics

- different versions of import tools used (tp/R3trans)

Make sure all relevant oracle parameters according to note

[1431798 |http://service.sap.com/sap/support/notes/1431798]Oracle 11.2.0: Database Parameter Settings

are implemented (especially important are "fix", event and "underscore"

parameters).

The script from note

1171650 Automated Oracle DB parameter check

will help you to identify missing/wrong paramters.

Please make sure too the datase statistics are up-to-date *and*

oracle system- and dictionary statistics are created correctly.

[838725 |http://service.sap.com/sap/support/notes/838725]Oracle dictionary statistics and system statistics

One more thing that can affect the performance:

The R3trans and tp processes in TABIM_UPG writes intensively to

DIR_PUT/tmp and DIR_PUT/log. If this drive is NFS mounted it may

cause some delay too.

Regards,

Aidan