cancel
Showing results for 
Search instead for 
Did you mean: 

SAP Freeze for 30 minutes

Former Member
0 Kudos

Dear gurus,

In our SAP production environment, SAP frequently hold for some minutes, ranging from 10 to 30 minutes.

When that happens, no users can logged onto SAP System (Waiting for system response in SAP Logon).

When that happen, I can't do anything except restarting SAP Server.

We rarely do an update record nor an intensive report program when that happened,

Where can I check the initial problem for this?

For information,

Our system running on AIX 5.2 Oracle 10.0.2.0 with 16GB Memory (SAP ECC 6.0).

The Detail:

Buffer size:

ABAP Programs: 900000 KB

Nametab: 89996 KB

Generic key table: 97657 KB

Single key table: 90000 KB

CUA information: 10000 KB

Screens: 19532 KB

No of Work Process:

Dialog: 25

Update (V1): 6

Update (V2): 3

Enqueue: 1

Background: 3

Spool: 4

Initial / max. swap requirement: 11154 MB / 13062 MB.

Even when the users is only 5, the memory could drop to only 200MB in the server (as in ST06).

I do not know what consume the high memory.

Also, when I go to DB02, an see the space, it shows blank record.

That is, the used space, total DB size, is 0. I have checked that SAPOSCOL is running.

Thanks for gurus help.

Regards,

Accepted Solutions (1)

Accepted Solutions (1)

Former Member
0 Kudos

Hi,

Check log file like dev_w0,dev_disp,alertsid.log.

Might be some dialog work process goes in PRIV mode.

give log file then suggest

SurendraJain

Answers (3)

Answers (3)

Former Member
0 Kudos

if u have Seprate Application servers then install Load Balancing the on ur all the R# production Servers. i hope this will happen due to excess no of users. also check the Background jobs as well as set the time out parameter also. so useless sessions are termminated automatically.

Thanks

presu

Former Member
0 Kudos

Hi Bobby,

Have you analysed from the network point of view ? We faced similar problem with our development server and after thorough analysis we came to know that the problem was with the network cable connecting to the main switch.

On seeing your WP table, I wonder on why you have only three WP's for BGD while having allocated 25 WP's for DIA and 4 for SPO. Can you increase the BGD Wp to atleast 6 as the BGD WP's are needed for the reports to run in the background.

The UNDO tablespace will be utilised only when the data is being deleted from the tablespace. So no need to resize the UNDO tablespace at this point of time as the same might be needed sometime later. For a PRD environment it is better to have the auto extend turned off for all the tablespaces except UNDo.

Regards,

Varadhu

Edited by: Varadhu on Jul 27, 2009 12:44 PM

anindya_bose
Active Contributor
0 Kudos

Hi Bobby

There could be more than one issues in your system.

> no users can logged onto SAP System (Waiting for system response in SAP Logon).

When this happens, check ORAARCH directory under SAPDATA_HOME. If this directory is full, database goes to dormat state and until and unless this directory is free ( free space is there) database wont allow any user to get it.

What is the frequency of your archive log job? Check https://service.sap.com/sap/support/notes/391

>Even when the users is only 5, the memory could drop to only 200MB in the server

Could you find any reason of this high memory consumption from OS level, you can check the PID from there and then from sm50 you can check which program is using maximum meory. Also check from ST02 what is the number of swapping for important buffers , like program buffer. What are the values of "Commit charge limit Kb" and "Commit charge free Kb" from ST06 when you see this memory bottleneck?

>Also, when I go to DB02, an see the space, it shows blank record.

Run RSORAT0D report and then check again. Clickking Rfresh on the DB02 first screen will run this report for you or else you can run it from SA38 or se38 directly

Former Member
0 Kudos

Thanks gurus,

I'll check the corresponding log and report back what I've found.

Thanks for the breakdown and the pre-solution.

Gurus, I also found that the PSAPUNDO is over allocated.

The current size is 92.660 MB, with usage 1%.

Can I resize that? Resize is using a BRTOOLS, am I right?

Best Regards,

Former Member
0 Kudos

Hi bobby,

You can resize PSAPUNDO via brools .Increase PSAPUNDO but it should not autoextend ON.

PSAPUNDO not need to decrease while its usage only 1%.

User are able to logon after sometime it means oraarch not a probelm because you have to delete archive file/oraarch backup with -cds option then user can only logon (but your not deleting)

SurendraJain

Former Member
0 Kudos

Dear SurendraJain,

Sorry I can't understand very well.

We're not expecting to increase the PSAPUNDO, but to reduce it.

And about ORAARCH, do you mean it's not the problem in our system?

Then can you kindly tell me another possible problem which cause our system to freeze for some times?

Thanks for your help.

Best Regards,

Former Member
0 Kudos

By decreasing PSAPUNDO no benifit .increase the value PSAPUNDO if it commit/rollback in wait state

SELECT NAME, XACTS "ACTIVE TRANSACTIONS"

FROM V$ROLLNAME, V$ROLLSTAT

WHERE STATUS = 'PENDING OFFLINE'

AND V$ROLLNAME.USN = V$ROLLSTAT.USN;

(Jst increase nothing any wrong)

Check Dialog mode status ,update status,exclusive lock during slow as suggested in previous.

check two suggestion given by anindy .

SurendraJain