Message from Database Analyser about collision rate of LOGQUEUE

[Former Member]

Approx. several a day I get the following messages :

===== #218 at 2009-10-07 01:58:36

• I SQL commands executed: 1160522

CON: PureSQL_CmdsLiveCache > INTERVAL * 5

VAL: 1160522 > 908 * 5

• I DB procedures called: 25953

CON: CallCnt > INTERVAL * 10

VAL: 25953 > 908 * 10

• W3 LOGQUE2: collision rate 27.4%, 58813 collisions, 536 waits (0.25%), 214664 accesses on region 46

CON: REGION_COLLISION_PERCENT["46"] > MAXCPU * 10

VAL: 27.4 > 2 * 10

• W3 LOGQUE1: collision rate 28.09%, 56918 collisions, 536 waits (0.26%), 202602 accesses on region 45

CON: REGION_COLLISION_PERCENT["45"] > MAXCPU * 10

VAL: 28.09 > 2 * 10

Few hours later the message is only an information because of less collision rate.

===== #223 at 2009-10-07 03:14:39

• I SQL commands executed: 963644

CON: PureSQL_CmdsLiveCache > INTERVAL * 5

VAL: 963644 > 913 * 5

• I DB procedures called: 20050

CON: CallCnt > INTERVAL * 10

VAL: 20050 > 913 * 10

• I LOGQUE1: collision rate 3.18%, 10159 collisions, 93 waits (0.03%), 318985 accesses on region 45

CON: REGION_COLLISION_PERCENT["45"] > ( 1 + MAXCPU / 3 )

VAL: 3.18 > ( 1 + 2 / 3 )

What can I do to reduce the collision rate ?

Our System is a ONEDB

===== #211 at 2009-10-07 00:12:22

• I Number of CPU's 4, processor type: HP IA64 (ia64)

CON: NUM_PHYSICAL_CPU > 1

VAL: 4 > 1

• I Physical memory 98270 MB, virtual memory 524288 MB, memory allocated from instance: 66179.44 MB

CON: NUM_PHYSICAL_CPU > 1

VAL: 4 > 1

• I Operating system: HP-UX B.11.23 U

CON: 1

VAL: 1

• I Kernel version: Kernel 7.6.04 Build 015-123-189-221

CON: 1

VAL: 1

• I Instance O2P (sap09298) is up since 2009-10-04 18:26:18

CON: 1

Thanks a lot

[former_member229109]

Hello Hermann Duessel ,

1. Please post more details about the database version and OS of the database server.

Please update with output of the following commands:

sdbregview -l

xinstinfo <SID-DB>

dbmcli -d <SID-DB> -u <dbm-user>,<pwd> param_directget INSTANCE_TYPE

2. Are you SAP customer?

High collision rates in the LOGQUE region may occur when you perform large number of parallel change operations in the database. Do you know what database applications were running at the time when the DB analyzer reported warnings 3 concerning high collision rate on regions LOGQUE* ?

Thank you and best regards, Natalia Khlopina

[Former Member]

Hello Natalia,

here you get the neccessary informations:

sdbregview -l

DB Analyzer /sapdb/programs 7.6.04.15 64 bit valid

Server Utilities /sapdb/programs 7.6.04.15 64 bit valid

SAP Utilities /sapdb/programs 7.6.04.15 64 bit valid

APO LC APPS /sapdb/O2P/db/sap 5.00.016 64 bit valid

Redist Python /sapdb/programs 7.6.04.15 32 bit valid

Base /sapdb/programs 7.6.04.15 64 bit valid

JDBC /sapdb/programs 7.6.05.06 valid

Messages /sapdb/programs MSG 0.6531 valid

PCR 7500 /sapdb/programs 7.5.00.50 64 bit valid

ODBC /sapdb/programs 7.6.04.15 64 bit valid

Database Kernel /sapdb/O2P/db 7.6.04.15 64 bit valid

Loader /sapdb/programs 7.6.04.15 64 bit valid

SQLDBC /sapdb/programs 7.6.04.15 64 bit valid

SQLDBC 76 /sapdb/programs 7.6.04.15 64 bit valid

Fastload API /sapdb/programs 7.6.04.15 64 bit valid

PCR 7403 /sapdb/programs 7.4.03.50 64 bit valid

xinstinfo O2P

IndepData : /sapdb/data

IndepPrograms : /sapdb/programs

InstallationPath : /sapdb/O2P/db

Kernelversion : KERNEL 7.6.04 BUILD 015-123-189-221

Rundirectory : /sapdb/data/wrk/O2P

dbmcli -d <SID-DB> -u <dbm-user>,<pwd> param_directget INSTANCE_TYPE

OK

INSTANCE_TYPE LVC

sap09298:o2padm 24>

To your 2 Question whether we are SAP customer?

Yes we are SAP customers,

but I can't tell you more about running database applications

because our customer rent apo-specialists who did the daily doing.

I only asked to this forum because several times a week i look to the DB-Analyser-Log

and saw these warnings. So I'm not sure whether these warnings could be a problem in the future

and whether perhaps i have to modify some parameters to solve this warnings.

Thank you and best regards,

Hermann Duessel

[former_member229109]

Hello Hermann Duessel,

1. As I already wrote, high collision rates in the LOGQUE region may occur when you perform large number of parallel change operations in the livecache. You have set MAXCPU to 2. You have 4 CPU on the liveCache server, may be another applications are running on the server. Correct?

With more liveCache applications running in parallel you need more CPU, so you could increase the MAXCPU liveCache parameter value.

2. If the liveCache application users on the system complained about long running jobs/transactions

=> Please create the ticket to the component 'BC-DB-LVC' or 'SV-BO-SCM' to check your system.

May be the liveCache parameters could be adjusted < use SAP note 1111426 first >,

or the application scenario could be changed.

Thank you and best regards, Natalia Khlopina

[Former Member]

Hello Natalia Khlopina,

No the server is exclusive for the ONEDB. So I can also modify the MAXCPU Parameter.

But I think I will watch the Logqueue the next weeks and open an ticket to check the system.

Thanks a lot for your help.

Best Regards

Hermann