cancel
Showing results for 
Search instead for 
Did you mean: 

SAP instance startup problem - MCOD

Former Member
0 Kudos

Hi All,

We have installed two SAP systems with MCOD database concept ie.. ERP and CRM system uses single database.

Installation completed successfully...

Now the problem is If I start ERP system first then CRM system will not come up and happening vice versa.

Please suggest me a solution ASAP .How can I distribute the memory to the two instances ..If the problem is not related to the Memory what are all the other things needs to be checked.

Thanks in advance

Accepted Solutions (1)

Accepted Solutions (1)

Former Member
0 Kudos

First of all, check via R3trans -d whether you can connect to the database before issuing startsap.

If R3trans -d doesn't show a returncode 0, then check file trans.log.

If R3trans -d is ok and the startup of the second instance fails, then look at the developer trace files in the work directory.

dev_disp and dev_w0 might provide a detailed error message. Maybe the total number of database connections is too limited, or you have not enough RAM, or .... (could be quite a long list of potential problems).

Regards,

Mark

Answers (1)

Answers (1)

Former Member
0 Kudos

Hi Mark,

Thanks for the quick response...

R3trans -d is giving return code zero for both systems and we have 28 GB RAM.

As I mentioned in the thread both systems are starting but one at a time.If I start ERP, I wont be able to start CRM and vice versa.

This is something related to memory .Can you please help me on this??

Thanking You,

Former Member
0 Kudos

It is not necessarily caused by a memory shortage. It could be some other limitation as well. Do you find any error messages in the files dev_disp or dev_w0? When you are in the directory work, you can also perform

grep ORA- *

to search for Oracle error messages in all files.

Regards,

Mark

tomas-krojzl
Active Contributor
0 Kudos

Hello,

If it is related to memory please run for both systems as <sid>adm

sappfpar check pf=<full path to instance profile>

this will give you overview for ABAP memory allocation.

Note that there might be specific settings not taken in account by this tool (for example AIX special memory management)

Another possible cause can be amount of maximum allowed sessions in database. Please check database configuration.

Tomas

former_member188883
Active Contributor
0 Kudos

Hi,

Please share following inputs

1) sappfpar -check pf=<path of instance profile>

for ECC and CRM system

2) db_cache_size for ECC and CRM systems

Regards,

Deepak Kori

Former Member
0 Kudos

Please find the outputs

1.sappfpar outputs

ERP

Nr of operating system shared memory segments: 33

Shared memory resource requirements estimated

================================================================

Total Nr of shared segments required.....: 33

System-imposed number of shared memories.: 1000

Shared memory segment size required min..: 319488000 ( 304.7 MB)

System-imposed maximum segment size......: 49928994816 (47616.0 MB)

Swap space requirements estimated

================================================

Shared memory....................: 1326.9 MB

..in pool 10 126.4 MB, 97% used

..in pool 40 104.1 MB, 97% used

..not in pool: 1090.4 MB

Processes........................: 401.5 MB

Extended Memory .................: 4096.0 MB

Extended Memory 2 ...............: 8976.0 MB

.. 32 VMs, 792.0 MB address space

-


Total, minimum requirement.......: 14800.4 MB

Process local heaps, worst case..: 1907.3 MB

Total, worst case requirement....: 16707.7 MB

Errors detected..................: 0

Warnings detected................: 2

CRM

Nr of operating system shared memory segments: 30

Shared memory resource requirements estimated

================================================================

Total Nr of shared segments required.....: 30

System-imposed number of shared memories.: 1000

Shared memory segment size required min..: 319488000 ( 304.7 MB)

Currently available maximum segment size.: 49928994816 (47616.0 MB)

Swap space requirements estimated

================================================

Shared memory....................: 1343.8 MB

..in pool 10 126.4 MB, 97% used

..in pool 40 104.1 MB, 97% used

..not in pool: 1107.3 MB

Processes........................: 401.5 MB

Extended Memory .................: 4096.0 MB

-


Total, minimum requirement.......: 5841.2 MB

Process local heaps, worst case..: 1907.3 MB

Total, worst case requirement....: 7748.6 MB

Errors detected..................: 0

Warnings detected................: 2

2.

Database buffer cache size is 3.8 GB (I guess you are refering to data buffer size).Since its single MCOD DB there is only one data buffer cache

Thanking You.

Edited by: SAP Basis2411 on Dec 21, 2011 4:43 AM

Former Member
0 Kudos

Imho it's time to look at the trace files dev_disp and dev_w0, as twice suggested by Mark.

nicholas_chang
Active Contributor
0 Kudos

Hi,

Besides check the developer trace, you can look at oracle log as well - alertSID.log to ensure sessions and processes are not in maximum usage.

BTW, what's your swap file size? and What's the OS platform?

Cheers,

Nicholas Chang

tomas-krojzl
Active Contributor
0 Kudos

Hello,

Another possible cause can be amount of maximum allowed sessions in database. Please check database configuration.

Can you please also look at parameter sessions and processes on database level?

show parameter sessions
show parameter processes

These parameters are controlling how many sessions/processes are allowed in database.

Try to compare this with total amount of work processes connecting to this database - please provide amount of ALL work processes for both systems - including all application servers.

Tomas

Former Member
0 Kudos

Hi Thomas,

I have already increased the no of processes and sessions oracle side and same problem was coming

Thanks

Edited by: SAP Basis2411 on Dec 22, 2011 7:34 AM

former_member189725
Active Contributor
0 Kudos

What are the instance numbers you selected for the systems ?

Former Member
0 Kudos

ERP -- 00

CRM -- 05

Edited by: SAP Basis2411 on Dec 22, 2011 10:13 AM

tomas-krojzl
Active Contributor
0 Kudos

Hello,

Imho it's time to look at the trace files dev_disp and dev_w0, as twice suggested by Mark.

I agree - I think we guessed a lot 😄

Tomas

Former Member
0 Kudos

Please find the error message in dev_disp file,Please note all the filesystems are having enough free spaces.

DpCommInitTable: init table for 500 entries

DpRqQInit: keep protect_queue / slots_per_queue 0 / 2001 in sys_adm

rdisp/queue_size_check_value : -> on,50,30,40,500,50,500,80

      • ERROR => e=28 semget(3059979,1,2016) (28: No space left on device) [evtux.c 640]

      • ERROR => DpWpEvtInit: EvtCreate (rc=1) [dpxxwp.c 451]

      • DP_FATAL_ERROR => DpSapEnvInit: DpWpEvtInit

      • DISPATCHER EMERGENCY SHUTDOWN ***

DpHalt: stop work processes

DpHalt: terminate gui connections

DpHalt: wait for end of work processes

DpHalt: not attached to the message server

DpHalt: cleanup EM

EsCleanup( )

EmCleanup() -> 0

Es2Cleanup: Cleanup ES2

SemRq( 65, 1, -1) enter

SemKeyPermission( 65 ) = 0740 (octal)

      • ERROR => e=28 semget(20565,1,2016) (28: No space left on device) [semux.c 498]

      • ERROR => SemRq: Implicit SemInit failed. Key=65 [semux.c 1129]

      • ERROR => Es2Cleanup: SemRq SEM_ES2_ADM (rc=1) [es2xx.c 1082]

Former Member
0 Kudos

Hi Guys,

Thanks for all your co-operation.

Solved by changing the memory parameters in /etc/sysctl.conf file.Now both the systems are up and running

Parameters changed

kernel.sem

kernel.msgmni

Note : Server restart is required after the parameter change