on 07-05-2011 9:07 AM
Dear All Experts,
Our SAP Hardware Configuration:
(SAP ECC 6.0 with Oracle 10.2.0.4 on Windows Server 2003)
XEON E5450 @ 3GHz (Total 8 Cores)
RAM: 20 GB
FOR ORACLE ERROR (ORA-12518: TNS:listener could not hand off client connection)
We have deployed a new Application server in the landscape and when we increase our Work Processes, we get the above error (ORA-12518).
As per SAP note #830576, we have calculated the SAP Work Processes to total 80 on Production Landscape:
----
*Formula: PARALLEL_MAX_SERVERS = #DB-CPU-Cores * 10*
---
Our hardware has 8 CPU Cores
*Calculation: PARALLEL_MAX_SERVERS = 8 * 10 = 80*
========
*Formula: PROCESSES = ABAP work processes * 2 +#J2EE server processes <max-connections> +PARALLEL_MAX_SERVERS + 40
---
We Require 80 Work Processes
Calculation: PRCOESSES = (80 x 2) + (0 x 0) + 80 + 40 = 280
----
Is our Above Calculation Correct?
it was the First Question, 2nd Question is: We Want to Configure Zero Memory Management on our New Application Server. As per SAP Note 88416 and Hardware ( total 11 GB Memory):
9600 MB considered to be PM because remaining 1.5 GB for OS (IS IT OK OR NOT?)
Load of 100 Users on this Application Server, Therefore [BE]=100
Profile parameter | Standard 32-bit | Standard 64-bit | Unit |
em/initial_size_MB | [PM] | [9600] | Mbyte |
--- em/max_size_MB | --- 20000 | --- 100000 (MAX. 64000 Acceptable??) | Mbyte |
em/address_space_MB | 512 | 4096 | Mbyte |
ztta/roll_first | 1 | 1 | Byte |
--- ztta/roll_area | --- 2000000 | --- 3000000 | Byte |
ztta/roll_extension | 2000000000 | 2000000000 | Byte |
abap/heap_area_dia | 2000000000 | 2000000000 | Byte |
abap/heap_area_nondia | 2000000000 | 0 | Byte |
abap/heap_area_total | 2000000000 | 9600 x 1048576=10,066,329,600 | Byte |
- rdisp/ROLL_MAXFS | [BE] * 100 | 100 x 100 = 10000 | 8KB Block |
--- rdisp/ROLL_SHM | [BE] * 100 | 100 x 100 = 10000 | 8KB Block |
--- rdisp/PG_MAXFS | --- 32768 | --- 32768 | 8KB Block |
rdisp/PG_SHM | [BE] * 50 | 100 x 50 = 5000 | 8KB Block |
-
Please Suggest me if ABOVE Calculations are Correct,
Thanks in Advance,
Regards,
ZAMAN.
Hi Zaman,
>> FOR ORACLE ERROR (ORA-12518: TNS:listener could not hand off client connection)
This error happens, because of network problem or no longer enough memory can be allocated by Oracle
1) Firstly, check event log on your OS, related by the error.
2) You can set DIRECT_HANDOFF_TTC_LISTENER=OFF parameter, in the listener.ora. But this is a workaround...
3) You can add more dispatchers to the system, but I am not sure about it.
>> Calculation: PRCOESSES = (80 x 2) + (0 x 0) + 80 + 40 = 280
It is correct
The other calculations are looking correct, but may need to configure these parameters, when you face with the memory problem, again.
Best regards,
Orkun Gedik
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Thank you Orkun adn Sunny for your replies,
This happened only when we added another Application Server into the landscape and I am considering INSUFFICIENT PARAMETER problem.
Below is the Current Value of the Processes Parameter in ORA file:
*.processes=80
Total Number of SAP Work Processes in our landscape including all Application Servers are 62 and if we decrease them to 58, we dont get Error in TMS, otherwise with 62, we cant run a Transport request.
Anyone else for suggestions?
Thanks in Advance,
Best Regards,
ZAMAN.
Due to changing the PROCESSES Parameter, I think i need to change other dependent parameters aswell, If anyone would review and suggest me if my calculations for the parameters below are OK:
Global DB Parameters including DB, APP & APP2:
==================
==================
-
PARALLEL_MAX_SERVERS = #DB-CPU-Cores x 10
---
Our hardware has 8 CPU Cores
Calculation: PARALLEL_MAX_SERVERS = 8 x 10 = 80
-
PROCESSES = ABAP work processes * 2 +#J2EE server processes <max-connections> +PARALLEL_MAX_SERVERS + 40
===
Currently, we have (DB)21 + (APP) 31 + (APP2) 6 = 58
We Require 80 Work Processes for all Application Servers
Calculation:
-
PROCESSES = (80 x 2) + (0 x 0) + 80 + 40 = 280
-
session = (1.1*processes) + 5
===
session = (1.1 x 280) + 5 = 313
-
transactions = 1.1 x sessions
===
transactions = 1.1 x 313 = 344
-
enqueue_resources = ((sessions - 10) x 2) + 55
===
enqueue_resources = ((313 - 10) x 2) + 55 = 661
-
dml_locks = 4 x transactions
===
dml_locks = 4 x 344 = 1376
-
PARALLEL_MAX_SERVERS = CPU_COUNT x PARALLEL_THREADS_PER_CPU x (2 if PGA_AGGREGATE_TARGET > 0; otherwise 1) x 5
===
PARALLEL_MAX_SERVERS = 8 x 2 x 2 x 5 = 160
-
As per default value for PARALLEL_THREADS_PER_CPU = 2
========================
========================
db_cache_advice = ready, to get approx. value for the parameter.
###########################
If any other parameter is required to change? Please Suggest me with your best solutions, Thank you.
Regards,
ZAMAN.
User | Count |
---|---|
78 | |
9 | |
9 | |
7 | |
7 | |
6 | |
6 | |
5 | |
5 | |
5 |
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.