The adjustment may be critical in cases where the operating system does not have sufficient resources. Increasing the parameters also results in a decreased address space for transactions.
For 32-bit systems in particular, you must determine the system status first, and then increase the parameter values if required.
To do this, proceed as follows:
Call transaction ST11 to analyze the developer traces, dev_rd (Gateway), dev_disp (Dispatcher), dev_rfc* (RFC) and dev_w* (work processes).
Call transaction SM21 (syslog) to determine any terminations or bottlenecks.
Call transaction SM51 - Select server - Goto - Server information - WP-CA-blocks or APPC-CA-blocks --> Are the values reaching the maximum (see rdisp/appc_ca_blk_no and rdisp/wp_ca_blk_no)?
See Note 3223 for further information about these parameters. The note describes how to calculate the parameter values in the memory.
Transaction SMGW (Gateway Monitor) - Goto - Logged on Clients
This displays the number of clients or terminals logged on to the gateway of the instance --> does this value correspond to the expected number, and is there sufficient space for additional logons (see also gw/max_conn, for example)?
The parameter settings must be adjusted for a high interface load.For 32-bit systems in particular, you must determine the system status first, and then increase the parameter values if required.
To do this, proceed as follows:
Call transaction ST11 to analyze the developer traces, dev_rd (Gateway), dev_disp (Dispatcher), dev_rfc* (RFC) and dev_w* (work processes).
Call transaction SM21 (syslog) to determine any terminations or bottlenecks.
Call transaction SM51 - Select server - Goto - Server information - WP-CA-blocks or APPC-CA-blocks --> Are the values reaching the maximum (see rdisp/appc_ca_blk_no and rdisp/wp_ca_blk_no)?
See Note 3223 for further information about these parameters. The note describes how to calculate the parameter values in the memory.
Transaction SMGW (Gateway Monitor) - Goto - Logged on Clients
This displays the number of clients or terminals logged on to the gateway of the instance --> does this value correspond to the expected number, and is there sufficient space for additional logons (see also gw/max_conn, for example)?
Set the following parameter settings in the instance profiles:
rdisp/tm_max_no: 2000 (higher values are possible)
Number of connections to an instance, including both dialog users and interfaces.
In case of SAP-RC=672 messages in the syslog, for example
17:38:50 DP Q0P maximum number of terminal connections reached
(number according to rdisp/tm_max_no = XXX)
rdisp/max_comm_entries: 2000 (higher values are possible)
Number of communications from and to an instance (not including dialog users).
rdisp/max_arq: 2000 (= rdisp/max_comm_entries, higher values are possible)
Limits the maximum number of asynchronous messages
used for internal communication to make information known throughout the system.
This parameter must be set to the same value as rdisp/max_comm_entries
see also Note 311930.
gw/max_conn: 2000 (higher values possible)
Number of logical connections to a gateway, that is, the number of
connected gateways and external programs.
gw/max_sys: 2000
Maximum number of Gateway clients
Note that an R/3 server is only a single client with regard to Gateway.
gw/max_overflow_size: 100000000 (100,000,000)
Note the following for AS400: For Release 4.5B or lower, this parameter can be set to a maximum of 16000000.
Maximum swap space for CPIC requests in the gateway, uses swap space
on the disk, will be used from Core Release 4.6D instead of gw/max_shm_req
for all TCP/IP connections.
gw/max_shm_req: 400
Maximum number of CPIC requests in the gateway that are in the shared memory.
As of Kernel Release 4.6D, it will only be used for SNA connections (for example, R2).
It will be ignored elsewhere.
If the system does not have any SNA connections, the parameter does not have to be set as of Release 4.6D.
rdisp/appc_ca_blk_no: 2000 + 2 * rdisp/wp_no_dia(+vb+vb2+enq+btc+spo)
For further details about the calculation, see Note 3223.
If the parameter rdisp/tm_max_no or rdisp/max_comm_entries or the number of work processes is increased, this value must be recalculated.
Buffer for TCP/IP communication. Note the following: This parameter uses space
in the shared memory.
Keep in mind the operating system-dependent requirements
for some of the 32-bit operating systems regarding pool size and
shared memory segments and/or the size of the entire shared memory area
(for example, Note 103747).
You can view the current size of the key containing the rdisp/wp_ca_blk_no
using transaction ST02 (Detail Analysis menu,
Storage, and choose "Shared Memory Detail").
Search for key 3, Dispatcher Communication Areas.
Here you can find the current size of key 3, which contains
rdisp/appc_ca_blk_no. Now calculate the value of key 3
after increasing rdisp/appc_ca_blk_no:
(New value rdisp/appc_ca_blk_no) - (old value rdisp/appc_ca_blk_no
from instance profile) = (delta from increase)
(Delta from increase) * 40KB = (value in KB by which key 3 is increased)
rdisp/wp_ca_blk_no: 2000 + 2 * rdisp/wp_no_dia(+vb+vb2+enq+btc+spo)
For further details about the calculation, see Note 3223.
If the parameter rdisp/tm_max_no or the number of work processes is increased, this value must be recalculated.
Buffer for DIAG communication
Note the following: This parameter uses space in the shared memory (see above).
Like disp/appc_ca_blk_no, it is also in key 3. The size of key 3
can be determined in transaction ST02, as described above.
The value by which key 3 increases after the increase of disp/wp_ca_blk_no is calculated as follows:
(New value rdisp/wp_ca_blk_no) - (current value rdisp/wp_ca_blk_no
from instance profile) = (delta from increase)
(Delta from increase) * 32KB = (value in KB by which key 3 is increased)
Note the following for rdisp/appc_ca_blk_no und rdisp/wp_ca_blk_no:
Both of these parameters require a lot of memory and should only be increased with caution. The pool size of pool 10 must subsequently be checked and adjusted if necessary.
Important: Increasing all the parameters mentioned by default to
the settings described also causes an
excess memory usage (of approximately 10MB) in key 2 (Dispatcher Administration Tables).
The current size of this key can also be determined using transaction ST02, as described above.
All values are only basic recommendations. An in-depth performance analysis is a basic prerequisite for increasing the values. Contact the SAP Support back office, if necessary.
0 comments:
Post a Comment