Memory pipes (MPI)

The ICM require two memory pipes (MPI) for each HTTP connection.

The MPI configuration in the ICM is limited by parameter

mpi/max_pipes. The default value for this paramter is 4000.

If you has 4000 for that parameter the operation system should provide 8000 operating system semaphores because eatch MPI require two semaphores.

To increase the parameter mpi/max_pipes you should take care the operation system configuration also. Follow the note 715400 to set the operation system as you need.

715400 Max. number of memory pipes (MPI) restricted to 4,000

ERROR MESSAGE:

ERROR => no more free event-flags. [mpixx_mt.c 4555]

ERROR => MpiEvtOpen(0) failed: 11 [mpix_mt.c 1169]

ERROR => IcmConnInitBuffer: MpiCreate2 failed (rc=11)

means that no more semaphores can be created because the ICM reach the

Operation System limits.

HTTPWatch traces

1. Install the free version of HTTPWatch 6.x (http://www.httpwatch.com)

on your Internet Explorer.

2. Start the browser, start the HTTPWatch, click on 'Record' button,

and please reproduce the issue.

3. On finish 'Save' the contents of HTTPWatch as a .hwl file

(Please note: Exporting as CSV or XML is not so helpful, so please use

the default save option).

4. Save .hwl file to this message.

Logon-Check Availability

Function module responsible: SCSM_RUN_RFC_LOGON_CHECK

All ABAP instance with the 'Logon-Check' is set in the CCMSPING availability configuration

will be monitored by the agent.

The SCSM_RUN_RFC_LOGON_CHECK will do this by creating an ARFC session

for each instance and it will wait for an answers of each of these ARFC.

If you have the note 929577, the sessions time out will be by default 2 seconds. If the note is not implementdthe timeout is determined by RFCPING_TIMEOUT.

Trace file's rotation and system restart

You have lost the trace file to perform an anlysis on the system behavior after an error/crash/hung situation.

Every system restart, the trace files are overwritten.
When you start the instance, all files from the services (message
server, dispatcher, work process, ICM and others) are truncated and only
the actual log status information are writen.

But, before truncate the actual trace files, the system create a backup
of this trace file with the .old extension.

Although exist the mechanism to create a backup of the trace file, if
you restart the system more than one time, the .old trace file will be
overwritten too.

A good pratice is make a copy of the work directory before the start up.

Parameter changes without restart

If you need to change some memory parameters but you cannot restart the instance, have solution.
You can change the parameters from RZ11 and run the remport RSMEMORY.

610965 Parameter in RZ11 cannot be switched

java start up

To start up the java stack in a dualstack system or only java.

jcontrol -DSAPSTART=1 pf=/usr/sap/xxx/SYS/profile/xxx_DVEBMGS00_dostname

This is the same than as sapcontrol.

Error in monitoring information

If you are not able to see the statistics information from ST03/ST03N, this can be a error in the timezone configuration.
Please note that the time can been seen at diffenret levels:

1. Operating System Time (or) System time
2. Time zone at Client level whcih is client dependent
3. User time zone whcih can be set using tocde SU3

Make sure if the timezone between the clients and operation system are the same.

926290 SWNCCOLL: SAP workload NW collector collects no data
741734 Incorrect times due to the time zone settings
481835 Analyzing the time zone settings
198411 Current data and information about time zones