Where is the VMC HRT result displayed

863354 - Management of the component VM Container (VMC)

The application server may be overloaded either by too many individual tasks to be carried out at the same time or by tasks that take too long (or both). The "snapshot analysis" described below is used to determine whether this is the case at the moment under consideration. The statistical set analysis described below can then be used to determine which applications have contributed to this. We limit ourselves here to aspects for applications that use the VM container. How it can be determined in retrospect whether an overload situation existed and whether applications using the VM container were involved is described in the sections Load Analysis (Part 2) and Load Analysis (Part 3).

Check the functionality of the VM container

You can check whether the VM container is switched on and basically functional by executing the report RSVMCRT_HEALTH_CHECK. As a result of the execution, this should only display green traffic lights. If the VM container is not active or has been deactivated, this is shown as a result by a gray traffic light (see also section Starting and Stopping). If some of the traffic lights displayed are red, there is an error and should be reported to SAP. The VM container provides so-called "virtual machines" (VMs) for the execution of Java code, which can be integrated into a work process if required. An overview of the available VMs and the memory areas they use in shared memory is displayed in the VM overview (transaction SM52). For the correct functioning of the VM container, the following conditions apply:

l There should always be enough VMs available (the difference between the number of VMs used and the maximum number of VMs should be more than 1). If necessary, the value REM for profile parameter rdisp / max_jvm must be increased.

l The VM Container uses an automatic garbage collection (called "Shared Garbage Collection" or "Shared GC") to manage the shared memory. This is not yet active after the application server has started, but begins to work after some applications have been executed. If possible, it should not work in escalation mode, which is indicated by the display Shared GCAktiv, VMs must contribute (not to be confused with the display Shared GCAktiv, VMs should contribute). If the shared GC is frequently in escalation mode, the configured memory area in the shared memory is too small and must be increased using the vmcj / option / ps parameter.

l As a rule, no errors (see number of errors that have occurred) should occur when executing applications. If so, further information is available after selecting the menu item VMC> Error info. In particular, it can be determined there in which work process the error occurred, so that further information from SAP support can possibly be found there in the developer trace.

Further steps for error analysis in the VM container: With the Java log display, which can be found in the VMC system administration (transaction SM53) in the navigation area under Log administration> Log display, you can view error messages that occurred during Java execution become. To select messages in the VM Container Framework, the category System (under /) should be checked in the log selection. To select messages in the application that uses the VM container, please tick the category Applications. In order to select only the error-relevant log entries, you should select the value ERROR as the severity level (Severity field).

Load analysis (part 1): Snapshot analysis

With the process overview (transaction SM50) you can find out which applications are currently being executed on an application server and how long it has taken to run until now. Are z. For example, if all dialog work processes are occupied, or if some very long-running applications are displayed for dialog work processes, then the system could be overloaded at the moment. You can tell whether an application is currently executing a Java application component in the VM container by the fact that the name of an ABAP program is not displayed in the Report column as usual, but rather the name of the Java function module is prefixed, e.g. B. "Function SPC_GET_PRICING_PROCEDURE". Further development-related information on this Java function module can be viewed under the given name of the module in the Function Builder (transaction SE37). If it turns out that the application is executing Java function modules in the VM container, you can use the VM overview (transaction SM52) to analyze the execution status in the VM container in more detail.

Load Analysis (Part 2): Statistics Set Analysis

For each dialog step, a statistics record with possibly several sub-items is generated, with sub-sets with information on its use being generated, especially when the VM container is used. If you want to analyze whether there were overload situations recently (for example on the same day), you can use the SAP workload single record analysis (transaction STAD) to determine the execution times and resource consumption with the help of this. Statistical sets are examined. In the list of statistics records displayed there, those with the use of the VM container can be recognized by the fact that times greater than 0 ms are displayed in the VMC elapsed time column. By double-clicking on the corresponding line and pressing the VMC pushbutton, the details of the execution in the VM container can be viewed.

Load analysis (part 3): load of past days, weeks and months

In the workload monitor (transaction ST03) as well as in the global workload analysis (transaction ST03G) you will find a wide range of information on time and resource consumption in the following analysis aggregates:

l System load overview (resource consumption per task type (dialog, booking, etc.))

l Top response time

l Transaction profile (resource consumption of used applications)