Monitoring Selected Processes with SAPOSCOL

Use

You can monitor the availability of selected processes with SAPOSCOL. These are displayed in the Monitored Processes subtree of the Operating System monitor and in the detailed analysis of the Operating System Monitor. The monitor displays the number of running processes and the CPU and memory usage for each name template, broken down by different users.

You can optionally set the configuration for process monitoring in the following configuration files (these files are specified in the following templates):

File dev_proc in the SAPOSCOL working directory (see SAPOSCOL Log Files)

Files of the procmon directory in the SAPOSCOL Working Directory. The naming convention for these templates depends on the CCMS agent with which the monitored host is connected to the central monitoring system.

CCMS Agent	Naming Convention
SAPCCMSR SAPCM3X	*procmon.ini
SAPCCMSR -j2ee	<SysID>_<InstNr>procmon.ini
SAPCCM4X	* ABAP_<SysID>_<InstNr>*procmon.ini

Note

<SysID> and <InstNr> are the system ID and the instance number of the monitored ABAP or Java instance. The J2EE Engine automatically generates suitable templates and stores these in the specified (cross-instance) directory. Since the shared memory of the CCMS agents SAPCCMSR -j2ee and SAPCCM4X are only to contain the process monitoring tree for the monitored Engine (ABAP or Java), system ID, and instance number are part of the naming convention for the templates.

The process monitoring is performed at collection intervals of one minute, by default. CCMS agents can read this data from the shared memory and display it in the SAP system.

Prerequisites

The data is transferred to the central monitoring system using a NetWeaver management agent, which must be registered with CEN (see Registering SAP NetWeaver Components and Hosts in CEN).

Process

To start monitoring selected processes, follow the procedure below:

Creating the Configuration Files

Create a template for the process monitoring ( dev_proc or *procmon.ini); the template must have the following structure:

The list of monitored processes begins with $PROC and ends with $..
Comment lines begin with the number sign ( #).

The lines with which you specify the processes to be monitored have the following structure:

<Namensmuster> [[USER=]<Benutzer>] [MTE_CLASS=<MTE-Klasse>] [MTE_NAME=<MTE-Name>] [CUSTOMGROUP=<Attributgruppe>]

The individual parts of the line have the following meaning:

Parameter	Meaning	Remarks
Namensmuster	Monitored process; you can use the wildcard character asterisk ( *) (see the example below); it is not possible to monitor all processes by entering only the wildcard character asterisk	mandatory max. 40 characters
Benutzer	User under whose name the process is running; you can use the wildcard character asterisk ( *)	optional, max. 20 characters
MTE-Klasse	MTE class to which the nodes are to belong; you can create your own monitors in which precisely the desired processes are displayed by assigning the process names to an MTE class Note To do this, use the rule CCMS_GET_MTE_BY_CLASS in a rule-based monitor, and specify the above MTE class there (see Rule Node: Rule Description and Use).	optional, max. 30 characters
MTE-Name	MTE name under which the monitoring object for the monitored process name is displayed in the alert monitor; if you do not set this parameter, Namensmuster is used as the MTE name	optional, max. 40 characters
Attributgruppe	Attribute group to which the attributes of a monitored process name are to belong; you can simplify the maintenance of the threshold values using an assignment to an attribute group	optional, max. 30 characters

Note

Enter the above parameters without quotation marks; the parameter values must not contain any spaces or special characters.

SAPOSCOL reads the contents of the configuration files every five minutes. After changing these files, you do not need to change SAPOSCOL or the responsible CCMS agent.

Setting the Threshold Values

As the expected CPU and memory usage is different for each process, ensure that you adjust the threshold values of the corresponding performance attributes. The simplest way to do this is directly in the Alert Monitor itself, by selecting the relevant performance attribute and choosing Properties (see Changing Properties and Method Assignments).

There are two options available to avoid accidentally changing the threshold values for other monitored processes:

Set the threshold values individually for the MTE (to do this, choose Edit Properties User for Individual MTE )
Use a specific attribute group in the configuration file for the desired monitored process. Your changes then affect only the processes that you have assigned to this group.

Result

The data for the monitored processes is displayed both in the Detail Analysis Menu of the operating system monitor and in the Operating System monitor in the Monitored Processes subtree (settings for MTE class, MTE name, and attribute group have no effect on the output in the operating system monitor).

Note

The output shown corresponds to monitoring set up using the configuration file in the example below.

The following table provides information about the monitoring tree elements (MTEs) of this monitor:

MTE	Meaning
Process Configuration State	Status of the process monitoring; this node exists, even if no processes are being monitored
Process Count	Number of running processes that fulfill the conditions for process name (superordinate node) and user (prefix of the MTE name)
CPU	Total of the CPU usage of the above processes, as a percentage
Resident Size	Total physical memory that is assigned to the above processes
VM Size	Total of the entire memory (physical and virtual) that is assigned to the above processes (only on Windows platforms)

Note

After changes in the configuration files, subtrees are created for new monitored processes at the next start of the data collection methods, which run once per minute; processes that are no longer monitored are first set to inactive after around five minutes (you can identify this due to the gray color) and then deleted after another five minutes.

Example

Configuration File dev_proc for a UNIX Host

Note

For another example, see Configuring Process Monitoring of the Database.