PRTG Manual: Windows Process Sensor
The Windows Process sensor monitors a Windows process via Windows Management Instrumentation (WMI) or Windows performance counters, as configured in the Windows Compatibility Options of the parent device.
The sensor can show the following process parameters:
- Absolute working set in bytes
- Private bytes
- Number of threads
- Number of handles
- Number of instances
- Average CPU usage (if multiple instances are running)
- Total CPU usage
For the CPU Usage (Total) value of a process, all CPU usage values are summed up. The total is divided by the number of all CPUs and the maximum value is 100%. This corresponds to the CPU usage of all instances of this specific process.
For the CPU Usage (average per instance) value, the summed up CPU usage value is divided by the number of all instances. It shows the average CPU usage of a single instance of the process on one CPU.
- Dutch: Windows Proces
- French: Processus Windows
- German: Windows Prozess
- Japanese: Windows プロセス
- Portuguese: Windows Processo
- Russian: Процесс Windows
- Simplified Chinese: Windows 进程
- Spanish: Proceso de Windows
- This sensor requires credentials for Windows systems in the settings of the parent device.
- This sensor requires Windows Server 2008 R2 or later on the probe system.
- This sensor requires that the Remote Registry Windows service runs on the target computer.
- This sensor requires WoW64 (Windows 32-bit on Windows 64-bit) for target systems that run Windows Server 2016.
- The sensor cannot show values above 4 GB for 64-bit processes if you use performance counters.
- This sensor can use a hybrid approach with Windows performance counters and WMI as fallback to query data.
- This sensor supports the IPv6 protocol.
- This sensor has a high performance impact.
You cannot add this sensor to the hosted probe of a PRTG hosted by Paessler instance. If you want to use this sensor, add it to a remote probe device.
Requirement |
Description |
---|---|
Windows credentials |
This sensor requires credentials for Windows systems in the settings of the parent device. Preferably, use Windows domain credentials.
|
Windows version |
For this sensor to work with Windows performance counters, make sure that a Windows version 2008 or later is installed on the probe system (on every cluster node, if on a cluster probe).
|
Remote Registry service |
For this sensor to work with Windows performance counters, make sure that the Remote Registry Windows service runs on the target computer. If this service does not run, a connection via performance counters is not possible. However, WMI connections might still work. To enable the service, log in to the respective computer and open the services manager (for example, via services.msc). In the list, find the respective service and set its Start Type to Automatic. |
Hybrid Approach: Performance Counters and WMI
By default, this sensor uses WMI to request monitoring data. You can change the default behavior to a hybrid approach in the Windows Compatibility Options of the parent device's settings on which you create this sensor: if you choose this option, the sensor first tries to query data via Windows performance counters and uses WMI as a fallback if performance counters are not available. When running in fallback mode, the sensor tries to connect via performance counters again after 24 hours.
Sensors using the WMI protocol have a high impact on the system performance. Try to stay below 200 WMI sensors per probe. Above this number, consider using multiple remote probes for load balancing.
For a general introduction to the technology behind WMI, see section Monitoring via WMI.
The Add Sensor dialog appears when you manually add a new sensor to a device. It only shows the setting fields that are required for creating the sensor. Therefore, you do not see all setting fields in this dialog. You can change (nearly) all settings in the sensor's Settings tab later.
Click the Settings tab of a sensor to change its settings.
Usually, a sensor connects to the IP Address or DNS Name of the parent device on which you created the sensor. See the device settings for details. For some sensors, you can explicitly define the monitoring target in the sensor settings. See below for details on available settings.
Setting |
Description |
---|---|
Sensor Name |
Enter a meaningful name to identify the sensor. By default, PRTG shows this name in the device tree, as well as in alarms, logs, notifications, reports, maps, libraries, and tickets.
|
Parent Tags |
Shows tags that the sensor inherits from its parent device, parent group, and parent probe. This setting is shown for your information only and cannot be changed here. |
Tags |
Enter one or more tags. Confirm each tag with the Spacebar key, a comma, or the Enter key. You can use tags to group objects and use tag-filtered views later on. Tags are not case-sensitive. Tags are automatically inherited.
The sensor has the following default tags that are automatically predefined in the sensor's settings when you add the sensor:
|
Priority |
Select a priority for the sensor. This setting determines the position of the sensor in lists. The highest priority is at the top of a list. Choose from the lowest priority ( |
Setting |
Description |
---|---|
Executable |
Enter the name of the process that you want to monitor. Provide the name of an executable file without the .exe extension (for example, enter firefox to monitor firefox.exe).
|
Setting |
Description |
---|---|
Result Handling |
Define what PRTG does with the sensor result:
|
Setting |
Description |
---|---|
Primary Channel |
Select a channel from the list to define it as the primary channel. In the device tree, the last value of the primary channel is always displayed below the sensor's name. The available options depend on what channels are available for this sensor.
|
Graph Type |
Define how different channels are shown for this sensor:
|
Stack Unit |
This setting is only visible if you enable Stack channels on top of each other as Graph Type. Select a unit from the list. All channels with this unit are stacked on top of each other. By default, you cannot exclude single channels from stacking if they use the selected unit. However, there is an advanced procedure to do so. |
By default, all of the following settings are inherited from objects that are higher in the hierarchy. You should change them centrally in the root group settings if necessary. To change a setting for this object only, click under the corresponding setting name to disable the inheritance. You then see the options described below.
For more information, see section Inheritance of Settings.
Click to interrupt the inheritance.
Setting |
Description |
---|---|
Scanning Interval |
Select a scanning interval from the dropdown list. The scanning interval determines the amount of time that the sensor waits between two scans. Choose from:
|
If a Sensor Query Fails |
Select the number of scanning intervals that the sensor has time to reach and to check a device again if a sensor query fails. Depending on the option that you select, the sensor can try to reach and to check a device again several times before the sensor shows the Down status. This can avoid false alarms if the monitored device only has temporary issues. For previous scanning intervals with failed requests, the sensor shows the Warning status. Choose from:
|
Schedules, Dependencies, and Maintenance Window
You cannot interrupt the inheritance for schedules, dependencies, and maintenance windows. The corresponding settings from the parent objects are always active. However, you can define additional schedules, dependencies, and maintenance windows. They are active at the same time as the parent objects' settings.
Setting |
Description |
---|---|
Schedule |
Select a schedule from the list. You can use schedules to monitor during a certain time span (days or hours) every week. Choose from:
|
Maintenance Window |
Select if you want to set up a one-time maintenance window. During a maintenance window, monitoring stops for the selected object and all child objects. They show the Paused status instead. Choose between:
|
Maintenance Begins |
This setting is only visible if you enable Set up a one-time maintenance window above. Use the date time picker to enter the start date and time of the one-time maintenance window. |
Maintenance Ends |
This setting is only visible if you enable Set up a one-time maintenance window above. Use the date time picker to enter the end date and time of the one-time maintenance window. |
Dependency Type |
Select a dependency type. You can use dependencies to pause monitoring for an object depending on the status of a different object. You can choose from:
|
Dependency |
This setting is only visible if you enable Select a sensor above. Click |
Dependency Delay (Sec.) |
This setting is only visible if you enable Select a sensor above. Define a time span in seconds for the dependency delay. After the master sensor for this dependency returns to the Up status, PRTG additionally delays the monitoring of the dependent objects by the time span you define. This can prevent false alarms, for example, after a server restart or to give systems more time for all services to start. Enter an integer value.
|
Click to interrupt the inheritance.
Setting |
Description |
---|---|
User Group Access |
Define the user groups that have access to the sensor. You see a table with user groups and group access rights. The table contains all user groups in your setup. For each user group, you can choose from the following group access rights:
|
Click to interrupt the inheritance.
Setting |
Description |
---|---|
Channel Unit Types |
For each type of channel, select the unit in which PRTG displays the data. If you define this setting on probe, group, or device level, these settings can be inherited to all sensors underneath. You can set units for the following channel types (if available):
|
KNOWLEDGE BASE
My Windows sensors do not work when using direct Performance Counter access. What can I do?
For more information about sensor settings, see the following sections: