The monitoring of physical resources is important to keep PPM running properly. You can monitor the system using the ARIS Cloud Controller commands.
File space
Monitor the file system of the PPM installation. When the usage exceeds 85%, increase the disk space.
If the space is consumed by normal usage of the application, for example, many databases, many documents, etc., the available disk space must be increased. Check whether disk space can be made available. Call the deleteLogFiles.bat script (<PPM installation>\server\support). This ensures that current log files are written properly. A maintenance window for the runnable should be used, so that it can be stopped while the script is running. Also check, whether heap dumps have been created. They can consume a large part of disc space.
We recommend increasing the available disk space, if the used space exceeds 75 % of the available disk space.
Memory usage
Every runnable has a JAVA-Xmx configuration parameter. It has to be monitored, if the memory usage is rising over time.
Peaks are not a problem, if they do not break the limit. If the system is suffering over a longer time, for example, the memory usage is higher than 70% for more than one hour, check for rarely occurring use cases. They might explain the high memory consumption. This can be:
If the defined limit for memory consumption is exceeded, there are two possible measures:
CPU load
A continuous high level of CPU usage can be an indicator that it might be necessary to scale up/out the system.
For the CPU load the same parameters as for the heap usage regarding peaks are true. So if a high CPU load is detected (> 60% for more than one hour), first it should be analyzed, if short peaks or exceptional use cases are the reason. If this is the case no direct reaction is needed.
If the defined limit for CPU usage is exceeded, there are two possible measures:
Thread pool size
For most runnables the ajpNumWorkerThreads parameter and the httpNumWorkerThreads parameter are set. Those values should be monitored.
If they regularly exceed 80% of the configuration parameter of the runnable, it has to be scaled out by adding a new runnable of the same type.