How to optimize your local machine for the Kiuwan Local Analyzer (KLA)


 

Every step in an analysis with the KLA is executed sequentially, following the order below: 

 

  1. For each technology

    1. rule analysis

    2. metrics analysis

    3. clone detection

  2. Report generation, encryption and uploading to Kiuwan cloud

 

Every step is executed by a “new” JVM and the Kiuwan configuration applies to all of those JVM instances. If your source code contains more than one technology, it will executre each step for each technology. 

Here are some things you should consider to properly configure your analyses. 

Java Virtual Machine

Every step is performed through the execution of a JVM process.

Java 8 (64 bits) or above —either JDK or JRE— is required. 

By default, KLA comes with pre-configured default values:

  • max memory to use during every single step

  • max duration time (timeout) of every single step

These parameters are configured through KLA configuration mechanism, please do not modify KLA scripts to include JVM flags such as -Xmx, use the mechanisms Kiuwan provides.

Please check /dev/random configuration of JVM. This may produce severe performance problems (java urandom Entropy).

Here is how to fix it: Analyses are very slow in Unix Linux, or halt when uploading results to Kiuwan

Single and parallel execution of analyses

If the machine running KLA analyses contains more available CPUs, you can run “simultaneous” (or parallel) analyses by executing additional instances of KLA.

In a parallel KLA execution scenario, every running analysis is completely independent from each other, so you can execute multiple analysis provided your machine has enough CPUs.

Memory configuration

By default, the KLA comes pre-configured with the following memory default values (analyzer.properties):

 

# Starting size for heap memory (128m = 128 Megabytes)
memory.start=128m
# Maximum size for heap memory (1024m = 1 Gigabyte)
memory.max=1024m
# Stack memory, per thread (1024k = 1 Megabyte)
stack.size=2048k

 

If your local analysis ends with an Out of Memory (OOM) error, you need to increase the max memory allocated to the JVM (by default, 1GB).

The troubleshooting links below can help you identify OOM errors:

You can configure Kiuwan to increase memory limits either for the whole installation or per application.

Depending on your available physical memory, OS and JVM version, if you increase the max memory the JVM might not start (please see https://www.kiuwan.com/docs/display/K5/Not+enough+Memory )

In these cases, stopping unneeded processes (or restarting the machine) can free unneeded allocated memory. Nevertheless, sometimes this does not free memory so you need to test with lower memory values.

 

Although you were not getting an OOM, if you notice the process is performing a high activity of JVM garbage collection, this situation may indicate your analysis needs more memory and the performance is suffering due to gc activity. In theses cases, try to increase the max memory, most probably the analysis performance will be faster.

 

IMPORTANT:

Do not increase indefinitely the memory.

If you see that your analysis needs more than 2GB to finish, it might be a clue of an existing memory leak or some other strange situation.

In this case, do not hesitate to contact Kiuwan Technical Support and report this situation.

 

Timeout configuration

By default, every step of a local analysis is configured to a default max execution time (60 minutes) (analyzer.properties):

 

# Timeout to use for max execution time of each analysis step (in msecs)
 timeout=3600000

 

The default value is often enough for most of the analyses, but depending on several circumstances (code size, memory, ruleset, etc) could not be enough and a timeout error will occur (Timeout killed the subprocess).

If this happens, you can increase the default value.