Jump to: navigation, search

Overload Control: Logging Level

This feature provides the ability to control SIP Server's CPU usage overload by decrementing the server's log level when the CPU usage overload threshold is reached. Overload is detected by per-thread CPU usage measurement. CPU usage is checked every 10 seconds. If the CPU usage of any core SIP Server thread exceeds the value specified in the log-reduce-cpu-threshold configuration option, the log level is decremented to allow SIP Server to handle traffic more efficiently. Once the load drops below 40% of the log-reduce-cpu-threshold configuration option setting, it remains at that level for approximately 300 seconds; after that the logging level is restored to the initially configured level.

When configuring the overload threshold, keep in mind the following:

  • The threshold value must not be configured too high; otherwise, the reduced logging can bring a risk not being enabled at all.
  • The threshold must not be configured too low; otherwise, the lack of logging will make troubleshooting impossible in case of any failure.

Genesys recommends monitoring the SIP Server usage during a typical load spike period, detecting both the start and finish of the period, so the overload threshold is set appropriately.

In HA deployments, the primary and backup SIP Servers monitor and process overload conditions independently. For example, the primary server might be overloaded, while the backup server is not.


log-reduce-cpu-threshold

Setting: overload section, Application level
Default Value: 0
Valid Values: 0, 5-100
Changes Take Effect: Immediately

Specifies the CPU usage overload threshold in percent. When the SIP Server CPU usage increases beyond the specified value, SIP Server is considered overloaded and the log level is decremented. The default value of 0 (zero) disables the dynamic overload control feature.

This page was last modified on December 9, 2016, at 09:48.

Feedback

Comment on this article:

blog comments powered by Disqus