Higher memory limits on ActiveGate in Kubernetes/OpenShift

[pawel_harbuz]

Hi,

As you problably know, recently Dynatrace has changed default Java version for ActiveGate from 11 to 17. The change has influence in the Garbage Collection area, which results in slightly higher memory usage on the ActiveGate level. The default limit for memory is 1.5 GB, which is not enough regarding to the change. In case of our Customers' environments almost everywhere is the default limit which made us to notify everyone to change the limits to at least 2 GB. Not increasing this limit can cause a delay or lack of metrics and traces which are sent using ActiveGate deployed as a pod. This post is only to warn everyone that it can cause problems. Please, be aware and notify your Customers.

Kind Regards,

Paweł

[Lukasz_Halman]

@pawel_harbuz Yes, ActiveGate since 1.295 uses java 17 and that has changed garbage collection behavior, previously used CMS GC has been removed in this version. After GC change results you can observe higher memory utilization, it is however not caused by increased memory requirements, but rather GC prioritizing application performance (lower CPU) over unnecessarily freeing heap space.

There is no reason to increase memory limits for the ActiveGate pod because of java 17. Rather the opposite, performance of ActiveGate has improved under the same resource conditions in our testing.

If you're experiencing problems and had to increase memory limits, please reach out to support to investigate your case.

[pawel_harbuz]

@Lukasz_HalmanActually, I've already reached Dynatrace One support and we confirmed that there is a need of the memory limit increase.