topic Re: Host anomaly detection settings are not applicable for PaaS/standalone deployments??? in Open Q&A

Host anomaly detection settings are not applicable for PaaS or standalone deployments

sujit_k_singh — Wed, 03 Apr 2024 11:15:56 GMT

Hi,

Under the host monitoring, i am seeing message "Host anomaly detection settings are not applicable for PaaS/standalone deployments." what does that means.

Re: Host anomaly detection settings are not applicable for PaaS/standalone deployments???

Mizső — Thu, 21 Mar 2024 06:53:38 GMT

Hi @sujit_k_singh,

I have only managed PaaS implemantations (some Solaris systems). I thnik at PaaS implemantation only the process is instrumented, there is not any host level metric collection. So I thnik that is why you can not change the host level anomaly detection rules eg. host level CPU saturation and so on...

I hope it helps.

Best regards,

Mizső

Re: Host anomaly detection settings are not applicable for PaaS/standalone deployments???

PacoPorro — Thu, 21 Mar 2024 07:11:08 GMT

Correct.
Platform as a service abstract user from the OS. Oneagent PaaS implementation is about the process not the OS

Re: Host anomaly detection settings are not applicable for PaaS/standalone deployments???

sujit_k_singh — Fri, 22 Mar 2024 01:21:35 GMT

@Mizső and @PacoPorro Thanks for your message... I am clear not even that make sense too.

Re: Host anomaly detection settings are not applicable for PaaS/standalone deployments???

Romanenkov_Al3x — Sat, 23 Mar 2024 09:07:23 GMT

Helpful, thanks!

Re: Host anomaly detection settings are not applicable for PaaS or standalone deployments

dan_m_smith — Tue, 17 Jun 2025 19:33:16 GMT

This raises the question of JVM level Anomaly Detection. I find it odd that these have always been defined at the Host Level and not the Process Level So on our Kubernetes platform running BottleRocket O/S (AppOnly deployment=Paas), we have no way to control the garbage collection anomaly detection settings. If it as defined at the Process level this would not be a problem.

Does anyone have a solution for this, other than Custom Metric Events and Process level tagging or naming.