OneAgent causing high load

[NicoleMT]

Hello everyone,

I would like to ask for guidance and hear if anyone has experienced a similar situation.

We have a production host where we observed a significant increase in system load while the Dynatrace OneAgent was enabled. What stands out is that after disabling the OneAgent, the service returned to normal behavior, and no other changes were made on the host or application.

We contacted Dynatrace Support, and according to their analysis, the OneAgent was not identified as the root cause of the observed load. As a next step, they recommended replicating the issue on a test host with the same characteristics to validate the behavior.
However, in our case, we do not have a test environment that exactly replicates the production scenario, so this approach is not currently possible.

Given this situation, we would like to ask the community:

• What alternatives do you recommend when the issue cannot be replicated in a test environment?

• Are there specific OneAgent metrics, configurations, or logs that should be reviewed to better assess its impact?

• Has anyone experienced a similar case where disabling the OneAgent reduced system load?

Any insights or shared experiences would be greatly appreciated.

Thank you in advance.

[AntonioSousa]

@NicoleMT ,
You reference that you saw a significant increase in system load. How did you measure it, before/during/after OneAgent installed? Given you're talking about load, it should be a Linux (or like) system. And how much was that increase?
I personally have seen an increase of some percentual points in CPU usage, due to some special cases. If it's really a Linux system, it should be pretty easy to single out if OneAgent is really responsible... 

[Julius_Loman]

Significantly higher load can happen, but it's very, very rare. If the question is about CPU usage (not directly just a load metric), I'd recommend looking at the OneAgent data on the host screen (the classic one) and use the Davis Causal Correlation Analysis. If it's on a deep monitored application, depending on the technology I'd try to isolate the issue using CPU profiling (if it's CPU related) and look into OneAgent logs from the application. Maybe a case when the application uses a supported framework/technology in a non-standard way and interferes with the OneAgent sensors.

If it's the Linux/Unix load metric which increased, typically I'd look into OneAgent os module logs and also check if you have any network filesystems connected to your host.

Generally, you need to provide much more detail about the observed system and your configuration. I'd start with running OneAgent in infrastructure mode first to rule out any application-specific problems.

[AntonioSousa]

I would just add that if you run it in Infrastructure mode, you also disable "ProcessAgent injection", just to be sure it doesn't interact with processes.