RTT/Duration Increasing for Heartbeat agent on Windows 10

I've noticed an add behavior for the Heartbeat agent running on Windows 10. I haven't tested with other OS yet but I figured I would ask if anyone hit that problem. Basically it seems the duration for the checks is increasing over time when I'm not actively using my computer. i.e. While I was gone over the weekend. As soon as I logged back in/started to use my PC again, the duration/RTT went back to normal. Here's what it looks like in Kibana when I'm charting the average duration.us field:

What power mode is your machine configured to use in the control panel? Maybe it's on the power saver mode?

Do you have metricbeat installed? It would be interesting to correlate those peaks to other system CPU/memory/process metrics.

Do the rtt metric values have a similar looking graph?

Hi Andrew!

FYI Here's what the rtt.us graph looks like:


Funny thing, I had metricbeat running on that particular machine up until last week so I don't have any other perf data captured for that time frame.

Right now the machine is using the balanced power plan, so there is CPU throttling happening. I just find strange that it would ramp up that slowly over time. Throttling happens pretty fast on those CPUs typically. It might be the NIC that goes into low power mode but then again, that should happen relatively quickly.


@andrewkroh I just ran some quick tests and it looks like it might be power management related. In the first test, I switched to the Power saver plan and locked the computer for a few minutes. I then saw the tcp_connect_rtt.us maximum go up. When I repeated the test using the High performance plan, I didn't see the same behavior:

I don't know if there's anything from the agent standpoint that could be done to prevent the computer running the agent to enter a low power state and therefore potentially skew the metrics captured.

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.