I have a setup where I have an open-telemetry collector scraping node-exporter prometheus endpoints and sending it to signoz otlp receiver. The metrics are being stored as temporality of

Unspecified

. I have a dashboard panel for node cpu that was created off these metrics using the v3/query and am able to use the sum_rate function to properly show the CPU utlization. But with the new dashboard panels (v4/query) treating this metric as a rate is not possible and the results are not correct.

@Srikanth Chekuri please have a look when you get some time

Hello, What version of collector are you using and what is the exact metric name that shows the incorrect temporality. All Prometheus metrics are either Cumulative or Unspecified.

The metric is

node_cpu_seconds_total

I am running the latest version of the collector (0.88.20). The metric temporatlity is correct. I feel like the metric explorer has lost a lot of functionality from 0.39 -> 0.43 where I used to be able to perform a sum_rate on the metric, but now that option is not given to me in the UI.

This is not related to temporality. The options are based on the metric type. If the metric type is observed as Counter, then you get the options for sum/avg/min/max - rate/increase. What is the recorded metric type for this

node_cpu_seconds_total

?

It is marked as a Gauge...I will look to see why it is marked that way