High CPU usage after upgrading to 0.29.2

[guoard]

Problem Statement

After upgrading windows_exporter from version 0.19.0 to 0.29.2, I noticed an increase in CPU usage.

Additionally, I found a single log entry with an error level:

level=error caller=http.go:169 msg="error gathering metrics: collected metric \"\\x00\\x00ndows_tcp_connection_failures_total\" { label:{name:\"af\"  value:\"ipv6\"}  counter:{value:168101}} was collected before with the same name and label values"

Despite this, windows_exporter_collector_duration_seconds showed an improvement, dropping from ~0.8s to <0.2s for all collectors.

Environment

• windows_exporter Version: 0.29.2
• Windows Server Version: 2012-R2

[jkroepke]

What I can say is that https://github.com/prometheus-community/windows_exporter/releases/tag/v0.30.0-rc.0 improves the CPU usage compared to 0.29 release. Based on internal test, the CPU load with all collectors enabled and scrape interval of 15 seconds has been lowered twice.

On a 4 core machine, a cpu usage up 2% is expected.

[guoard]

@jkroepke Thank you for your response! I’ll test it and share the results.