CPU readiness metric showing 0 in Alloy (vCenter integration) despite config changes

Grafana Alloy
,
1

Hello Team,
We are facing an issue in the alloy version - v1.9.2 with CPU readiness metrics in Grafana Alloy using the vCenter integration. The data is being ingested into Mimir successfully , however the CPU readiness metric consistently reports a value of 0.

Below are the troubleshooting steps and observations so far:

  1. Verified the Alloy configuration for the vCenter integration by running Alloy in debug mode; however, the root cause could not be identified.

  2. Updated and re-deployed the Alloy configuration multiple times, but the issue persists.

  3. Adjusted the collection interval to evaluate the impact:

    • Initially set to 30 seconds
    • Reduced to 10 seconds
    • Also tested with a 2-minute interval, which helped smoothen spikes, but the CPU readiness values are still consistently reported as 0.

  4. In Splunk, the CPU readiness metric is observed in milliseconds, whereas Alloy appears to scrape and forward only percentage-based CPU readiness values.

Could you please help us with the following:

  • Why is CPU readiness always reported as 0 in Alloy?
  • Does Alloy support CPU readiness metrics in milliseconds for vCenter?
  • Is there any specific configuration or metric mapping required to obtain accurate CPU readiness values?
2

Hi Community,

Looking for feedback from anyone who has implemented the vcenter receiver and is actively looking at this data in alloy or the downstream contrib collector.

For reference this is on the vcenter receiver configuration and the specific metric in Mimir is vcenter_vm_cpu_readiness_percent

This is documented on this page: otelcol.receiver.vcenter | Grafana Alloy documentation

vcenter.vm.cpu.readiness metric Enables the vcenter.vm.cpu.readiness metric

In the collector contrib opentelemetry-collector-contrib/receiver/vcenterreceiver/documentation.md at v0.147.0 · open-telemetry/opentelemetry-collector-contrib · GitHub

It is described as vcenter.vm.cpu.readiness

Percentage of time that the virtual machine was ready, but could not get scheduled to run on the physical CPU. It is a gauge type of metric.

This value is expected to be non-zero but it is returned for all VM in all vcenter instances as a zero value.

The vcenter cluster version is v8 family and all other metrics are being received with no issues It literally is just this metric but it is an important metric nontheless.

Regards,

Aron