An external-metric autoscaling path is introduced and later the HPA shows no data despite successful Prometheus queries.
A Prometheus adapter exposes an external metric and the HPA never scales
The metric is present in Prometheus, yet the adapter cannot map it to the workload because one ownership label vanished.
Scenario
What to check first
- Identify the primary failure signal in the The Metric Existed and the Label the Adapter Needed Was Missing scenario.
- Separate visible symptoms from the underlying technical dependency.
- Describe the safest recovery path and the follow-up prevention work.
Checking checklist
- Summarize the current impact and the last known change.
- Collect direct evidence from logs, runtime state, and configuration before changing anything.
- Separate immediate recovery from permanent prevention work.
Recovery and prevention
Trace labels end to end through relabeling and adapter mapping before rewriting HPA specs.
Questions worth viewing together
Community-field Kubernetes problem inspired by Prometheus adapter discussions where relabeling dropped the namespace label the HPA mapping depended on. Autoscaling adapters fail quietly when a single ownership label disappears between scrape and query.
Teams often blame HPA API registration when the adapter cannot map the metric due to relabel loss.
Metric relabel changes should be regression-tested against any autoscaling adapters that depend on those labels.
Similar cases seen in the field