Changes to HPA definition!
This release addresses several issues related to overlapping metrics
- External metrics are correctly namespaced
- Metric types are no longer identified by metric name allowing multiple
metrics of the same type in a single Horizontal Pod Autoscaler.
These changes results in small changes to how the metrics are defined on
the HPAs. The new format is documented in the README.md.
You can see how to update your manifests by comparing to the previous version.
The old format is still supported but considered deprecated and will be dropped in a future release.
All changes
- Use upstream version of custom-metrics-apiserver (#211)
- [prom collector ] Documenting scrape interval (#215)
- Fix openapi serving (#217)
- Use labels for mapping metric types to metrics (#219)
- Fix mapping of Prometheus query for new metric configuration (#228)
- UPDATED to use NewSessionWithConfig instead of NewSession (#230)
- Add helm chart manifests (#238, #253)
- Switch to compliant base image (zalando) (#250)
- Use non-deprecated Dockerfile syntax (#256)
- Don't check for metric name on AWS init (#251)
- Update dependencies (#224, #236, #249, #269)
- Use Github actions instead of travis (#270)
- Namespace external metrics (#259)
Docker image
Docker image is available in Zalando's Open Source registry:
docker run -it registry.opensource.zalan.do/teapot/kube-metrics-adapter:v0.1.10 --help
Thanks to everyone contributing to this release: @muaazsaleem, @EwanValentine, @tanersener, @jonathanbeber, @mikkeloscar