v0.3.0
What changed
- 11 e2e tests covering gRPC health, all profiles, scale-out/in, streaming, error paths, aggregation
- Helm chart: securityContext (privileged for NVML), optional hostPath mounts, updateStrategy, terminationGracePeriod
- CI: added e2e test job, docker build gated on e2e pass
- gofmt -s applied to all Go files
- Added Apache 2.0 LICENSE file
Install
helm install gpu-scaler deploy/helm/keda-gpu-scaler --namespace gpu-scaler --create-namespace