Reference
Edera release notes
v1.3.0

v1.3.0

Release overview

This release focuses on stability improvements, performance optimizations, and expanded observability.
Key highlights include faster OCI image pulls, Kubernetes CRI compliance improvements, PVH stability fixes, and expanded metrics via /metrics.
GovCloud AMIs are now available for AL2023 EKS, and all components have been upgraded to Rust v1.88.0.

What’s new

Kubernetes & Container Runtime

  • AWS Kubernetes 1.33 Support: Added support for EKS 1.33.
  • Enhanced CRI Implementation:
    • Launch host shim process to maintain stable cgroup controllers.
    • Reject host network pods for improved security.
  • Improved Container Runtime: Allow quick rollback from Edera runtime to standard containerd runtime.
  • Registry Authentication: Implement registry auth using the Kubernetes authentication model.
  • Performance Improvements:
    • Enhanced OCI fetch/extract performance.
    • Global retry mechanism for layer acquisition.
    • Use faster zlib-rs for flate2 compression.

Cloud Infrastructure

  • GovCloud Support: Bootstrap IAM configuration and build GovCloud AMIs for AL2023 EKS.

CLI & Developer Experience

  • Add primitive CLI markdown generation.
  • zone top now shows CPU usage and total CPU metrics.

Observability

  • Expose hypervisor metrics from the daemon and orchestrator.
  • Add CPU usage seconds total for zones and the host.
  • Include health counts in the orchestrator.
  • Add hypervisor-side performance counters.
  • Updated metric naming conventions for consistency.

GPU Support

  • PCI device initialization for GPU zone creation.
  • Proper PCI device cleanup during GPU zone destruction.
  • Handle daemon reloads when GPU devices are present.

Fixes and improvements

Zone Management

  • Fix zone destroy --all to properly handle single-zone scenarios.
  • Resolve failure when network backend is set to none due to default routes.
  • Expose zones without network devices to the runtime zone list.
  • Prevent CPU burn when zones fail to reach ready state.

System Integration

  • Proper termination signal handling and cleanup in all systemd daemons.
  • Use systemd overrides instead of direct kubelet unit editing.

Hardware Support

  • Fix PVH APIC corruption issues.

OCI & Image Processing

  • Signal last chunk properly during image import to enable resolution start.
  • Include host block device in addons loop devices for proper cleanup.

Performance & Reliability

  • Implement chunked streaming for List operations (Workload/Zone/NetworkReservation).
  • Canonicalize mount targets before mounting workloads.

Deprecations

  • Deprecated Amazon Linux 2 (AL2) AMIs.

Known issues

None reported in this release.

Upgrade notes

There are no known breaking changes in this release from the previous release v1.2.0.

Last updated on