Skip to Content
ChangelogNovember 2026 - V1.23.0

November 2026 - V1.23.0

This release expands workload optimization to CronJobs and Workload Groups, adds HPA-aware protection and MCP configuration tools, and improves Spot provisioning and memory visibility.

🚀 Highlights

HPA-Aware Workload Optimization

Workload Autoscaler detects HPAs directly targeting Deployments and StatefulSets and protects CPU and Memory settings according to their metric scope. Unprotected resources remain eligible for optimization; custom, external, or unknown metrics trigger conservative protection. See Coexisting with HPA .

CronJob and Workload Group Support

Optimize CronJobs and namespace-scoped Workload Groups defined by Pod selectors. Recommendations aggregate observations across matching Pods, extending resource sizing to recurring jobs and custom workloads. Both support recommendation-only (off) and creation-time (oncreate) modes; existing Pods are not resized or recreated by these targets.

Improved Spot Resilience

AWS adds an opt-in preference for Spot pools with longer observed node lifetimes, while preserving placement constraints and fallback allocation. Observed lifetimes guide selection rather than guarantee future availability; the feature is disabled by default.

New GKE Spot nodes request advance preemption notice, and Spot Handler now detects GCE preemption events to initiate interruption handling.

Expanded MCP Capabilities

AI assistants can now search and retrieve official CloudPilot documentation, inspect Workload Autoscaler audit history, and read scheduling diagnostics.

Optional write access adds NodePool, NodeClass, and autoscaling policy creation, plus updates to supported cluster settings, workload placement, and scheduled rebalances. Connections remain read-only unless write access is explicitly approved. Existing organization permissions and SaaS Operation Audit Logs  apply to organization-scoped actions.

⚙️ Enhancements

  • Add configurable memory headroom and a zero-request fallback to AWS provisioning and rebalance planning without changing Pod resource requests.
  • Report workload-level Cold Memory observations to help identify Memory Extension candidates. Collection is opt-in and requires supported nodes.
  • Update component dependencies, certificate bundles, and regional image mirrors.

🛠️ Bug Fixes

  • Correct topology-spread, DaemonSet overhead, host-port, and in-place resize accounting across AWS, GKE, and AKS.
  • Improve GKE image selection and recovery after failed zone launches, and avoid rejecting usable AKS subnets with a fixed IP threshold.
  • Apply workload update thresholds, shared rollout limits, and OOM recommendation ceilings consistently.
  • Improve JVM telemetry recovery after process exits and preserve independently available workload metrics.

📦 Upgrade Notes

Follow the Upgrade Guide  to update components, CRDs, and RBAC together. HPA protection requires controller read access to HPAs and LimitRanges; image-only upgrades do not add these permissions. See Workload Autoscaler Installation .

For questions or support, join our Slack community .

Last updated on