Changelog

Every controller, CLI, API, and dashboard release — documented. Each breaking change is called out explicitly. Subscribe to get release notes the day they ship.

v0.8.1
fix Fixed race condition in burn rate evaluation when multiple AnalysisRuns completed within the same 30-second window. Previously could cause duplicate rollback triggers.
feat kubectl kbs rollout status now shows the full burn rate history for the current analysis window, not just the final verdict.
fix Corrected Datadog metric aggregation when rollupMethod was set to avg over intervals > 5 minutes. Values were being truncated at 4 decimal places.
v0.8.0
feat Multi-cluster rollout orchestration is now available in the Scale tier. A single Rollout spec can now coordinate canary rollout across up to 10 clusters using the multiCluster.targets field.
feat New REST API endpoints: GET /v1/rollouts/{id}/history and POST /v1/rollouts/{id}/pause. Full reference in API docs.
feat Dashboard now shows the 90-day uptime sparkline per service, calculated from AnalysisRun outcomes.
breaking errorBudgetPolicy.windowMinutes now defaults to 60 instead of 30. Existing configurations that relied on the 30-minute default will need to add windowMinutes: 30 explicitly.
v0.7.2
fix Fixed memory leak in the controller when processing Rollouts with more than 20 canarySteps. In practice only affected teams with very granular progressive delivery configurations.
fix Rollback now correctly waits for old-version pod readiness before marking the Rollout as Healthy. Previously, status could briefly show Healthy while pods were still terminating.
feat Added support for Linkerd traffic weight annotations as an alternative to Istio VirtualService for traffic splitting.
v0.7.0
feat PagerDuty notification integration. Set rollbackPolicy.notifyPagerDuty: true and provide an API key in the secret referenced by pagerdutySecretRef.
feat kubectl kbs get rollouts now shows a compact table with current traffic weight, analysis status, and last verdict timestamp.
breaking The v1alpha1.Rollout.spec.strategy field is now required. Rollouts that omit it will fail validation. Add strategy: canary to existing resources.
v0.6.3
feat Initial Helm chart published to charts.kubestead.com. Install with helm repo add kubestead https://charts.kubestead.com && helm install kubestead-controller kubestead/controller.
feat REST API v1 launched with rollout CRUD operations and AnalysisRun query endpoints. API key auth via Authorization: Bearer <key>.
fix Fixed RBAC template to include get/list/watch permissions on analysisruns resource — was causing permission denied errors on fresh installs without custom ClusterRoles.