github dragonflydb/dragonfly-operator v1.7.0

4 hours ago

Release v1.7.0

Docker image: ghcr.io/dragonflydb/operator:v1.7.0
Helm chart: oci://ghcr.io/dragonflydb/dragonfly-operator/helm

⚠️ Breaking Change

  • Operator-managed NetworkPolicy has been removed (revert of the feature added in #473/#514, via #560).
    • The spec.networkPolicyEnabled field has been removed from the Dragonfly CRD.
    • Upgrading the operator itself is safe for existing Dragonfly CRs that already have this field set — it is simply ignored (unknown JSON fields are dropped silently when the operator reads the object; nothing crashes or errors).
    • However, re-applying an old manifest that still sets networkPolicyEnabled (e.g. via GitOps/CI re-sync) may be rejected: kubectl 1.27+ defaults to --validate=strict, which performs client-side schema validation and will error on the now-unknown field. Direct API/patch updates bypass this and just have the field silently pruned server-side. Recommendation: drop networkPolicyEnabled from your stored manifests going forward to avoid this.
    • The operator will no longer create, update, or reconcile away a NetworkPolicy for Dragonfly instances.
    • This was reverted because it broke common setups where clients/workloads live in a different namespace than the Dragonfly instance (#546), and it blocked external Prometheus/VictoriaMetrics scrapers from reaching /metrics (#495).
    • Action required: if you relied on the operator-managed policy (default or networkPolicyEnabled: true), you must now create and maintain your own NetworkPolicy for equivalent restrictions.
    • ⚠️ Leftover NetworkPolicy objects are NOT auto-deleted on upgrade. If you were running v1.6.0/v1.6.1 with networkPolicyEnabled unset or true, the operator had already created a NetworkPolicy object (owned by, but not garbage-collected on operator upgrade from, the Dragonfly CR). This revert only removes the reconcile logic — it does not clean up existing objects. After upgrading, you must manually delete any operator-created policies to fully restore pre-#473 behavior:
   kubectl delete networkpolicy <dragonfly-instance-name> -n <namespace>

What's Changed

  • chore(actions): bump golang.org/x/net from 0.49.0 to 0.55.0 by @dependabot[bot] in #557
  • chore(actions): bump actions/checkout from 6 to 7 by @dependabot[bot] in #554
  • chore(actions): bump github.com/onsi/ginkgo/v2 from 2.28.1 to 2.32.0 by @dependabot[bot] in #548
  • chore(deps): Bump deps by @moredure in #558
  • feat(chart): support for a namespaced scope deployment by @cjabrantes in #556
  • chore(actions): bump github.com/onsi/gomega from 1.41.0 to 1.42.1 by @dependabot[bot] in #562
  • fix(ginkgo): Bump ginkgo cli by @moredure in #563
  • chore(actions): bump k8s.io/client-go from 0.36.2 to 0.36.3 by @dependabot[bot] in #570
  • chore(actions): bump github.com/go-logr/logr from 1.4.3 to 1.4.4 by @dependabot[bot] in #569
  • feat(chart): add optional PodDisruptionBudget for the operator pod by @somaz94 in #550
  • fix: drop removed annotations and labels from live resources on update by @AruneshDwivedi in #564
  • Improve error message in reconcileResources by @cthtrifork in #571
  • fix(resources): disable service links on Dragonfly pods by @chrisRedwine in #590
  • chore(controller): stop reporting spurious VolumeClaimTemplates changes on every reconcile by @miledxz in #587
  • fix: move status.phase off Ready when no healthy master is available by @miledxz in #591
  • chore(df): bump version to v2.0.0 by @Abhra303 in #594
  • chore(actions): bump github.com/stretchr/testify from 1.11.1 to 1.12.1 by @dependabot[bot] in #580
  • fix(release): push helm chart after operator image is pushed by @Abhra303 in #595
  • revert(operator): remove operator-managed NetworkPolicy by @miledxz in #560
  • feat(release): update manifest files and bump operator version by @Abhra303 in #597

New Contributors

Full Changelog: v1.6.1...v1.7.0

Don't miss a new dragonfly-operator release

NewReleases is sending notifications on new releases.