spoke-dr

GitOps

spoke-dr-cluster-config recorded Synced/Healthy.

Storage

ODF desired/live spec uses compact LVMS topology and StorageCluster is ready.

OADP

lab-dpa reconciled, BSL available, latest scheduled daily completed in the recorded run. A historical partially failed backup remains in history.

Local app middleware

Demo middleware was retired from spoke desired state. Non-core app middleware is not tracked as part of the OpenShift core operations scope.

AI platform

RHOAI operator installed, but no DataScienceCluster or user AI workloads exist.

User workload metrics

Enabled; user workload Prometheus, Thanos ruler, and the Prometheus operator should run here.

Vault / ESO

SecretStore/rke2-vault is Ready=True and ExternalSecret/eso-vault-smoke is synced through the kubernetes-spoke-dr Vault auth mount.

• servicemeshoperator3.v3.3.2 and kiali-operator.v2.22.2 CSVs recorded Succeeded.
• Istio/default, IstioCNI/default, ZTunnel/default, Kiali, OSSM Console, CNI pods, and ingress gateway are recorded running.
• Ambient components are pinned to OSSM 3.3 version v1.28.5.
• No application namespace has opted into ambient yet; use istio.io/dataplane-mode=ambient for app onboarding.
• RHOAI ServiceMesh capability is recorded False/MissingOperator because it expects the old OSSM v2 operator; this does not block OSSM 3 itself.

spoke-dr remains platform standby — operators, OSSM 3, ESO, OADP, and storage stay ready, but no application workloads run in normal operation. lab-workloads/clusters/spoke-dr/kustomization.yaml is intentionally resources: []; the workload ApplicationSet matches spoke-dc only. Full rationale and alternatives in ADR-0001.

This avoids the cost of cross-cluster session replication, dual image promotion, and split-brain handling that hot-standby would imply. The trade-off is non-trivial activation time during a regional spoke-dc loss; that's accepted in exchange for operational simplicity.