Skip to content

Supported resource kinds

Every kind below has a list screen and a detail screen. Custom resources are handled separately — see custom resources.

Pods, Deployments, ReplicaSets, ReplicationControllers, StatefulSets, DaemonSets, Jobs, CronJobs, HorizontalPodAutoscalers.

ReplicaSets and ReplicationControllers list as their own kinds, each row naming the controller above it — the only way to tell two rollouts of one deployment apart.

ConfigMaps, Secrets, ResourceQuotas, LimitRanges, PriorityClasses, RuntimeClasses, Leases.

Quotas and limit ranges flatten to one row per resource, rather than one row per object with a nested table.

Services, Ingresses, IngressClasses, NetworkPolicies, Endpoints.

PersistentVolumes, PersistentVolumeClaims, StorageClasses, VolumeSnapshots, CSIDrivers, CSINodes, CSIStorageCapacity.

A CSIDriver’s attachRequired defaults to true when unset, unlike its other flags — so an omitted spec does not read as a driver that needs no attach step.

ServiceAccounts, Roles, RoleBindings, ClusterRoles, ClusterRoleBindings.

PodDisruptionBudgets, ValidatingWebhookConfigurations, MutatingWebhookConfigurations, ValidatingAdmissionPolicies, APIServices, FlowSchemas.

A webhook configuration reads Blocking when any of its webhooks fails closed — which is the API’s own default when failurePolicy is unset — because such a configuration is a dependency of writing at all. An unavailable APIService explains why a whole API group’s kinds have vanished.

A PodDisruptionBudget allowing no disruption reads Blocked, since a drain stops on it and no count says so on its own.

Nodes, Namespaces, Events, CustomResourceDefinitions.

Namespaces and Nodes are also manifest kinds, so they can be read, created and deleted through the ordinary write path — and both ask you to type the name before deleting.

Releases and their revisions. See Helm releases.