Well-Known Labels, Annotations and Taints

Kubernetes reserves all labels and annotations in the kubernetes.io and k8s.io namespaces.

This document serves both as a reference to the values and as a coordination point for assigning values.

Labels, annotations and taints used on API objects

app.kubernetes.io/component

Example: app.kubernetes.io/component: "database"

Used on: All Objects (typically used on workload resources ).

The component within the architecture.

One of the recommended labels .

app.kubernetes.io/created-by (deprecated)

Example: app.kubernetes.io/created-by: "controller-manager"

Used on: All Objects (typically used on workload resources ).

The controller/user who created this resource.

app.kubernetes.io/instance

Example: app.kubernetes.io/instance: "mysql-abcxzy"

Used on: All Objects (typically used on workload resources ).

A unique name identifying the instance of an application. To assign a non-unique name, use app.kubernetes.io/name .

One of the recommended labels .

app.kubernetes.io/managed-by

Example: app.kubernetes.io/managed-by: "helm"

Used on: All Objects (typically used on workload resources ).

The tool being used to manage the operation of an application.

One of the recommended labels .

app.kubernetes.io/name

Example: app.kubernetes.io/name: "mysql"

Used on: All Objects (typically used on workload resources ).

The name of the application.

One of the recommended labels .

app.kubernetes.io/part-of

Example: app.kubernetes.io/part-of: "wordpress"

Used on: All Objects (typically used on workload resources ).

The name of a higher-level application this one is part of.

One of the recommended labels .

app.kubernetes.io/version

Example: app.kubernetes.io/version: "5.7.21"

Used on: All Objects (typically used on workload resources ).

The current version of the application.

Common forms of values include:

One of the recommended labels .

applyset.kubernetes.io/additional-namespaces (alpha)

Example: applyset.kubernetes.io/additional-namespaces: "namespace1,namespace2"

Used on: Objects being used as ApplySet parents.

Use of this annotation is alpha. For Kubernetes version 1.27, you can use this annotation on Secrets, ConfigMaps, or custom resources if the CustomResourceDefinition defining them has the applyset.kubernetes.io/is-parent-type label.

Part of the specification used to implement ApplySet-based pruning in kubectl . This annotation is applied to the parent object used to track an ApplySet to extend the scope of the ApplySet beyond the parent object's own namespace (if any). The value is a comma-separated list of the names of namespaces other than the parent's namespace in which objects are found.

applyset.kubernetes.io/contains-group-resources (alpha)

Example: applyset.kubernetes.io/contains-group-resources: "certificates.cert-manager.io,configmaps,deployments.apps,secrets,services"

Used on: Objects being used as ApplySet parents.

Use of this annotation is alpha. For Kubernetes version 1.27, you can use this annotation on Secrets, ConfigMaps, or custom resources if the CustomResourceDefinition defining them has the applyset.kubernetes.io/is-parent-type label.

Part of the specification used to implement ApplySet-based pruning in kubectl . This annotation is applied to the parent object used to track an ApplySet to optimize listing of ApplySet member objects. It is optional in the ApplySet specification, as tools can perform discovery or use a different optimization. However, as of Kubernetes version 1.27, it is required by kubectl. When present, the value of this annotation must be a comma separated list of the group-kinds, in the fully-qualified name format, i.e. <resource>.<group> .

applyset.kubernetes.io/id (alpha)

Example: applyset.kubernetes.io/id: "applyset-0eFHV8ySqp7XoShsGvyWFQD3s96yqwHmzc4e0HR1dsY-v1"

Used on: Objects being used as ApplySet parents.

Use of this label is alpha. For Kubernetes version 1.27, you can use this label on Secrets, ConfigMaps, or custom resources if the CustomResourceDefinition defining them has the applyset.kubernetes.io/is-parent-type label.

Part of the specification used to implement ApplySet-based pruning in kubectl . This label is what makes an object an ApplySet parent object. Its value is the unique ID of the ApplySet, which is derived from the identity of the parent object itself. This ID must be the base64 encoding (using the URL safe encoding of RFC4648) of the hash of the group-kind-name-namespace of the object it is on, in the form: <base64(sha256(<name>.<namespace>.<kind>.<group>))> . There is no relation between the value of this label and object UIDs.

applyset.kubernetes.io/is-parent-type (alpha)

Example: applyset.kubernetes.io/is-parent-type: "true"

Used on: Custom Resource Definition (CRD)

Use of this label is alpha. Part of the specification used to implement ApplySet-based pruning in kubectl . You can set this label on a CustomResourceDefinition (CRD) to identify the custom resource type it defines (not the CRD itself) as an allowed parent for an ApplySet. The only permitted value for this label is "true" ; if you want to mark a CRD as not being a valid parent for ApplySets, omit this label.

applyset.kubernetes.io/part-of (alpha)

Example: applyset.kubernetes.io/part-of: "applyset-0eFHV8ySqp7XoShsGvyWFQD3s96yqwHmzc4e0HR1dsY-v1"

Used on: All objects.

Use of this label is alpha. Part of the specification used to implement ApplySet-based pruning in kubectl . This label is what makes an object a member of an ApplySet. The value of the label must match the value of the applyset.kubernetes.io/id label on the parent object.

applyset.kubernetes.io/tooling (alpha)

Example: applyset.kubernetes.io/tooling: "kubectl/v1.27"

Used on: Objects being used as ApplySet parents.

Use of this annotation is alpha. For Kubernetes version 1.27, you can use this annotation on Secrets, ConfigMaps, or custom resources if the CustomResourceDefinition defining them has the applyset.kubernetes.io/is-parent-type label.

Part of the specification used to implement ApplySet-based pruning in kubectl . This annotation is applied to the parent object used to track an ApplySet to indicate which tooling manages that ApplySet. Tooling should refuse to mutate ApplySets belonging to other tools. The value must be in the format <toolname>/<semver> .

cluster-autoscaler.kubernetes.io/safe-to-evict

Example: cluster-autoscaler.kubernetes.io/safe-to-evict: "true"

Used on: Pod

When this annotation is set to "true" , the cluster autoscaler is allowed to evict a Pod even if other rules would normally prevent that. The cluster autoscaler never evicts Pods that have this annotation explicitly set to "false" ; you could set that on an important Pod that you want to keep running. If this annotation is not set then the cluster autoscaler follows its Pod-level behavior.

config.kubernetes.io/local-config

Example: config.kubernetes.io/local-config: "true"

Used on: All objects

This annotation is used in manifests to mark an object as local configuration that should not be submitted to the Kubernetes API.

A value of "true" for this annotation declares that the object is only consumed by client-side tooling and should not be submitted to the API server.

A value of "false" can be used to declare that the object should be submitted to the API server even when it would otherwise be assumed to be local.

This annotation is part of the Kubernetes Resource Model (KRM) Functions Specification, which is used by Kustomize and similar third-party tools. For example, Kustomize removes objects with this annotation from its final build output.

internal.config.kubernetes.io/* (reserved prefix)

Used on: All objects

This prefix is reserved for internal use by tools that act as orchestrators in accordance with the Kubernetes Resource Model (KRM) Functions Specification. Annotations with this prefix are internal to the orchestration process and are not persisted to the manifests on the filesystem. In other words, the orchestrator tool should set these annotations when reading files from the local filesystem and remove them when writing the output of functions back to the filesystem.

A KRM function must not modify annotations with this prefix, unless otherwise specified for a given annotation. This enables orchestrator tools to add additional internal annotations, without requiring changes to existing functions.

internal.config.kubernetes.io/path

Example: internal.config.kubernetes.io/path: "relative/file/path.yaml"

Used on: All objects

This annotation records the slash-delimited, OS-agnostic, relative path to the manifest file the object was loaded from. The path is relative to a fixed location on the filesystem, determined by the orchestrator tool.

This annotation is part of the Kubernetes Resource Model (KRM) Functions Specification, which is used by Kustomize and similar third-party tools.

A KRM Function should not modify this annotation on input objects unless it is modifying the referenced files. A KRM Function may include this annotation on objects it generates.

internal.config.kubernetes.io/index

Example: internal.config.kubernetes.io/index: "2"

Used on: All objects

This annotation records the zero-indexed position of the YAML document that contains the object within the manifest file the object was loaded from. Note that YAML documents are separated by three dashes ( --- ) and can each contain one object. When this annotation is not specified, a value of 0 is implied.

This annotation is part of the Kubernetes Resource Model (KRM) Functions Specification, which is used by Kustomize and similar third-party tools.

A KRM Function should not modify this annotation on input objects unless it is modifying the referenced files. A KRM Function may include this annotation on objects it generates.

kubernetes.io/arch

Example: kubernetes.io/arch: "amd64"

Used on: Node

The Kubelet populates this with runtime.GOARCH as defined by Go. This can be handy if you are mixing arm and x86 nodes.

kubernetes.io/os

Example: kubernetes.io/os: "linux"

Used on: Node, Pod

For nodes, the kubelet populates this with runtime.GOOS as defined by Go. This can be handy if you are mixing operating systems in your cluster (for example: mixing Linux and Windows nodes).

You can also set this label on a Pod. Kubernetes allows you to set any value for this label; if you use this label, you should nevertheless set it to the Go runtime.GOOS string for the operating system that this Pod actually works with.

When the kubernetes.io/os label value for a Pod does not match the label value on a Node, the kubelet on the node will not admit the Pod. However, this is not taken into account by the kube-scheduler. Alternatively, the kubelet refuses to run a Pod where you have specified a Pod OS, if this isn't the same as the operating system for the node where that kubelet is running. Just look for Pods OS for more details.

kubernetes.io/metadata.name

Example: kubernetes.io/metadata.name: "mynamespace"

Used on: Namespaces

The Kubernetes API server (part of the control plane ) sets this label on all namespaces. The label value is set to the name of the namespace. You can't change this label's value.

This is useful if you want to target a specific namespace with a label selector .

kubernetes.io/limit-ranger

Example: kubernetes.io/limit-ranger: "LimitRanger plugin set: cpu, memory request for container nginx; cpu, memory limit for container nginx"

Used on: Pod

Kubernetes by default doesn't provide any resource limit, that means unless you explicitly define limits, your container can consume unlimited CPU and memory. You can define a default request or default limit for pods. You do this by creating a LimitRange in the relevant namespace. Pods deployed after you define a LimitRange will have these limits applied to them. The annotation kubernetes.io/limit-ranger records that resource defaults were specified for the Pod, and they were applied successfully. For more details, read about LimitRanges .

addonmanager.kubernetes.io/mode

Example: addonmanager.kubernetes.io/mode: "Reconcile"

Used on: All objects

To specify how an add-on should be managed, you can use the addonmanager.kubernetes.io/mode label. This label can have one of three values: Reconcile , EnsureExists , or Ignore .

  • Reconcile : Addon resources will be periodically reconciled with the expected state. If there are any differences, the add-on manager will recreate, reconfigure or delete the resources as needed. This is the default mode if no label is specified.
  • EnsureExists : Addon resources will be checked for existence only but will not be modified after creation. The add-on manager will create or re-create the resources when there is no instance of the resource with that name.
  • Ignore : Addon resources will be ignored. This mode is useful for add-ons that are not compatible with the add-on manager or that are managed by another controller.

For more details, see Addon-manager

beta.kubernetes.io/arch (deprecated)

This label has been deprecated. Please use kubernetes.io/arch instead.

beta.kubernetes.io/os (deprecated)

This label has been deprecated. Please use kubernetes.io/os instead.

kube-aggregator.kubernetes.io/automanaged

Example: kube-aggregator.kubernetes.io/automanaged: "onstart"

Used on: APIService

The kube-apiserver sets this label on any APIService object that the API server has created automatically. The label marks how the control plane should manage that APIService. You should not add, modify, or remove this label by yourself.

There are two possible values:

  • onstart : The APIService should be reconciled when an API server starts up, but not otherwise.
  • true : The API server should reconcile this APIService continuously.

service.alpha.kubernetes.io/tolerate-unready-endpoints (deprecated)

Used on: StatefulSet

This annotation on a Service denotes if the Endpoints controller should go ahead and create Endpoints for unready Pods. Endpoints of these Services retain their DNS records and continue receiving traffic for the Service from the moment the kubelet starts all containers in the pod and marks it Running , til the kubelet stops all containers and deletes the pod from the API server.

kubernetes.io/hostname

Example: kubernetes.io/hostname: "ip-172-20-114-199.ec2.internal"

Used on: Node

The Kubelet populates this label with the hostname. Note that the hostname can be changed from the "actual" hostname by passing the --hostname-override flag to the kubelet .

This label is also used as part of the topology hierarchy. See topology.kubernetes.io/zone for more information.

kubernetes.io/change-cause

Example: kubernetes.io/change-cause: "kubectl edit --record deployment foo"

Used on: All Objects

This annotation is a best guess at why something was changed.

It is populated when adding --record to a kubectl command that may change an object.

kubernetes.io/description

Example: kubernetes.io/description: "Description of K8s object."

Used on: All Objects

This annotation is used for describing specific behaviour of given object.

kubernetes.io/enforce-mountable-secrets

Example: kubernetes.io/enforce-mountable-secrets: "true"

Used on: ServiceAccount

The value for this annotation must be true to take effect. This annotation indicates that pods running as this service account may only reference Secret API objects specified in the service account's secrets field.

node.kubernetes.io/exclude-from-external-load-balancers

Example: node.kubernetes.io/exclude-from-external-load-balancers

Used on: Node

Kubernetes automatically enables the ServiceNodeExclusion feature gate on the clusters it creates. With this feature gate enabled on a cluster, you can add labels to particular worker nodes to exclude them from the list of backend servers. The following command can be used to exclude a worker node from the list of backend servers in a backend set- kubectl label nodes <node-name> node.kubernetes.io/exclude-from-external-load-balancers=true

controller.kubernetes.io/pod-deletion-cost

Example: controller.kubernetes.io/pod-deletion-cost: "10"

Used on: Pod

This annotation is used to set Pod Deletion Cost which allows users to influence ReplicaSet downscaling order. The annotation parses into an int32 type.

cluster-autoscaler.kubernetes.io/enable-ds-eviction

Example: cluster-autoscaler.kubernetes.io/enable-ds-eviction: "true"

Used on: Pod

This annotation controls whether a DaemonSet pod should be evicted by a ClusterAutoscaler. This annotation needs to be specified on DaemonSet pods in a DaemonSet manifest. When this annotation is set to "true" , the ClusterAutoscaler is allowed to evict a DaemonSet Pod, even if other rules would normally prevent that. To disallow the ClusterAutoscaler from evicting DaemonSet pods, you can set this annotation to "false" for important DaemonSet pods. If this annotation is not set, then the Cluster Autoscaler follows its overall behaviour (i.e evict the DaemonSets based on its configuration).

kubernetes.io/ingress-bandwidth

Example: kubernetes.io/ingress-bandwidth: 10M

Used on: Pod

You can apply quality-of-service traffic shaping to a pod and effectively limit its available bandwidth. Ingress traffic (to the pod) is handled by shaping queued packets to effectively handle data. To limit the bandwidth on a pod, write an object definition JSON file and specify the data traffic speed using kubernetes.io/ingress-bandwidth annotation. The unit used for specifying ingress rate is bits per second, as a Quantity . For example, 10M means 10 megabits per second.

kubernetes.io/egress-bandwidth

Example: kubernetes.io/egress-bandwidth: 10M

Used on: Pod

Egress traffic (from the pod) is handled by policing, which simply drops packets in excess of the configured rate. The limits you place on a pod do not affect the bandwidth of other pods. To limit the bandwidth on a pod, write an object definition JSON file and specify the data traffic speed using kubernetes.io/egress-bandwidth annotation. The unit used for specifying egress rate is bits per second, as a Quantity . For example, 10M means 10 megabits per second.

beta.kubernetes.io/instance-type (deprecated)

node.kubernetes.io/instance-type

Example: node.kubernetes.io/instance-type: "m3.medium"

Used on: Node

The Kubelet populates this with the instance type as defined by the cloudprovider . This will be set only if you are using a cloudprovider . This setting is handy if you want to target certain workloads to certain instance types, but typically you want to rely on the Kubernetes scheduler to perform resource-based scheduling. You should aim to schedule based on properties rather than on instance types (for example: require a GPU, instead of requiring a g2.2xlarge ).

failure-domain.beta.kubernetes.io/region (deprecated)

See topology.kubernetes.io/region .

failure-domain.beta.kubernetes.io/zone (deprecated)

See topology.kubernetes.io/zone .

pv.kubernetes.io/bind-completed

Example: pv.kubernetes.io/bind-completed: "yes"

Used on: PersistentVolumeClaim

When this annotation is set on a PersistentVolumeClaim (PVC), that indicates that the lifecycle of the PVC has passed through initial binding setup. When present, that information changes how the control plane interprets the state of PVC objects. The value of this annotation does not matter to Kubernetes.

pv.kubernetes.io/bound-by-controller

Example: pv.kubernetes.io/bound-by-controller: "yes"

Used on: PersistentVolume, PersistentVolumeClaim

If this annotation is set on a PersistentVolume or PersistentVolumeClaim, it indicates that a storage binding (PersistentVolume → PersistentVolumeClaim, or PersistentVolumeClaim → PersistentVolume) was installed by the controller . If the annotation isn't set, and there is a storage binding in place, the absence of that annotation means that the binding was done manually. The value of this annotation does not matter.

pv.kubernetes.io/provisioned-by

Example: pv.kubernetes.io/provisioned-by: "kubernetes.io/rbd"

Used on: PersistentVolume

This annotation is added to a PersistentVolume(PV) that has been dynamically provisioned by Kubernetes. Its value is the name of volume plugin that created the volume. It serves both user (to show where a PV comes from) and Kubernetes (to recognize dynamically provisioned PVs in its decisions).

pv.kubernetes.io/migrated-to

Example: pv.kubernetes.io/migrated-to: pd.csi.storage.gke.io

Used on: PersistentVolume, PersistentVolumeClaim

It is added to a PersistentVolume(PV) and PersistentVolumeClaim(PVC) that is supposed to be dynamically provisioned/deleted by its corresponding CSI driver through the CSIMigration feature gate. When this annotation is set, the Kubernetes components will "stand-down" and the external-provisioner will act on the objects.

statefulset.kubernetes.io/pod-name

Example:

statefulset.kubernetes.io/pod-name: "mystatefulset-7"

When a StatefulSet controller creates a Pod for the StatefulSet, the control plane sets this label on that Pod. The value of the label is the name of the Pod being created.

See Pod Name Label in the StatefulSet topic for more details.

scheduler.alpha.kubernetes.io/node-selector

Example: scheduler.alpha.kubernetes.io/node-selector: "name-of-node-selector"

Used on: Namespace

The PodNodeSelector uses this annotation key to assign node selectors to pods in namespaces.

topology.kubernetes.io/region

Example:

topology.kubernetes.io/region: "us-east-1"

See topology.kubernetes.io/zone .

topology.kubernetes.io/zone

Example:

topology.kubernetes.io/zone: "us-east-1c"

Used on: Node, PersistentVolume

On Node: The kubelet or the external cloud-controller-manager populates this with the information as provided by the cloudprovider . This will be set only if you are using a cloudprovider . However, you should consider setting this on nodes if it makes sense in your topology.

On PersistentVolume: topology-aware volume provisioners will automatically set node affinity constraints on PersistentVolumes .

A zone represents a logical failure domain. It is common for Kubernetes clusters to span multiple zones for increased availability. While the exact definition of a zone is left to infrastructure implementations, common properties of a zone include very low network latency within a zone, no-cost network traffic within a zone, and failure independence from other zones. For example, nodes within a zone might share a network switch, but nodes in different zones should not.

A region represents a larger domain, made up of one or more zones. It is uncommon for Kubernetes clusters to span multiple regions, While the exact definition of a zone or region is left to infrastructure implementations, common properties of a region include higher network latency between them than within them, non-zero cost for network traffic between them, and failure independence from other zones or regions. For example, nodes within a region might share power infrastructure (e.g. a UPS or generator), but nodes in different regions typically would not.

Kubernetes makes a few assumptions about the structure of zones and regions:

  1. regions and zones are hierarchical: zones are strict subsets of regions and no zone can be in 2 regions
  2. zone names are unique across regions; for example region "africa-east-1" might be comprised of zones "africa-east-1a" and "africa-east-1b"

It should be safe to assume that topology labels do not change. Even though labels are strictly mutable, consumers of them can assume that a given node is not going to be moved between zones without being destroyed and recreated.

Kubernetes can use this information in various ways. For example, the scheduler automatically tries to spread the Pods in a ReplicaSet across nodes in a single-zone cluster (to reduce the impact of node failures, see kubernetes.io/hostname ). With multiple-zone clusters, this spreading behavior also applies to zones (to reduce the impact of zone failures). This is achieved via SelectorSpreadPriority .

SelectorSpreadPriority is a best effort placement. If the zones in your cluster are heterogeneous (for example: different numbers of nodes, different types of nodes, or different pod resource requirements), this placement might prevent equal spreading of your Pods across zones. If desired, you can use homogenous zones (same number and types of nodes) to reduce the probability of unequal spreading.

The scheduler (through the VolumeZonePredicate predicate) also will ensure that Pods, that claim a given volume, are only placed into the same zone as that volume. Volumes cannot be attached across zones.

If PersistentVolumeLabel does not support automatic labeling of your PersistentVolumes, you should consider adding the labels manually (or adding support for PersistentVolumeLabel ). With PersistentVolumeLabel , the scheduler prevents Pods from mounting volumes in a different zone. If your infrastructure doesn't have this constraint, you don't need to add the zone labels to the volumes at all.

volume.beta.kubernetes.io/storage-provisioner (deprecated)

Example: volume.beta.kubernetes.io/storage-provisioner: "k8s.io/minikube-hostpath"

Used on: PersistentVolumeClaim

This annotation has been deprecated.

volume.beta.kubernetes.io/storage-class (deprecated)

Example: volume.beta.kubernetes.io/storage-class: "example-class"

Used on: PersistentVolume, PersistentVolumeClaim

This annotation can be used for PersistentVolume(PV) or PersistentVolumeClaim(PVC) to specify the name of StorageClass . When both storageClassName attribute and volume.beta.kubernetes.io/storage-class annotation are specified, the annotation volume.beta.kubernetes.io/storage-class takes precedence over the storageClassName attribute.

This annotation has been deprecated. Instead, set the storageClassName field for the PersistentVolumeClaim or PersistentVolume.

volume.beta.kubernetes.io/mount-options (deprecated)

Example : volume.beta.kubernetes.io/mount-options: "ro,soft"

Used on: PersistentVolume

A Kubernetes administrator can specify additional mount options for when a PersistentVolume is mounted on a node.

This annotation has been deprecated.

volume.kubernetes.io/storage-provisioner

Used on: PersistentVolumeClaim

This annotation will be added to dynamic provisioning required PVC.

volume.kubernetes.io/selected-node

Used on: PersistentVolumeClaim

This annotation is added to a PVC that is triggered by a scheduler to be dynamically provisioned. Its value is the name of the selected node.

volumes.kubernetes.io/controller-managed-attach-detach

Used on: Node

If a node has set the annotation volumes.kubernetes.io/controller-managed-attach-detach on itself, then its storage attach and detach operations are being managed by the volume attach/detach controller running within the kube-controller-manager .

The value of the annotation isn't important; if this annotation exists on a node, then storage attaches and detaches are controller managed.

node.kubernetes.io/windows-build

Example: node.kubernetes.io/windows-build: "10.0.17763"

Used on: Node

When the kubelet is running on Microsoft Windows, it automatically labels its node to record the version of Windows Server in use.

The label's value is in the format "MajorVersion.MinorVersion.BuildNumber".

service.kubernetes.io/headless

Example: service.kubernetes.io/headless: ""

Used on: Service

The control plane adds this label to an Endpoints object when the owning Service is headless.

kubernetes.io/service-name

Example: kubernetes.io/service-name: "my-website"

Used on: EndpointSlice

Kubernetes associates EndpointSlices with Services using this label.

This label records the name of the Service that the EndpointSlice is backing. All EndpointSlices should have this label set to the name of their associated Service.

kubernetes.io/service-account.name

Example: kubernetes.io/service-account.name: "sa-name"

Used on: Secret

This annotation records the name of the ServiceAccount that the token (stored in the Secret of type kubernetes.io/service-account-token ) represents.

kubernetes.io/service-account.uid

Example: kubernetes.io/service-account.uid: da68f9c6-9d26-11e7-b84e-002dc52800da

Used on: Secret

This annotation records the unique ID of the ServiceAccount that the token (stored in the Secret of type kubernetes.io/service-account-token ) represents.

kubernetes.io/legacy-token-last-used

Example: kubernetes.io/legacy-token-last-used: 2022-10-24

Used on: Secret

The control plane only adds this label for Secrets that have the type kubernetes.io/service-account-token . The value of this label records the date (ISO 8601 format, UTC time zone) when the control plane last saw a request where the client authenticated using the service account token.

If a legacy token was last used before the cluster gained the feature (added in Kubernetes v1.26), then the label isn't set.

endpointslice.kubernetes.io/managed-by

Example: endpointslice.kubernetes.io/managed-by: "controller"

Used on: EndpointSlices

The label is used to indicate the controller or entity that manages an EndpointSlice. This label aims to enable different EndpointSlice objects to be managed by different controllers or entities within the same cluster.

endpointslice.kubernetes.io/skip-mirror

Example: endpointslice.kubernetes.io/skip-mirror: "true"

Used on: Endpoints

The label can be set to "true" on an Endpoints resource to indicate that the EndpointSliceMirroring controller should not mirror this resource with EndpointSlices.

service.kubernetes.io/service-proxy-name

Example: service.kubernetes.io/service-proxy-name: "foo-bar"

Used on: Service

The kube-proxy has this label for custom proxy, which delegates service control to custom proxy.

experimental.windows.kubernetes.io/isolation-type (deprecated)

Example: experimental.windows.kubernetes.io/isolation-type: "hyperv"

Used on: Pod

The annotation is used to run Windows containers with Hyper-V isolation. To use Hyper-V isolation feature and create a Hyper-V isolated container, the kubelet should be started with feature gates HyperVContainer=true and the Pod should include the annotation experimental.windows.kubernetes.io/isolation-type: hyperv .

ingressclass.kubernetes.io/is-default-class

Example: ingressclass.kubernetes.io/is-default-class: "true"

Used on: IngressClass

When a single IngressClass resource has this annotation set to "true" , new Ingress resource without a class specified will be assigned this default class.

kubernetes.io/ingress.class (deprecated)

storageclass.kubernetes.io/is-default-class

Example: storageclass.kubernetes.io/is-default-class: "true"

Used on: StorageClass

When a single StorageClass resource has this annotation set to "true" , new PersistentVolumeClaim resource without a class specified will be assigned this default class.

alpha.kubernetes.io/provided-node-ip

Example: alpha.kubernetes.io/provided-node-ip: "10.0.0.1"

Used on: Node

The kubelet can set this annotation on a Node to denote its configured IPv4 address.

When kubelet is started with the --cloud-provider flag set to any value (includes both external and legacy in-tree cloud providers), it sets this annotation on the Node to denote an IP address set from the command line flag ( --node-ip ). This IP is verified with the cloud provider as valid by the cloud-controller-manager.

batch.kubernetes.io/job-completion-index

Example: batch.kubernetes.io/job-completion-index: "3"

Used on: Pod

The Job controller in the kube-controller-manager sets this annotation for Pods created with Indexed completion mode .

kubectl.kubernetes.io/default-container

Example: kubectl.kubernetes.io/default-container: "front-end-app"

The value of the annotation is the container name that is default for this Pod. For example, kubectl logs or kubectl exec without -c or --container flag will use this default container.

kubectl.kubernetes.io/default-logs-container (deprecated)

Example: kubectl.kubernetes.io/default-logs-container: "front-end-app"

The value of the annotation is the container name that is the default logging container for this Pod. For example, kubectl logs without -c or --container flag will use this default container.

endpoints.kubernetes.io/over-capacity

Example: endpoints.kubernetes.io/over-capacity:truncated

Used on: Endpoints

The control plane adds this annotation to an Endpoints object if the associated Service has more than 1000 backing endpoints. The annotation indicates that the Endpoints object is over capacity and the number of endpoints has been truncated to 1000.

If the number of backend endpoints falls below 1000, the control plane removes this annotation.

batch.kubernetes.io/job-tracking (deprecated)

Example: batch.kubernetes.io/job-tracking: ""

Used on: Jobs

The presence of this annotation on a Job indicates that the control plane is tracking the Job status using finalizers . The control plane uses this annotation to safely transition to tracking Jobs using finalizers, while the feature is in development. You should not manually add or remove this annotation.

job-name (deprecated)

Example: job-name: "pi"

Used on: Jobs and Pods controlled by Jobs

controller-uid (deprecated)

Example: controller-uid: "$UID"

Used on: Jobs and Pods controlled by Jobs

batch.kubernetes.io/job-name

Example: batch.kubernetes.io/job-name: "pi"

Used on: Jobs and Pods controlled by Jobs

This label is used as a user-friendly way to get Pods corresponding to a Job. The job-name comes from the name of the Job and allows for an easy way to get Pods corresponding to the Job.

batch.kubernetes.io/controller-uid

Example: batch.kubernetes.io/controller-uid: "$UID"

Used on: Jobs and Pods controlled by Jobs

This label is used as a programmatic way to get all Pods corresponding to a Job.
The controller-uid is a unique identifer that gets set in the selector field so the Job controller can get all the corresponding Pods.

scheduler.alpha.kubernetes.io/defaultTolerations

Example: scheduler.alpha.kubernetes.io/defaultTolerations: '[{"operator": "Equal", "value": "value1", "effect": "NoSchedule", "key": "dedicated-node"}]'

Used on: Namespace

This annotation requires the PodTolerationRestriction admission controller to be enabled. This annotation key allows assigning tolerations to a namespace and any new pods created in this namespace would get these tolerations added.

scheduler.alpha.kubernetes.io/tolerationsWhitelist

Example: scheduler.alpha.kubernetes.io/tolerationsWhitelist: '[{"operator": "Exists", "effect": "NoSchedule", "key": "dedicated-node"}]'

Used on: Namespace

This annotation is only useful when the (alpha) PodTolerationRestriction admission controller is enabled. The annotation value is a JSON document that defines a list of allowed tolerations for the namespace it annotates. When you create a Pod or modify its tolerations, the API server checks the tolerations to see if they are mentioned in the allow list. The pod is admitted only if the check succeeds.

scheduler.alpha.kubernetes.io/preferAvoidPods (deprecated)

Used on: Nodes

This annotation requires the NodePreferAvoidPods scheduling plugin to be enabled. The plugin is deprecated since Kubernetes 1.22. Use Taints and Tolerations instead.

The taints listed below are always used on Nodes

node.kubernetes.io/not-ready

Example: node.kubernetes.io/not-ready: "NoExecute"

The node controller detects whether a node is ready by monitoring its health and adds or removes this taint accordingly.

node.kubernetes.io/unreachable

Example: node.kubernetes.io/unreachable: "NoExecute"

The node controller adds the taint to a node corresponding to the NodeCondition Ready being Unknown .

node.kubernetes.io/unschedulable

Example: node.kubernetes.io/unschedulable: "NoSchedule"

The taint will be added to a node when initializing the node to avoid race condition.

node.kubernetes.io/memory-pressure

Example: node.kubernetes.io/memory-pressure: "NoSchedule"

The kubelet detects memory pressure based on memory.available and allocatableMemory.available observed on a Node. The observed values are then compared to the corresponding thresholds that can be set on the kubelet to determine if the Node condition and taint should be added/removed.

node.kubernetes.io/disk-pressure

Example: node.kubernetes.io/disk-pressure :"NoSchedule"

The kubelet detects disk pressure based on imagefs.available , imagefs.inodesFree , nodefs.available and nodefs.inodesFree (Linux only) observed on a Node. The observed values are then compared to the corresponding thresholds that can be set on the kubelet to determine if the Node condition and taint should be added/removed.

node.kubernetes.io/network-unavailable

Example: node.kubernetes.io/network-unavailable: "NoSchedule"

This is initially set by the kubelet when the cloud provider used indicates a requirement for additional network configuration. Only when the route on the cloud is configured properly will the taint be removed by the cloud provider.

node.kubernetes.io/pid-pressure

Example: node.kubernetes.io/pid-pressure: "NoSchedule"

The kubelet checks D-value of the size of /proc/sys/kernel/pid_max and the PIDs consumed by Kubernetes on a node to get the number of available PIDs that referred to as the pid.available metric. The metric is then compared to the corresponding threshold that can be set on the kubelet to determine if the node condition and taint should be added/removed.

node.kubernetes.io/out-of-service

Example: node.kubernetes.io/out-of-service:NoExecute

A user can manually add the taint to a Node marking it out-of-service. If the NodeOutOfServiceVolumeDetach feature gate is enabled on kube-controller-manager , and a Node is marked out-of-service with this taint, the pods on the node will be forcefully deleted if there are no matching tolerations on it and volume detach operations for the pods terminating on the node will happen immediately. This allows the Pods on the out-of-service node to recover quickly on a different node.

node.cloudprovider.kubernetes.io/uninitialized

Example: node.cloudprovider.kubernetes.io/uninitialized: "NoSchedule"

Sets this taint on a node to mark it as unusable, when kubelet is started with the "external" cloud provider, until a controller from the cloud-controller-manager initializes this node, and then removes the taint.

node.cloudprovider.kubernetes.io/shutdown

Example: node.cloudprovider.kubernetes.io/shutdown: "NoSchedule"

If a Node is in a cloud provider specified shutdown state, the Node gets tainted accordingly with node.cloudprovider.kubernetes.io/shutdown and the taint effect of NoSchedule .

feature.node.kubernetes.io/*

Used on: Nodes Example: feature.node.kubernetes.io/network-sriov.capable: "true"

These features are advertised as labels in the Kubernetes Node object on nodes running NFD. All built-in labels use the feature.node.kubernetes.io label namespace and have the format feature.node.kubernetes.io/<feature-name>: <true> . NFD has many extension points for creating vendor and application-specific labels. For details, see the customization guide .

nfd.node.kubernetes.io/master.version

Example: nfd.node.kubernetes.io/master.version: "v0.6.0"

Used on: Nodes

For node(s) where the Node Feature Discovery (NFD) master is scheduled, this annotation records the version of the NFD master. It is used for informative use only.

nfd.node.kubernetes.io/worker.version

Example: nfd.node.kubernetes.io/worker.version: "v0.4.0"

Used on: Nodes

This annotation records the version for a Node Feature Discovery's worker if there is one running on a node. It's used for informative use only.

nfd.node.kubernetes.io/feature-labels

Example: nfd.node.kubernetes.io/feature-labels: "cpu-cpuid.ADX,cpu-cpuid.AESNI,cpu-hardware_multithreading,kernel-version.full"

Used on: Nodes

This annotation records a comma-separated list of node feature labels managed by Node Feature Discovery (NFD). NFD uses this for an internal mechanism. You should not edit this annotation yourself.

nfd.node.kubernetes.io/extended-resources

Example: nfd.node.kubernetes.io/extended-resources: "accelerator.acme.example/q500,example.com/coprocessor-fx5"

Used on: Nodes

This annotation records a comma-separated list of extended resources managed by Node Feature Discovery (NFD). NFD uses this for an internal mechanism. You should not edit this annotation yourself.

pod-security.kubernetes.io/enforce

Example: pod-security.kubernetes.io/enforce: "baseline"

Used on: Namespace

Value must be one of privileged , baseline , or restricted which correspond to Pod Security Standard levels. Specifically, the enforce label prohibits the creation of any Pod in the labeled Namespace which does not meet the requirements outlined in the indicated level.

See Enforcing Pod Security at the Namespace Level for more information.

pod-security.kubernetes.io/enforce-version

Example: pod-security.kubernetes.io/enforce-version: "1.27"

Used on: Namespace

Value must be latest or a valid Kubernetes version in the format v<MAJOR>.<MINOR> . This determines the version of the Pod Security Standard policies to apply when validating a submitted Pod.

See Enforcing Pod Security at the Namespace Level for more information.

pod-security.kubernetes.io/audit

Example: pod-security.kubernetes.io/audit: "baseline"

Used on: Namespace

Value must be one of privileged , baseline , or restricted which correspond to Pod Security Standard levels. Specifically, the audit label does not prevent the creation of a Pod in the labeled Namespace which does not meet the requirements outlined in the indicated level, but adds an audit annotation to that Pod.

See Enforcing Pod Security at the Namespace Level for more information.

pod-security.kubernetes.io/audit-version

Example: pod-security.kubernetes.io/audit-version: "1.27"

Used on: Namespace

Value must be latest or a valid Kubernetes version in the format v<MAJOR>.<MINOR> . This determines the version of the Pod Security Standard policies to apply when validating a submitted Pod.

See Enforcing Pod Security at the Namespace Level for more information.

pod-security.kubernetes.io/warn

Example: pod-security.kubernetes.io/warn: "baseline"

Used on: Namespace

Value must be one of privileged , baseline , or restricted which correspond to Pod Security Standard levels. Specifically, the warn label does not prevent the creation of a Pod in the labeled Namespace which does not meet the requirements outlined in the indicated level, but returns a warning to the user after doing so. Note that warnings are also displayed when creating or updating objects that contain Pod templates, such as Deployments, Jobs, StatefulSets, etc.

See Enforcing Pod Security at the Namespace Level for more information.

pod-security.kubernetes.io/warn-version

Example: pod-security.kubernetes.io/warn-version: "1.27"

Used on: Namespace

Value must be latest or a valid Kubernetes version in the format v<MAJOR>.<MINOR> . This determines the version of the Pod Security Standard policies to apply when validating a submitted Pod. Note that warnings are also displayed when creating or updating objects that contain Pod templates, such as Deployments, Jobs, StatefulSets, etc.

See Enforcing Pod Security at the Namespace Level for more information.

rbac.authorization.kubernetes.io/autoupdate

Example: rbac.authorization.kubernetes.io/autoupdate: "false"

Used on: ClusterRole, ClusterRoleBinding, Role, RoleBinding

When this annotation is set to "true" on default RBAC objects created by the kube-apiserver, they are automatically updated at server start to add missing permissions and subjects (extra permissions and subjects are left in place). To prevent autoupdating a particular role or rolebinding, set this annotation to "false" . If you create your own RBAC objects and set this annotation to "false" , kubectl auth reconcile (which allows reconciling arbitrary RBAC objects in a manifest ) respects this annotation and does not automatically add missing permissions and subjects.

kubernetes.io/psp (deprecated)

Example: kubernetes.io/psp: restricted

Used on: Pod

This annotation was only relevant if you were using PodSecurityPolicies . Kubernetes v1.27 does not support the PodSecurityPolicy API.

When the PodSecurityPolicy admission controller admitted a Pod, the admission controller modified the Pod to have this annotation. The value of the annotation was the name of the PodSecurityPolicy that was used for validation.

seccomp.security.alpha.kubernetes.io/pod (non-functional)

Older versions of Kubernetes allowed you to configure seccomp behavior using this annotation . See Restrict a Container's Syscalls with seccomp to learn the supported way to specify seccomp restrictions for a Pod.

container.seccomp.security.alpha.kubernetes.io/[NAME] (non-functional)

Older versions of Kubernetes allowed you to configure seccomp behavior using this annotation . See Restrict a Container's Syscalls with seccomp to learn the supported way to specify seccomp restrictions for a Pod.

snapshot.storage.kubernetes.io/allow-volume-mode-change

Example: snapshot.storage.kubernetes.io/allow-volume-mode-change: "true"

Used on: VolumeSnapshotContent

Value can either be true or false . This determines whether a user can modify the mode of the source volume when a PersistentVolumeClaim is being created from a VolumeSnapshot.

Refer to Converting the volume mode of a Snapshot and the Kubernetes CSI Developer Documentation for more information.

scheduler.alpha.kubernetes.io/critical-pod (deprecated)

Example: scheduler.alpha.kubernetes.io/critical-pod: ""

Used on: Pod

This annotation lets Kubernetes control plane know about a pod being a critical pod so that the descheduler will not remove this pod.

Annotations used for audit

See more details on the Audit Annotations page.

kubeadm

kubeadm.alpha.kubernetes.io/cri-socket

Example: kubeadm.alpha.kubernetes.io/cri-socket: unix:///run/containerd/container.sock

Used on: Node

Annotation that kubeadm uses to preserve the CRI socket information given to kubeadm at init / join time for later use. kubeadm annotates the Node object with this information. The annotation remains "alpha", since ideally this should be a field in KubeletConfiguration instead.

kubeadm.kubernetes.io/etcd.advertise-client-urls

Example: kubeadm.kubernetes.io/etcd.advertise-client-urls: https://172.17.0.18:2379

Used on: Pod

Annotation that kubeadm places on locally managed etcd pods to keep track of a list of URLs where etcd clients should connect to. This is used mainly for etcd cluster health check purposes.

kubeadm.kubernetes.io/kube-apiserver.advertise-address.endpoint

Example: kubeadm.kubernetes.io/kube-apiserver.advertise-address.endpoint: https://172.17.0.18:6443

Used on: Pod

Annotation that kubeadm places on locally managed kube-apiserver pods to keep track of the exposed advertise address/port endpoint for that API server instance.

kubeadm.kubernetes.io/component-config.hash

Used on: ConfigMap

Example: kubeadm.kubernetes.io/component-config.hash: 2c26b46b68ffc68ff99b453c1d30413413422d706483bfa0f98a5e886266e7ae

Annotation that kubeadm places on ConfigMaps that it manages for configuring components. It contains a hash (SHA-256) used to determine if the user has applied settings different from the kubeadm defaults for a particular component.

node-role.kubernetes.io/control-plane

Used on: Node

A marker label to indicate that the node is used to run control plane components. The kubeadm tool applies this label to the control plane nodes that it manages. Other cluster management tools typically also set this taint.

You can label control plane nodes with this label to make it easier to schedule Pods only onto these nodes, or to avoid running Pods on the control plane. If this label is set, EndpointSlice controller ignores that node while calculating Topology Aware Hints.

node-role.kubernetes.io/control-plane

Used on: Node

Taint that kubeadm applies on control plane nodes to restrict placing pods and allow only specific pods to schedule on them.

Example: node-role.kubernetes.io/control-plane:NoSchedule

If this Taint is applied, control plane nodes allow only critical workloads to schedule on them. You can manually remove this taint with the following command on a specific node. 

kubectl taint nodes <node-name> node-role.kubernetes.io/control-plane:NoSchedule-

node-role.kubernetes.io/master (deprecated)

Used on: Node

Example: node-role.kubernetes.io/master:NoSchedule

Taint that kubeadm previously applied on control plane nodes to allow only critical workloads to schedule on them. Replaced by node-role.kubernetes.io/control-plane ; kubeadm no longer sets or uses this deprecated taint.

Last modified March 31, 2023 at 7:58 PM PST: Documenting node discover annotations (9458ee79df)