KUBERNETES(1)(kubernetes)

📦
version 1.37.2 · 10 versions
, in arch rolling, fedora 43, fedora 44, nixpkgs unstable

💖 KUBERNETES(1)(kubernetes) 💖

okay so bestie, welcome to the kube-controller-manager manual page — it lives in section 1 of the kubernetes manual, and it was originally written by Eric Paris, Jan 2015 (iconic of him, honestly). This is part 1 of 5, so don't you dare close the tab, we're just getting started ✨

🌸 NAME 🌸

kube-controller-manager -

ngl bestie the description here is just… blank. The source page really does end at that little dash. No cap. Moving on 💅

🦋 SYNOPSIS 🦋

kube-controller-manager [OPTIONS]

✨ DESCRIPTION ✨

okay so picture this: the Kubernetes controller manager is a daemon that has the core control loops shipped with Kubernetes just… embedded in it. Living inside it. Main character energy fr fr.

If you've ever done robotics or automation, you already know what a control loop is: it's a non-terminating loop that regulates the state of a system. It just goes and goes and goes, forever, keeping things in check. 🌀

In Kubernetes, a controller is exactly that kind of control loop — it watches the shared state of the cluster through the apiserver, and then makes changes trying to move the current state towards the desired state. It's giving glow-up arc for your cluster, honestly.

Controllers that ship with Kubernetes today include the replication controller, the endpoints controller, the namespace controller, and the serviceaccounts controller. Obsessed with all of them, truly. 👑

💅 OPTIONS 💅

  • --allocate-node-cidrs=false — should CIDRs for Pods actually be allocated and set on the cloud provider? Requires --cluster-cidr though, so don't come crying if you skip it.
  • --allow-metric-labels=[] — the map from metric-label to value allow-list of this label. The key's format is ,. The value's format is ,... — e.g. metric1,label1='v1,v2,v3', metric1,label2='v1,v2,v3' metric2,label1='v1,v2,v3'.
  • --allow-metric-labels-manifest="" — the path to the manifest file that contains the allow-list mapping. The format of the file is the same as the flag --allow-metric-labels, i.e.:
allowListMapping:
  "metric1,label1": "value11,value12"
  "metric2,label2": ""
heads up bestie: the flag --allow-metric-labels will override the manifest file. 🫶
  • --allow-untagged-cloud=false — allow the cluster to run without the cluster-id on cloud instances. This is a legacy mode of operation and a cluster-id will be required in the future. So like… it works, but it's giving expired trend.
  • --attach-detach-reconcile-sync-period=1m0s — the reconciler sync wait time between volume attach detach. This duration must be larger than one second, and increasing this value from the default may allow for volumes to be mismatched with pods. So don't get cute with it. 😭
  • --authentication-kubeconfig="" — kubeconfig file pointing at the 'core' kubernetes server with enough rights to create tokenreviews.authentication.k8s.io. This is optional. If empty, all token requests are considered to be anonymous and no client CA is looked up in the cluster.
  • --authentication-skip-lookup=false — if false, the authentication-kubeconfig will be used to lookup missing authentication configuration from the cluster.
  • --authentication-token-webhook-cache-ttl=10s — the duration to cache responses from the webhook token authenticator.
  • --authentication-tolerate-lookup-failure=false — if true, failures to look up missing authentication configuration from the cluster are not considered fatal. Note that this can result in authentication that treats all requests as anonymous. 💀
  • --authorization-always-allow-paths=[/healthz,/readyz,/livez] — a list of HTTP paths to skip during authorization, i.e. these are authorized without contacting the 'core' kubernetes server.
  • --authorization-kubeconfig="" — kubeconfig file pointing at the 'core' kubernetes server with enough rights to create subjectaccessreviews.authorization.k8s.io. This is optional. If empty, all requests not skipped by authorization are forbidden. periodt.
  • --authorization-webhook-cache-authorized-ttl=10s — the duration to cache 'authorized' responses from the webhook authorizer.
  • --authorization-webhook-cache-unauthorized-ttl=10s — the duration to cache 'unauthorized' responses from the webhook authorizer.
  • --bind-address=0.0.0.0 — the IP address on which to listen for the --secure-port port. The associated interface(s) must be reachable by the rest of the cluster, and by CLI/web clients. If blank or an unspecified address (0.0.0.0 or ::), all interfaces and IP address families will be used.
  • --cert-dir="" — the directory where the TLS certs are located. If --tls-cert-file and --tls-private-key-file are provided, this flag will be ignored.
  • --cidr-allocator-type="RangeAllocator" — type of CIDR allocator to use.
  • --client-ca-file="" — if set, any request presenting a client certificate signed by one of the authorities in the client-ca-file is authenticated with an identity corresponding to the CommonName of the client certificate.
  • --cloud-config="" — the path to the cloud provider configuration file. Empty string for no configuration file.
  • --cloud-provider="" — the provider for cloud services. Empty string for no provider.
  • --cluster-cidr="" — CIDR Range for Pods in cluster. Only used when --allocate-node-cidrs=true; if false, this option will be ignored.
  • --cluster-name="kubernetes" — the instance prefix for the cluster.
  • --cluster-signing-cert-file="" — filename containing a PEM-encoded X509 CA certificate used to issue cluster-scoped certificates. If specified, no more specific --cluster-signing-* flag may be specified. pick a lane bestie 🎀
  • --cluster-signing-duration=8760h0m0s — the max length of duration signed certificates will be given. Individual CSRs may request shorter certs by setting spec.expirationSeconds.
  • --cluster-signing-key-file="" — filename containing a PEM-encoded RSA or ECDSA private key used to sign cluster-scoped certificates. If specified, no more specific --cluster-signing-* flag may be specified.
  • --cluster-signing-kube-apiserver-client-cert-file="" — filename containing a PEM-encoded X509 CA certificate used to issue certificates for the kubernetes.io/kube-apiserver-client signer. If specified, --cluster-signing-{cert,key}-file must not be set.
  • --cluster-signing-kube-apiserver-client-key-file="" — filename containing a PEM-encoded RSA or ECDSA private key used to sign certificates for the kubernetes.io/kube-apiserver-client signer. If specified, --cluster-signing-{cert,key}-file must not be set.
  • --cluster-signing-kubelet-client-cert-file="" — filename containing a PEM-encoded X509 CA certificate used to issue certificates for the kubernetes.io/kube-apiserver-client-kubelet signer. If specified, --cluster-signing-{cert,key}-file must not be set.
  • --cluster-signing-kubelet-client-key-file="" — filename containing a PEM-encoded RSA or ECDSA private key used to sign certificates for the kubernetes.io/kube-apiserver-client-kubelet signer. If specified, --cluster-signing-{cert,key}-file must not be set.
  • --cluster-signing-kubelet-serving-cert-file="" — filename containing a PEM-encoded X509 CA certificate used to issue certificates for the kubernetes.io/kubelet-serving signer. If specified, --cluster-signing-{cert,key}-file must not be set.
  • --cluster-signing-kubelet-serving-key-file="" — filename containing a PEM-encoded RSA or ECDSA private key used to sign certificates for the kubernetes.io/kubelet-serving signer. If specified, --cluster-signing-{cert,key}-file must not be set.
  • --cluster-signing-legacy-unknown-cert-file="" — filename containing a PEM-encoded X509 CA certificate used to issue certificates for the kubernetes.io/legacy-unknown signer. If specified, --cluster-signing-{cert,key}-file must not be set.
  • --cluster-signing-legacy-unknown-key-file="" — filename containing a PEM-encoded RSA or ECDSA private key used to sign certificates for the kubernetes.io/legacy-unknown signer. If specified, --cluster-signing-{cert,key}-file must not be set.
  • --concurrent-cron-job-syncs=5 — the number of cron job objects that are allowed to sync concurrently. Larger number = more responsive jobs, but more CPU (and network) load.
  • --concurrent-daemonset-syncs=2 — the number of daemonset objects that are allowed to sync concurrently. Larger number = more responsive daemonsets, but more CPU (and network) load.
  • --concurrent-deployment-syncs=5 — the number of deployment objects that are allowed to sync concurrently. Larger number = more responsive deployments, but more CPU (and network) load.
  • --concurrent-device-taint-eviction-syncs=8 — the number of operations (evicting pods, updating DeviceTaintRule status) allowed to run concurrently. Greater number = more responsive, but more CPU (and network) load.
  • --concurrent-disruption-syncs=1 — the number of PDB objects that are allowed to sync concurrently. Larger number = more responsive PDB updates, but more CPU (and network) load.
  • --concurrent-endpoint-syncs=5 — the number of endpoint syncing operations that will be done concurrently. Larger number = faster endpoint updating, but more CPU (and network) load.
  • --concurrent-ephemeralvolume-syncs=5 — the number of ephemeral volume syncing operations that will be done concurrently. Larger number = faster ephemeral volume updating, but more CPU (and network) load.
  • --concurrent-gc-syncs=20 — the number of garbage collector workers that are allowed to sync concurrently.
  • --concurrent-horizontal-pod-autoscaler-syncs=5 — the number of horizontal pod autoscaler objects that are allowed to sync concurrently. Larger number = more responsive horizontal pod autoscaler objects processing, but more CPU (and network) load.
  • --concurrent-job-syncs=5 — the number of job objects that are allowed to sync concurrently. Larger number = more responsive jobs, but more CPU (and network) load.
  • --concurrent-namespace-syncs=10 — the number of namespace objects that are allowed to sync concurrently. Larger number = more responsive namespace termination, but more CPU (and network) load.
  • --concurrent-replicaset-syncs=5 — the number of replica sets that are allowed to sync concurrently. Larger number = more responsive replica management, but more CPU (and network) load.
  • --concurrent-resource-quota-syncs=5 — the number of resource quotas that are allowed to sync concurrently. Larger number = more responsive quota management, but more CPU (and network) load.
  • --concurrent-resourceclaim-syncs=50 — the number of operations (creating or deleting ResourceClaims) allowed to run concurrently. Larger number = more responsive, but more CPU (and network) load.
  • --concurrent-service-endpoint-syncs=5 — the number of service endpoint syncing operations that will be done concurrently. Larger number = faster endpoint slice updating, but more CPU (and network) load. Defaults to 5.
  • --concurrent-serviceaccount-token-syncs=5 — the number of service account token objects that are allowed to sync concurrently. Larger number = more responsive token generation, but more CPU (and network) load.
  • --concurrent-statefulset-syncs=5 — the number of statefulset objects that are allowed to sync concurrently. Larger number = more responsive statefulsets, but more CPU (and network) load.
  • --concurrent-ttl-after-finished-syncs=5 — the number of ttl-after-finished-controller workers that are allowed to sync concurrently.

okay and there is so much more coming — this list does not stop here, bestie. keep scrolling, the next part is right there 🦋✨

  • --concurrent-validating-admission-policy-status-syncs=5 — the number of ValidatingAdmissionPolicyStatusController workers that are allowed to sync concurrently. That's it, that's the whole thing ✨
  • --concurrent_rc_syncs=5 — the number of replication controllers allowed to sync concurrently. Bigger number = more responsive replica management, but also more CPU (and network) load. So like, pick your struggle 💅
  • --configure-cloud-routes=true — should the CIDRs allocated by allocate-node-cidrs be configured on the cloud provider. Yes or no bestie 🫶
  • --contention-profiling=false — enable block profiling, but only if profiling is already enabled. It's giving "extra credit, conditional on the first thing" 📊
  • --controller-shutdown-timeout=10s — how long to wait for the controllers to shut down before terminating the executable. A little grace period before the lights go off 😭
  • --controller-start-interval=0s — the interval between starting controller managers. Stagger them if you must ⏱️
  • --controllers=[] — the list of controllers to enable, and this one is lowkey iconic. '' enables all on-by-default controllers, 'foo' enables the controller named 'foo', and '-foo' disables the controller named 'foo'. Here's the full roster of all controllers:
    • All controllers: bootstrap-signer-controller, certificatesigningrequest-approving-controller, certificatesigningrequest-cleaner-controller, certificatesigningrequest-signing-controller, clusterrole-aggregation-controller, cronjob-controller, daemonset-controller, deployment-controller, device-taint-eviction-controller, disruption-controller, endpoints-controller, endpointslice-controller, endpointslice-mirroring-controller, ephemeral-volume-controller, garbage-collector-controller, horizontal-pod-autoscaler-controller, job-controller, kube-apiserver-serving-clustertrustbundle-publisher-controller, legacy-serviceaccount-token-cleaner-controller, namespace-controller, node-ipam-controller, node-lifecycle-controller, persistentvolume-attach-detach-controller, persistentvolume-binder-controller, persistentvolume-expander-controller, persistentvolume-protection-controller, persistentvolumeclaim-protection-controller, pod-garbage-collector-controller, podcertificaterequest-cleaner-controller, podgroup-protection-controller, replicaset-controller, replicationcontroller-controller, resourceclaim-controller, resourcepoolstatusrequest-controller, resourcequota-controller, root-ca-certificate-publisher-controller, selinux-warning-controller, service-cidr-controller, serviceaccount-controller, serviceaccount-token-controller, statefulset-controller, storage-version-migrator-controller, storageversion-garbage-collector-controller, taint-eviction-controller, token-cleaner-controller, ttl-after-finished-controller, ttl-controller, validatingadmissionpolicy-status-controller, volumeattributesclass-protection-controller
    • Disabled-by-default controllers: bootstrap-signer-controller, selinux-warning-controller, token-cleaner-controller
  • --disable-attach-detach-reconcile-sync=false — disable the volume attach detach reconciler sync.
bestie. ⚠️ Disabling this may cause volumes to be mismatched with pods. Use wisely. Like, wisely wisely. 🙈
  • --disable-force-detach-on-timeout=false — prevent force detaching volumes based on maximum unmount time and node status.
🚨 If this flag is set to true, the non-graceful node shutdown feature must be used to recover from node failure. See https://k8s.io/docs/storage-disable-force-detach-on-timeout/
  • --disable-http2-serving=false — if true, HTTP2 serving will be disabled [default=false]. Simple as 🌸
  • --disabled-metrics=[] — this flag is your escape hatch for misbehaving metrics 🕳️ You must provide the fully qualified metric name in order to disable it. Disclaimer: disabling metrics is higher in precedence than showing hidden metrics. iykyk 👑
  • --emulated-version=[] — the versions that different components emulate their capabilities (APIs, features, ...) of. If set, the component will emulate the behavior of this version instead of the underlying binary version. Version format could only be major.minor, for example:
--emulated-version=wardle=1.2,kube=1.31

Options are: kube=1.34..1.37(default:1.37). If the component is not specified, defaults to "kube" 💜

  • --enable-dynamic-provisioning=true — enable dynamic provisioning for environments that support it. If your env supports it, let it cook 🔥
  • --enable-garbage-collector=true — enables the generic garbage collector.
MUST be synced with the corresponding flag of the kube-apiserver. Fr fr, they gotta match or it's chaos 💀
  • --enable-hostpath-provisioner=false — enable HostPath PV provisioning when running without a cloud provider. This allows testing and development of provisioning features.
🚫 HostPath provisioning is not supported in any way, won't work in a multi-node cluster, and should not be used for anything other than testing or development. Not for prod, bestie, I'm begging 🙏
  • --enable-leader-migration=false — whether to enable controller leader migration. The passing-of-the-crown feature 👑
  • --endpoint-updates-batch-period=0s — the length of the endpoint updates batching period. Processing of pod changes will be delayed by this duration to join them with potential upcoming updates and reduce the overall number of endpoints updates. Larger number = higher endpoint programming latency, but lower number of endpoints revision generated. Trade-offs, always trade-offs 🦋
  • --endpointslice-updates-batch-period=0s — the length of the endpoint slice updates batching period. Processing of pod changes will be delayed by this duration to join them with potential upcoming updates and reduce the overall number of endpoints updates. Larger number = higher endpoint programming latency, but lower number of endpoints revision generated. Same vibe, slice edition 🌈
  • --external-cloud-volume-plugin="" — the plugin to use when the cloud provider is set to external. Can be empty, and should only be set when cloud-provider is external. Currently used to allow node-ipam-controller, persistentvolume-binder-controller, persistentvolume-expander-controller and attach-detach-controller to work for in tree cloud providers ✨
  • --feature-gates= — okay bestie, this is THE big one, the main character of the flags. It's a comma-separated list of component:key=value pairs that describe feature gates for alpha/experimental features across different components. If you don't name a component, it defaults to "kube". You can invoke this flag over and over again (it stacks, it's generous like that). For example:
--feature-gates 'wardle:featureA=true,wardle:featureB=false' --feature-gates 'kube:featureC=true'

And the options are… *deep breath*… literally all of these 💅:

  • kube:APIResponseCompression=true|false (BETA - default=true)
  • kube:APIServerIdentity=true|false (BETA - default=true)
  • kube:APIServerWebhookAuthenticationToken=true|false (ALPHA - default=false)
  • kube:APIServingWithRoutine=true|false (ALPHA - default=false)
  • kube:AllAlpha=true|false (ALPHA - default=false)
  • kube:AllBeta=true|false (BETA - default=false)
  • kube:AllowParsingUserUIDFromCertAuth=true|false (BETA - default=true)
  • kube:AllowUnsafeMalformedObjectDeletion=true|false (BETA - default=true)
  • kube:AtomicFIFO=true|false (BETA - default=true)
  • kube:AtomicWriteVolumeUserFields=true|false (ALPHA - default=false)
  • kube:AuthorizePodWebsocketUpgradeCreatePermission=true|false (BETA - default=true)
  • kube:CBORServingAndStorage=true|false (ALPHA - default=false)
  • kube:CPUManagerPolicyAlphaOptions=true|false (ALPHA - default=false)
  • kube:CPUManagerPolicyBetaOptions=true|false (BETA - default=true)
  • kube:CRDObservedGenerationTracking=true|false (BETA - default=true)
  • kube:CRIListStreaming=true|false (ALPHA - default=false)
  • kube:CSIVolumeHealth=true|false (ALPHA - default=false)
  • kube:ClearingNominatedNodeNameAfterBinding=true|false (BETA - default=true)
  • kube:ClientsAllowCARotation=true|false (BETA - default=true)
  • kube:ClientsAllowCBOR=true|false (ALPHA - default=false)
  • kube:ClientsAllowTLSCacheGC=true|false (BETA - default=true)
  • kube:ClientsPreferCBOR=true|false (ALPHA - default=false)
  • kube:CloudControllerManagerWatchBasedRoutesReconciliation=true|false (ALPHA - default=false)
  • kube:CloudControllerManagerWebhook=true|false (ALPHA - default=false)
  • kube:ComponentFlagz=true|false (BETA - default=true)
  • kube:ComponentStatusz=true|false (BETA - default=true)
  • kube:CompositePodGroup=true|false (ALPHA - default=false)
  • kube:ConcurrentWatchObjectDecode=true|false (BETA - default=true)
  • kube:ConsistentListFromCacheSkipTimeoutFallback=true|false (ALPHA - default=false)
  • kube:ConstrainedImpersonation=true|false (BETA - default=true)
  • kube:ContainerCheckpoint=true|false (BETA - default=true)
  • kube:ContainerRestartRules=true|false (BETA - default=true)
  • kube:ContainerStopSignals=true|false (ALPHA - default=false)
  • kube:ContextualLogging=true|false (BETA - default=true)
  • kube:ControllerManagerReleaseLeaderElectionLockOnExit=true|false (ALPHA - default=false)
  • kube:CoordinatedLeaderElection=true|false (BETA - default=false)
  • kube:CrossNamespaceVolumeDataSource=true|false (ALPHA - default=false)
  • kube:DRAConsumableCapacity=true|false (BETA - default=true)
  • kube:DRADerivedAttributes=true|false (ALPHA - default=false)
  • kube:DRADeviceBindingConditions=true|false (BETA - default=true)
  • kube:DRADeviceCompatibilityGroups=true|false (ALPHA - default=false)
  • kube:DRAFractionalCapacityRange=true|false (BETA - default=false), or DRAFractionalCapacityRange=true|false (BETA - default=true) if --min-compatibility-version>=1.37
  • kube:DRAListTypeAttributes=true|false (ALPHA - default=false)
  • kube:DRANodeAllocatableResources=true|false (ALPHA - default=false)
  • kube:DRAOptionalNodeOperations=true|false (ALPHA - default=false)
  • kube:DRAPartitionableDevices=true|false (BETA - default=true)
  • kube:DRAPartitionableDevicesType=true|false (ALPHA - default=false)
  • kube:DRAResourceClaimGranularStatusAuthorization=true|false (BETA - default=true)
  • kube:DRAResourcePoolStatus=true|false (ALPHA - default=false)
  • kube:DRASchedulerFilterTimeout=true|false (BETA - default=true)
  • kube:DRAWorkloadResourceClaims=true|false (BETA - default=false)
  • kube:DeclarativeValidationBeta=true|false (BETA - default=true)
  • kube:DefaultPodSysctls=true|false (ALPHA - default=false)
  • kube:DeploymentReplicaSetTerminatingReplicas=true|false (BETA - default=true)
  • kube:DetectCacheInconsistency=true|false (BETA - default=true)
  • kube:EmptyDirVolumeMode=true|false (ALPHA - default=false)
  • kube:EnvFiles=true|false (BETA - default=true)
  • kube:EtcdRangeStream=true|false (BETA - default=true)
  • kube:EventedPLEG=true|false (ALPHA - default=false)
  • kube:EvictionRequestAPI=true|false (ALPHA - default=false)
  • kube:ExcludeAdmissionWebhookVirtualResources=true|false (BETA - default=true)
  • kube:ExtendWebSocketsToKubelet=true|false (BETA - default=true)
  • kube:GRPCContainerProbeTLS=true|false (ALPHA - default=false)
  • kube:GenericWorkload=true|false (BETA - default=false)
  • kube:GracefulNodeShutdown=true|false (BETA - default=true)
  • kube:GracefulNodeShutdownBasedOnPodPriority=true|false (BETA - default=true)
  • kube:H2CContainerProbe=true|false (ALPHA - default=false)
  • kube:HPAGeneration=true|false (BETA - default=true)
  • kube:HPAOptimizedSelectorStore=true|false (BETA - default=true)
  • kube:HPAScaleToZero=true|false (BETA - default=true)
  • kube:HugepageAwareEviction=true|false (BETA - default=true)
  • kube:ImageVolumeWithDigest=true|false (ALPHA - default=false)
  • kube:InOrderInformersBatchProcess=true|false (BETA - default=true)
  • kube:InPlacePodLevelResourcesVerticalScaling=true|false (BETA - default=true)
  • kube:InPlacePodVerticalScalingExclusiveCPUs=true|false (ALPHA - default=false)
  • kube:InPlacePodVerticalScalingExclusiveMemory=true|false (ALPHA - default=false)
  • kube:InPlacePodVerticalScalingMemoryBackedVolumes=true|false (ALPHA - default=false)
  • kube:InPlacePodVerticalScalingSchedulerPreemption=true|false (ALPHA - default=false)
  • kube:InterPodAffinityHostnameFastPath=true|false (ALPHA - default=false)
  • kube:KubeProxyNFTablesLocalhostNodePorts=true|false (ALPHA - default=false)
  • kube:KubeletAllocatedPodsEndpoint=true|false (ALPHA - default=false)
  • kube:KubeletCrashLoopBackOffMax=true|false (BETA - default=true)
  • kube:KubeletEnsureSecretPulledImages=true|false (BETA - default=true)
  • kube:KubeletInUserNamespace=true|false (BETA - default=true)
  • kube:KubeletSeparateDiskGC=true|false (BETA - default=true)
  • kube:KubeletServiceAccountTokenForCredentialProviders=true|false (BETA - default=true)
  • kube:ListFromCacheSnapshot=true|false (BETA - default=true)
  • kube:LocalStorageCapacityIsolationFSQuotaMonitoring=true|false (BETA - default=false)
  • kube:LoggingAlphaOptions=true|false (ALPHA - default=false)
  • kube:LoggingBetaOptions=true|false (BETA - default=true)
  • kube:ManifestBasedAdmissionControlConfig=true|false (BETA - default=true)
  • kube:MatchLabelKeysInPodTopologySpread=true|false (BETA - default=true)
  • kube:MatchLabelKeysInPodTopologySpreadSelectorMerge=true|false (BETA - default=true)
  • kube:MaxUnavailableStatefulSet=true|false (BETA - default=true)
  • kube:MemoryQoS=true|false (BETA - default=true)
  • kube:MutablePVNodeAffinity=true|false (ALPHA - default=false)
  • kube:MutablePodResourcesForSuspendedJobs=true|false (BETA - default=true)
  • kube:MutableSchedulingDirectivesForSuspendedJobs=true|false (BETA - default=true)
  • kube:NFTablesNetlink=true|false (BETA - default=true)
  • kube:NativeHistograms=true|false (BETA - default=true)
  • kube:NodeControllerLeaseCircuitBreaker=true|false (BETA - default=true)
  • kube:NodeLifecycleConditions=true|false (ALPHA - default=false)
  • kube:NominatedNodeNameForExpectation=true|false (BETA - default=true)
  • kube:OpenAPIEnums=true|false (BETA - default=true)
  • kube:OpportunisticBatching=true|false (BETA - default=true)
  • kube:PersistentVolumeClaimUnusedSinceTime=true|false (BETA - default=true)
  • kube:PodAndContainerStatsFromCRI=true|false (BETA - default=false)
  • kube:PodDeletionCost=true|false (BETA - default=true)
  • kube:PodGroupPreemptionPolicy=true|false (ALPHA - default=false)
  • kube:PodLevelResourceManagers=true|false (BETA - default=false)
  • kube:PodLevelResources=true|false (BETA - default=true)
  • kube:PodLevelResourcesFixDefaulting=true|false (BETA - default=true)
  • kube:PodLevelResourcesFixKubeletQOSClass=true|false (BETA - default=true)
  • kube:PodLogsQuerySplitStreams=true|false (ALPHA - default=false)
  • kube:PodTopologyLabelsAdmission=true|false (BETA - default=true)
  • kube:PodsAPI=true|false (BETA - default=true)
  • kube:PortForwardWebsockets=true|false (BETA - default=true)
  • kube:QOSReserved=true|false (ALPHA - default=false)
  • kube:ReduceDefaultCrashLoopBackOffDecay=true|false (ALPHA - default=false)
  • kube:ReloadKubeletClientCAFile=true|false (BETA - default=true)
  • kube:ReloadKubeletServerCertificateFile=true|false (BETA - default=true)
  • kube:RemoteRequestHeaderUID=true|false (BETA - default=true)
  • kube:ResourceHealthStatus=true|false (BETA - default=true)
  • kube:ResourceHealthStatusMessage=true|false (BETA - default=true)
  • kube:RestartAllContainersOnContainerExits=true|false (BETA - default=true)
  • kube:RotateKubeletServerCertificate=true|false (BETA - default=true)
  • kube:RuntimeClassInImageCriApi=true|false (ALPHA - default=false)
  • kube:SchedulerAsyncAPICalls=true|false (BETA - default=false)
  • kube:SchedulerAsyncPreemption=true|false (BETA - default=true)
  • kube:SchedulerPopFromBackoffQ=true|false (BETA - default=true)
  • kube:SchedulerPreQueueingHints=true|false (ALPHA - default=false)
  • kube:ServiceAccountNodeAudienceRestriction=true|false (BETA - default=true)
  • kube:ShardedListAndWatch=true|false (ALPHA - default=false)
  • kube:SizeBasedListCostEstimate=true|false (BETA - default=true)
  • kube:StaleControllerConsistencyDaemonSet=true|false (BETA - default=true)
  • kube:StaleControllerConsistencyHPA=true|false (BETA - default=true)
  • kube:StaleControllerConsistencyJob=true|false (BETA - default=true)
  • kube:StaleControllerConsistencyReplicaSet=true|false (BETA - default=true)
  • kube:StaleControllerConsistencyStatefulSet=true|false (BETA - default=true)
  • kube:StatefulSetRecreateStrategy=true|false (ALPHA - default=false)
  • kube:StatefulSetSemanticRevisionComparison=true|false (BETA - default=true)
  • kube:StorageCapacityScoring=true|false (BETA - default=true)
  • kube:StorageVersionAPI=true|false (ALPHA - default=false)
  • kube:StorageVersionHash=true|false (BETA - default=true)
  • kube:StrictIPCIDRValidation=true|false (BETA - default=true)
  • kube:StructuredAuthenticationConfigurationEgressSelector=true|false (BETA - default=true)
  • kube:StructuredAuthenticationConfigurationJWKSMetrics=true|false (BETA - default=true)
  • kube:TaintTolerationComparisonOperators=true|false (ALPHA - default=false)
  • kube:TokenRequestServiceAccountUIDValidation=true|false (BETA - default=true)
  • kube:TopologyAwareWorkloadScheduling=true|false (ALPHA - default=false)
  • kube:TopologyManagerPolicyAlphaOptions=true|false (ALPHA - default=false)
  • kube:TopologyManagerPolicyBetaOptions=true|false (BETA - default=true)
  • kube:TranslateStreamCloseWebsocketRequests=true|false (BETA - default=true)
  • kube:UnauthenticatedHTTP2DOSMitigation=true|false (BETA - default=true)
  • kube:UnknownVersionInteroperabilityProxy=true|false (BETA - default=true)
  • kube:UnlockWhileProcessingFIFO=true|false (BETA -
okay that last one is exactly where the original text runs out, bestie — we're not inventing the rest, we report what's there 💀
  • kube:UserNamespacesHostNetworkSupport=true|false (ALPHA - default=false)
  • kube:VolumeBindMountOptions=true|false (ALPHA - default=false)
  • kube:VolumeLimitScaling=true|false (BETA - default=true)
  • kube:WatchList=true|false (BETA - default=true)
  • kube:WatchListClient=true|false (BETA - default=true)
  • kube:WatchListCompression=true|false (BETA - default=true)
  • kube:WebhookRoundTripLoadBalancing=true|false (BETA - default=true)
  • kube:WindowsCPUAndMemoryAffinity=true|false (ALPHA - default=false)
  • kube:WindowsGracefulNodeShutdown=true|false (BETA - default=true)
  • kube:WorkloadWithJob=true|false (ALPHA - default=false)

okay bestie the feature gate saga is finally over, we made it, and now we're sliding into the flags proper. buckle up because there are a LOT of them and they all matter 💅

  • --flex-volume-plugin-dir="/usr/libexec/kubernetes/kubelet-plugins/volume/exec/" — the full path of the directory where the flex volume plugin goes looking for extra third-party volume plugins. like, that's the whole vibe. ✨
  • -h, --help=false — shows the help for kube-controller-manager. classic. iconic. we love a help flag.
  • --horizontal-pod-autoscaler-cpu-initialization-period=5m0s — the period after a pod starts where CPU samples might just get skipped entirely. she's warming up, leave her alone 🫶
  • --horizontal-pod-autoscaler-downscale-stabilization=5m0s — the period the autoscaler looks backwards through, and it will NOT scale down below any recommendation it already made during that window. she doesn't backtrack, periodt.
  • --horizontal-pod-autoscaler-initial-readiness-delay=30s — the period after a pod starts where any readiness changes get treated as initial readiness. first impressions era 💜
  • --horizontal-pod-autoscaler-sync-period=15s — how often it syncs the pod count in the horizontal pod autoscaler. check-in szn 🎀
  • --horizontal-pod-autoscaler-tolerance=0.1 — the minimum change (measured from 1.0) in the desired-to-actual metrics ratio before the HPA considers scaling at all. tiny changes? not the moment. 💅
  • --http2-max-streams-per-connection=0 — the limit the server hands to clients for max streams in one HTTP/2 connection. Zero means "just use golang's default, bestie". 🌸
  • --kube-api-burst=30 — the burst to use while talking with the kubernetes apiserver. she's chatty. 🦋
  • --kube-api-content-type="application/vnd.kubernetes.protobuf" — the content type of the requests sent to the apiserver. keep it exactly like that, no editing fr fr.
  • --kube-api-qps=20 — the QPS to use while talking with the kubernetes apiserver. throttled but make it cute. ✨
  • --kubeconfig="" — path to the kubeconfig file holding your authorization + master location info (and lowkey, the master location can get overridden by the master flag). 👑
  • --large-cluster-size-threshold=50 — the number of nodes from which node-lifecycle-controller starts treating the cluster as "large" for eviction logic purposes. and here's the tea: --secondary-node-eviction-rate gets implicitly overridden to 0 for clusters this size or smaller. Notice: if nodes live in multiple zones, this threshold gets treated as the zone node size threshold for each zone, to determine node eviction rate independently. 😭
  • --leader-elect=true — starts a leader election client and gains leadership before running the main loop. turn this on when you're running replicated components for high availability. main character energy, but like, democratically 👑
  • --leader-elect-lease-duration=15s — how long non-leader candidates wait after spotting a leadership renewal before they try to snatch the leadership of a led-but-unrenewed leader slot. basically: the max duration a leader can be stopped before another candidate replaces her. only applies if leader election is enabled. 💀
  • --leader-elect-renew-deadline=10s — the interval between the acting master's attempts to renew a leadership slot before it stops leading. must be less than the lease duration. only applies if leader election is enabled. ⏰
  • --leader-elect-resource-lock="leases" — the type of resource object used for locking during leader election. supported options are 'leases'. that's it, that's the list 🎀
  • --leader-elect-resource-name="kube-controller-manager" — the name of the resource object used for locking during leader election. 💅
  • --leader-elect-resource-namespace="kube-system" — the namespace of that resource object used for locking during leader election. 🏠
  • --leader-elect-retry-period=2s — how long clients should wait between attempting acquisition and renewal of a leadership. only applies if leader election is enabled. 🫶
  • --leader-migration-config="" — path to the config file for controller leader migration, or empty to use the value reflecting the default configuration of the controller manager. the config file should be of type LeaderMigrationConfiguration, group controllermanager.config.k8s.io, version v1alpha1. 📄
  • --legacy-service-account-token-clean-up-period=8760h0m0s — the amount of time since the last usage of a legacy service account token before it can be deleted. she's had her moment, now she's gone 🥀
  • --log-flush-frequency=5s — the maximum number of seconds between log flushes. 🚽 (sorry not sorry)
  • --log-text-info-buffer-size=0 — [Alpha] in text format with split output streams, info messages can be buffered for a while to boost performance. the default value of zero bytes disables buffering. you can write the size as number of bytes (512), multiples of 1000 (1K), multiples of 1024 (2Ki), or powers of those (3M, 4G, 5Mi, 6Gi). enable the LoggingAlphaOptions feature gate to use this. 🧚
  • --log-text-split-stream=false — [Alpha] in text format, write error messages to stderr and info messages to stdout. the default is a single stream to stdout. enable the LoggingAlphaOptions feature gate to use this. 💜
  • --logging-format="text" — sets the log format. permitted formats: "text". that's the whole menu, don't ask for more 🙈
  • --master="" — the address of the Kubernetes API server (it overrides any value sitting in kubeconfig). 👑
  • --max-endpoints-per-slice=100 — the max number of endpoints added to an EndpointSlice. more endpoints per slice = fewer endpoint slices, but bigger resources. defaults to 100. 🍓
  • --min-compatibility-version=[] — the min version of control plane components the server should be compatible with. must be less or equal to the emulated-version. version format can only be major.minor, for example: '--min-compatibility-version=wardle=1.2,kube=1.31'. options are: kube=1.34..1.37(default:1.36) — and if the component isn't specified, it defaults to "kube". 🎯
  • --min-resync-period=12h0m0s — the resync period in reflectors will be random between MinResyncPeriod and 2*MinResyncPeriod. surprise! 🌈
  • --mirroring-concurrent-service-endpoint-syncs=5 — the number of service endpoint syncing operations done concurrently by the endpointslice-mirroring-controller. bigger number = faster endpoint slice updating, but more CPU (and network) load. defaults to 5. 🔥
  • --mirroring-endpointslice-updates-batch-period=0s — the length of the EndpointSlice updates batching period for the endpointslice-mirroring-controller. EndpointSlice change processing gets delayed by this duration to join them with potential upcoming updates and cut down the overall number of EndpointSlice updates. bigger number = higher endpoint programming latency, but a lower number of endpoints revisions generated. 🦋
  • --mirroring-max-endpoints-per-subset=1000 — the max number of endpoints the endpointslice-mirroring-controller adds to an EndpointSlice. more endpoints per slice = fewer endpoint slices, but bigger resources. defaults to 100. (yes bestie, the text says 100 even though the flag says 1000, we report, we don't editorialize 💅)
  • --namespace-sync-period=5m0s — the period for syncing namespace life-cycle updates. 🌸
  • --node-cidr-mask-size=0 — mask size for the node cidr in the cluster. default is 24 for IPv4 and 64 for IPv6. 🎀
  • --node-cidr-mask-size-ipv4=0 — mask size for the IPv4 node cidr in a dual-stack cluster. default is 24. 🫶
  • --node-cidr-mask-size-ipv6=0 — mask size for the IPv6 node cidr in a dual-stack cluster. default is 64. 💜
  • --node-eviction-rate=0.1 — number of nodes per second on which pods get deleted when a node fails and a zone is healthy (see --unhealthy-zone-threshold for the definition of healthy/unhealthy). "zone" means the entire cluster in non-multizone clusters. 💀
  • --node-monitor-grace-period=50s — how long we let a running Node be unresponsive before we mark it unhealthy. must be N times more than kubelet's nodeStatusUpdateFrequency, where N is the number of retries allowed for kubelet to post node status. this value should also be greater than the sum of HTTP2_PING_TIMEOUT_SECONDS and HTTP2_READ_IDLE_TIMEOUT_SECONDS. 😭
  • --node-monitor-period=5s — the period for syncing NodeStatus in NodeLifecycleController. ⏱️
  • --node-startup-grace-period=1m0s — how long we let a starting Node be unresponsive before marking it unhealthy. 🌱
  • --node-sync-period=0s — > ⚠️ This flag is deprecated and will be removed in future releases. see node-monitor-period for Node health checking, or route-reconciliation-period for the cloud provider's route configuration settings. she's leaving, don't get attached 💀
  • --permit-address-sharing=false — if true, SO_REUSEADDR gets used when binding the port. this lets you bind wildcard IPs like 0.0.0.0 and specific IPs in parallel, and it avoids waiting for the kernel to release sockets stuck in TIME_WAIT state. [default=false] 🎀
  • --permit-port-sharing=false — if true, SO_REUSEPORT gets used when binding the port, which lets more than one instance bind on the same address and port. [default=false] ✨
  • --profiling=true — enable profiling via the web interface at host:port/debug/pprof/. 📊
  • --pv-recycler-increment-timeout-nfs=30 — the increment of time added per Gi to ActiveDeadlineSeconds for an NFS scrubber pod. ⏳
  • --pv-recycler-minimum-timeout-hostpath=60 — the minimum ActiveDeadlineSeconds to use for a HostPath Recycler pod. This is for development and testing only and will not work in a multi-node cluster. 🚨
  • --pv-recycler-minimum-timeout-nfs=300 — the minimum ActiveDeadlineSeconds to use for an NFS Recycler pod. 🌸
  • --pv-recycler-pod-template-filepath-hostpath="" — the file path to a pod definition used as a template for HostPath persistent volume recycling. This is for development and testing only and will not work in a multi-node cluster. 🙈
  • --pv-recycler-pod-template-filepath-nfs="" — the file path to a pod definition used as a template for NFS persistent volume recycling. 🍓
  • --pv-recycler-timeout-increment-hostpath=30 — the increment of time added per Gi to ActiveDeadlineSeconds for a HostPath scrubber pod. This is for development and testing only and will not work in a multi-node cluster. 💀
  • --pvclaimbinder-sync-period=15s — the period for syncing persistent volumes and persistent volume claims. 🔁
  • --requestheader-allowed-names=[] — list of client certificate common names allowed to provide usernames in the headers specified by --requestheader-username-headers. if it's empty, any client certificate validated by the authorities in --requestheader-client-ca-file is allowed. 👑
  • --requestheader-client-ca-file="" — the root certificate bundle you use to verify client certs on incoming requests before you trust the usernames hiding in the headers that --requestheader-username-headers points at. WARNING: bestie, do not just assume authorization already happened for incoming requests. Generally don't depend on that. 🙈
  • --requestheader-extra-headers-prefix=[x-remote-extra-] — list of request header prefixes to go snooping through. X-Remote-Extra- is the suggested one. 💅
  • --requestheader-group-headers=[x-remote-group] — list of request headers to inspect for groups. X-Remote-Group is suggested. 👑
  • --requestheader-uid-headers=[] — list of request headers to inspect for UIDs. X-Remote-Uid is suggested. Heads up: this one requires the RemoteRequestHeaderUID feature to be enabled. 🔥
  • --requestheader-username-headers=[x-remote-user] — list of request headers to inspect for usernames. X-Remote-User is the common one, fr fr. 🫶
  • --resource-quota-sync-period=5m0s — the period for syncing quota usage status in the system. ⏰
  • --root-ca-file="" — if you set it, this root certificate authority gets included in the service account's token secret. It has to be a valid PEM-encoded CA bundle, no cap. 🎀
  • --route-reconciliation-period=10s — the period for reconciling routes that the cloud provider created for Nodes. 🌈
  • --secondary-node-eviction-rate=0.01 — number of nodes per second on which pods get deleted when a node fails and a zone is unhealthy (peep --unhealthy-zone-threshold for what counts as healthy/unhealthy). A zone is the entire cluster in non-multizone clusters. Lowkey important: this value is implicitly overridden to 0 if the cluster size is smaller than --large-cluster-size-threshold. 💀
  • --secure-port=10257 — the port where HTTPS gets served with authentication and authorization. If it's 0, don't serve HTTPS at all. periodt. 🔒
  • --service-account-private-key-file="" — enables legacy secret-based tokens when set. It's the filename holding a PEM-encoded private RSA or ECDSA key used to sign service account tokens. 🔑
  • --service-cluster-ip-range="" — CIDR Range for Services in cluster. Only used when --allocate-node-cidrs=true; if that's false, this option gets ignored. 💅
  • --show-hidden-metrics-for-version="" — the previous version you want to show hidden metrics for. Only the previous minor version is meaningful, other values won't be allowed. The format is ., e.g.: '1.16'. The whole point of this format is making sure you get the chance to notice if the next release hides additional metrics, instead of being jump-scared when they're permanently removed in the release after that. 😭
  • --terminated-pod-gc-threshold=12500 — number of terminated pods that can exist before the terminated pod garbage collector starts deleting terminated pods. If it's <= 0, the terminated pod garbage collector is disabled. bye bestie 🧹
  • --tls-cert-file="" — file containing the default x509 Certificate for HTTPS. (CA cert, if any, concatenated after server cert.) If HTTPS serving is enabled and --tls-cert-file and --tls-private-key-file are not provided, a self-signed certificate and key are generated for the public address and saved to the directory specified by --cert-dir. ✨
  • --tls-cipher-suites=[] — comma-separated list of cipher suites for the server. If omitted, the default Go cipher suites will be used.

Preferred values: TLS_AES_128_GCM_SHA256, TLS_AES_256_GCM_SHA384, TLS_CHACHA20_POLY1305_SHA256, TLS_ECDHE_ECDSA_WITH_AES_128_CBC_SHA, TLS_ECDHE_ECDSA_WITH_AES_128_GCM_SHA256, TLS_ECDHE_ECDSA_WITH_AES_256_CBC_SHA, TLS_ECDHE_ECDSA_WITH_AES_256_GCM_SHA384, TLS_ECDHE_ECDSA_WITH_CHACHA20_POLY1305, TLS_ECDHE_ECDSA_WITH_CHACHA20_POLY1305_SHA256, TLS_ECDHE_RSA_WITH_AES_128_CBC_SHA, TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256, TLS_ECDHE_RSA_WITH_AES_256_CBC_SHA, TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384, TLS_ECDHE_RSA_WITH_CHACHA20_POLY1305, TLS_ECDHE_RSA_WITH_CHACHA20_POLY1305_SHA256.

Insecure values: TLS_ECDHE_ECDSA_WITH_AES_128_CBC_SHA256, TLS_ECDHE_ECDSA_WITH_RC4_128_SHA, TLS_ECDHE_RSA_WITH_3DES_EDE_CBC_SHA, TLS_ECDHE_RSA_WITH_AES_128_CBC_SHA256, TLS_ECDHE_RSA_WITH_RC4_128_SHA, TLS_RSA_WITH_3DES_EDE_CBC_SHA, TLS_RSA_WITH_AES_128_CBC_SHA, TLS_RSA_WITH_AES_128_CBC_SHA256, TLS_RSA_WITH_AES_128_GCM_SHA256, TLS_RSA_WITH_AES_256_CBC_SHA, TLS_RSA_WITH_AES_256_GCM_SHA384, TLS_RSA_WITH_RC4_128_SHA.
  • --tls-curve-preferences=[] — comma-separated list of numeric Go crypto/tls CurveID values, as the allowed key exchange mechanisms for the server. The supported values depend on the Go version you're using. See https://pkg.go.dev/crypto/tls#CurveID for values supported for each Go version. The order of the list is ignored — key exchange mechanisms are chosen by Go from this list using an internal preference order. If omitted, the default Go curves will be used. 🌀
  • --tls-min-version="" — minimum TLS version supported. Possible values: VersionTLS10, VersionTLS11, VersionTLS12, VersionTLS13. 🛡️
  • --tls-private-key-file="" — file containing the default x509 private key matching --tls-cert-file. 🔐
  • --tls-sni-cert-key=[] — a pair of x509 certificate and private key file paths, optionally suffixed with a list of domain patterns which are fully qualified domain names, possibly with prefixed wildcard segments. The domain patterns also allow IP addresses, but IPs should only be used if the apiserver has visibility to the IP address requested by a client. If no domain patterns are provided, the names of the certificate are extracted. Non-wildcard matches trump over wildcard matches, explicit domain patterns trump over extracted names. For multiple key/certificate pairs, use the --tls-sni-cert-key multiple times. Examples: "example.crt,example.key" or "foo.crt,foo.key:*.foo.com,foo.com". 💖
  • --unhealthy-zone-threshold=0.55 — fraction of Nodes in a zone which needs to be not Ready (minimum 3) for the zone to be treated as unhealthy. 📉
  • --use-service-account-credentials=false — if true, use individual service account credentials for each controller. 🙋
  • -v, --v=0 — number for the log level verbosity. 🗣️
  • --version=false — --version, --version=raw prints version information and quits; --version=vX.Y.Z... sets the reported version. 👀
  • --vmodule= — comma-separated list of pattern=N settings for file-filtered logging (only works for text log format). 🍓

📜 HISTORY 📜

January 2015, originally compiled by Eric Paris (eparis at redhat dot com) based on the kubernetes source material, but hopefully they have been automatically generated since! 💀✨

<center>okay that's a wrap bestie 🫶 this whole thing lives under the Manuals manual, User section, titled KUBERNETES(1)(kubernetes) — and the era it was born in is that January 2015 energy. slay, you read the whole page. 👑</center>

📦 where this page comes from