Kubernetes v1.30.0 is LIVE!

605 views
Skip to first unread message

Mark Rossetti

unread,
Apr 17, 2024, 5:04:17 PM4/17/24
to kubernetes-announce, d...@kubernetes.io
Kubernetes Community,
Kubernetes v1.30.0 has been built and pushed using Golang version 1.22.2.
The release notes have been updated in CHANGELOG-1.30.md, with a pointer to them on GitHub:
v1.30.0
Downloads for v1.30.0
Source Code
filename sha512 hash
kubernetes.tar.gz 1abe2ea09c08787dfa85450d026499a7a716bb771ff5af0aff1ac651b0059c0aac7a4e8558ef1eddffd4ccf1871daa9086d0be2e3ed8bed47f8c9930e52578ef
kubernetes-src.tar.gz 69a4ae4e4d227f784e241078b5667b7b7eb3cf787d237475cb311e6d417ceb2057624db6d2a5dded81e97db6250fb2e642388b564107fa86ac3c4b5e75980843
Client Binaries
filename sha512 hash
kubernetes-client-darwin-amd64.tar.gz 3a98e32c6e912b0d7e8af31a1e49be20983dd6ddfbc54024cf58eab551cba5398fb04d69138c39b3bc672c962b5a038f4404ac603a6e338218f00511e0bbf41e
kubernetes-client-darwin-arm64.tar.gz 98fa1d19bb9250ff7c9536090a3d939fc2c7cc7253efd91e1fba4613acf44557bf5cdae9d7fd4fc5c209eeacc8bc2b584d281b92fc6e3d1ccea7cf43853d64cc
kubernetes-client-linux-386.tar.gz 608388b0233e08b7b68fb62721275c271356db4c95b2686821a0b6e2c002e26b5d453ea470adddc9b3c17c623d29cec0a2ff66adb75b1e74f9a2f77101e8679e
kubernetes-client-linux-amd64.tar.gz 7983b34128518610a63b500bc9a612b5c334180555af534940a3ec839d4e0b476c9d7742cb6cc57c9ba15de609561fafef5854b3708e48c1e9f822480f3e60b6
kubernetes-client-linux-arm.tar.gz 36955e5c1637d0b1f679f2ae7e8bf780b9e70fc8a7e64c5194aa8a44035982b97bf8ca631dbc1af4c46d91ff686fdd24bd49b5076fc1617a18bc496b0aee6367
kubernetes-client-linux-arm64.tar.gz edc3cc941996235d578c59f80e88cdadbc5259a9b3505073deac8632ef3cc08f53bc70959304f4aed5ccf2f99cab02f097730cdab9e8e691f5ffe8217be497fe
kubernetes-client-linux-ppc64le.tar.gz 45a0119f4e5953eebe3cdf2e714d2cce376a800ca8194030880c0b3b5a5f9e462aa2e73bab97d16c9094dac290e3e55ba3df006fc7cd50f1ee1acdedc4dba7a1
kubernetes-client-linux-s390x.tar.gz 6a15fe0ad5ca4fbff89ea38d0390cf9c822971a50bb18bc37c23dfe3655a7072ed60f87408a47733c7ecb64cebe6201b43140ce6978f2a3c7753ff98bb0e4156
kubernetes-client-windows-386.tar.gz aea2df663dc841a513ceb060fb7368f492ec864c75c87569594c1d1a09cae2caa6e67846517fd246214395716ec7809e5b9bd6a92adae7afed1a4b087e9a3831
kubernetes-client-windows-amd64.tar.gz 5483baf908e22de9347fa965ffb1487afec12dc3c3b8642895ac91811154e2cee5de4a3218369f55929badff0466ae9437b59c0922a3699c2bec44a7c4841c47
kubernetes-client-windows-arm64.tar.gz fe922ecb57a0313ff461438db130416ad3dc4c849f243cf4f29010288351afe3a9ec786b1c820784bd51ca8f7a74f7436797565339d0d0583d93b0a91d5d50aa
Server Binaries
filename sha512 hash
kubernetes-server-linux-amd64.tar.gz 320c934728b8720b08da5a2d4c95cea111e4bc2efa508a8e48393f45037b0cb3c5331e3ec1f744f83dd57644c6fdd684bdcad9c2fa01108ed2cca6b1046cb821
kubernetes-server-linux-arm64.tar.gz ddf0c4a47f1cd1ad44e9773b2cf8255c6fb11ef84fa42e73dd9479b9a8b8c94a8dd5ba6581ee3aa152354d9a2fe1ad219ba5289c64aa6b9ddd7ddbca000525d4
kubernetes-server-linux-ppc64le.tar.gz 5a1e839e38f5b4329dd8d0bac38c0dade336952821c8a4dc550221290481dc8d65f8b2995fc4330c7d0ea6a68fcbe81ae4085edd80b4194761bc86dac3ed380b
kubernetes-server-linux-s390x.tar.gz 30fa3358a0b0e32ed7dfcaaba7d4403cdcf3a305124e7419c898db87c9ba1dc1ae5fc8f71c6caf744d35ea85d3526069596204c18f23449a759703b2ae797bbf
Node Binaries
filename sha512 hash
kubernetes-node-linux-amd64.tar.gz e3e8f443a5397029a26a30852f47a9406dc17622638a7b03d37b384283880503af85d134d2a5dfdcfd7a0a5cb11eef8312b37dec77c90062e29d7e0ea1da6026
kubernetes-node-linux-arm64.tar.gz d1299f54a2ec4343eb23be36071e3560f339d535b4c2eeb60c09cf40b3357f4371ca198aa4af0c41bde73bef1221a839c846a9f9b497f47f2fdf553238a7dab9
kubernetes-node-linux-ppc64le.tar.gz cd78d5811ac15aefb3392c9bc10ad97c69d9d87add1237ebfc8c72d12be6137d8a4933967a5294932ceff58ffadaddf541cff8d0ae3b663359ef5f6a9ad98b84
kubernetes-node-linux-s390x.tar.gz 26cec1a3f76289394966f98eab1e27a6dc122ac58f1d7c1ea94eb6b96e74377c449aa282d788655a1f752ff31add47be5389543a75e52cb4de947b7c613fe1e6
kubernetes-node-windows-amd64.tar.gz b6eed059c8e9c816ca5e26d9fd8154dd904eefd5024e71af59757b801bfd9b232b63d6c74c824b59891a2689b209d1dd7622f86ec3fd8530ebeffab36e3b504c
Container Images
All container images are available as manifest lists and support the described architectures. It is also possible to pull a specific architecture directly by adding the "-$ARCH" suffix to the container image name.
name architectures
registry.k8s.io/conformance:v1.30.0 amd64, arm64, ppc64le, s390x
registry.k8s.io/kube-apiserver:v1.30.0 amd64, arm64, ppc64le, s390x
registry.k8s.io/kube-controller-manager:v1.30.0 amd64, arm64, ppc64le, s390x
registry.k8s.io/kube-proxy:v1.30.0 amd64, arm64, ppc64le, s390x
registry.k8s.io/kube-scheduler:v1.30.0 amd64, arm64, ppc64le, s390x
registry.k8s.io/kubectl:v1.30.0 amd64, arm64, ppc64le, s390x
Changelog since v1.29.0
Changes by Kind
Deprecation
  • kubectl: Removed the deprecated flag prune-whitelist for apply. Please use the flag prune-allowlist instead. (#120246, @pacoxu)
  • The deprecated SecurityContextDeny admission plugin, which has been deprecated since v1.27, has been removed. It is recommended to use the Pod Security Admission plugin instead, which has been available since v1.25. Refer to the Kubernetes documentation for more information. (#122612, @mtardy)
API Change
  • Added (alpha) support for the managedBy field on Jobs. Jobs with a custom value of this field - any value other than kubernetes.io/job-controller - were skipped by the job controller, and their reconciliation was delegated to an external controller, indicated by the value of the field. Jobs that didn't have this field at all, or where the field value was the reserved string kubernetes.io/job-controller, were reconciled by the built-in job controller. (#123273, @mimowo)
  • Added alpha-level support for the SuccessPolicy in Jobs. (#123412, @tenzen-y)
  • Added the CEL library for IP Addresses and CIDRs. This was made available for use starting from version 1.31. (#121912, @JoelSpeed)
  • Allowed container runtimes to fix an image garbage collection bug by adding an image_id field to the CRI Container message. (#123508, @saschagrunert)
  • Dynamic Resource Allocation: DRA drivers can now use "structured parameters" to let the scheduler handle claim allocation. (#123516, @pohly)
  • Fixed accidental enablement of the new alpha optionalOldSelf API field in CustomResourceDefinition validation rules, which should only have been allowed to be set when the CRDValidationRatcheting feature gate is enabled. (#122329, @jpbetz)
  • Implemented the prescore extension point for the volumeBinding plugin. It now returns skip if it doesn't do anything in Score. (#115768, @AxeZhan)
  • Kubelet would fail if NodeSwap was used with LimitedSwap and cgroupv1 node. (#123738, @kannon92)
  • Promoted AdmissionWebhookMatchConditions to GA. The feature is now stable, and the feature gate is now locked to default. (#123560, @ivelichkovich)
  • Structured Authentication Configuration now supports DiscoveryURL. If specified, discoveryURL overrides the URL used to fetch discovery information. This is for scenarios where the well-known and jwks endpoints are hosted at a different location than the issuer (such as locally in the cluster). (#123527, @aramase)
  • The StorageVersionMigration API, previously available as a Custom Resource Definition (CRD), is now a built-in API in Kubernetes. (#123344, @nilekhc)
  • When configuring a JWT authenticator:
    If username.expression used 'claims.email', then 'claims.email_verified' must have been used in username.expression or extra[*].valueExpression or claimValidationRules[*].expression. An example claim validation rule expression that matches the validation automatically applied when username.claim is set to 'email' is 'claims.?email_verified.orValue(true)'. (#123737, @enj)
  • readOnly volumes now support recursive read-only mounts for kernel versions >= 5.12." (#123180, @AkihiroSuda)
  • cri-api: Implemented KEP-3857: Recursive Read-only (RRO) mounts. (#123272, @AkihiroSuda)
  • kube-apiserver: the AuthenticationConfiguration type accepted in --authentication-config files has been promoted to apiserver.config.k8s.io/v1beta1. (#123696, @aramase)
  • kubelet allowed specifying a custom root directory for pod logs (instead of the default /var/log/pods) using the podLogsDir key in kubelet configuration. (#112957, @mxpv)
  • resource.k8s.io/ResourceClaim (alpha API): The strategic merge patch strategy for the status.reservedFor array was changed so that a strategic-merge-patch can now add individual entries. This change may break clients using strategic merge patch to update status, which rely on the previous behavior (replacing the entire array). (#122276, @pohly)
  • Added a CBOR implementation of runtime.Serializer. Until CBOR graduates to Alpha, API servers will refuse to start if configured with CBOR support. (#122881, @benluddy)
  • Added a alpha feature, behind the RelaxedEnvironmentVariableValidation feature gate. When that gate is enabled, Kubernetes allows almost all printable ASCII characters to be used in the names of environment variables for containers in Pods. (#123385, @HirazawaUi)
  • Added a new (alpha) field, trafficDistribution, to the Service spec to express preferences for traffic distribution to endpoints. Enabled through the ServiceTrafficDistribution feature gate. (#123487, @gauravkghildiyal)
  • Added audienceMatchPolicy field to AuthenticationConfiguration and support for configuring multiple audiences. The "audienceMatchPolicy" can be empty (or unset) when a single audience is specified in the "audiences" field. The "audienceMatchPolicy" must be set to "MatchAny" when multiple audiences are specified in the "audiences" field. (#123165, @aramase)
  • Added consistent vanity import to files and provided tooling for verifying and updating them. (#120642, @jcchavezs)
  • Added the disable-force-detach CLI option for kube-controller-manager. By default, it's set to false. When enabled, it prevents force detaching volumes based on maximum unmount time and node status. If activated, the non-graceful node shutdown feature must be used to recover from node failure. Additionally, if a pod needs to be forcibly terminated at the risk of corruption, the appropriate VolumeAttachment object must be deleted. (#120344, @rohitssingh)
  • Added to MutableFeatureGate the ability to override the default setting of feature gates, to allow default-enabling a feature on a component-by-component basis instead of for all affected components simultaneously. (#122647, @benluddy)
  • Aggregated discovery supports both v2beta1 and v2 types and feature is promoted to GA. (#122882, @Jefftree)
  • Alpha support for field selectors on custom resources has been added. With the CustomResourceFieldSelectors feature gate enabled, the CustomResourceDefinition API now allows specifying selectableFields. Listing a field there enables filtering custom resources for that CustomResourceDefinition in list or watch requests. (#122717, @jpbetz)
  • AppArmor profiles can now be configured through fields on the PodSecurityContext and container SecurityContext. The beta AppArmor annotations are deprecated, and AppArmor status is no longer included in the node ready condition. (#123435, @tallclair)
  • Contextual logging is now in beta and enabled by default. Check out the KEP and official documentation for more details. (#122589, @pohly)
  • Enabled concurrent log rotation in kubelet. You can now configure the maximum number of concurrent rotations with the containerLogMaxWorkers setting, and adjust the monitoring interval with containerLogMonitorInterval. (#114301, @harshanarayana)
  • Graduated pod scheduling gates to general availability. The PodSchedulingReadiness feature gate no longer has any effect, and the .spec.schedulingGates field is always available within the Pod and PodTemplate APIs. (#123575, @Huang-Wei)
  • Graduated support for minDomains in pod topology spread constraints, to general availability. The MinDomainsInPodTopologySpread feature gate no longer has any effect, and the field is always available within the Pod and PodTemplate APIs. (#123481, @sanposhiho)
  • In kubelet configuration, the .memorySwap.swapBehavior field now accepts a new value NoSwap, which becomes the default if unspecified. The previously accepted UnlimitedSwap value has been dropped. (#122745, @kannon92)
  • Kube-apiserver: the AuthorizationConfiguration type accepted in --authorization-config files has been promoted to apiserver.config.k8s.io/v1beta1. (#123640, @liggitt)
  • OIDC authentication will now fail if the username asserted based on a CEL expression config is the empty string. Previously the request would be authenticated with the username set to the empty string. (#123568, @enj)
  • Removed note that hostAliases are not supported on hostNetwork Pods from the PodSpec API. The feature has been supported since v1.8. (#122422, @neolit123)
  • Structured Authentication Configuration now supports configuring multiple JWT authenticators. The maximum allowed JWT authenticators in the authentication configuration is 64. (#123431, @aramase)
  • Text logging in Kubernetes components now uses textlogger. The same split streams of info and error log entries with buffering of info entries is now also supported for text output (off by default, alpha feature). Previously, this was only supported for JSON. Performance is better also without split streams. (#114672, @pohly)
  • The API server now detects and fails on startup if there are conflicting issuers between JWT authenticators and service account configurations. Previously, such configurations would run but could be inconsistently effective depending on the credential. (#123561, @enj)
  • The JWT authenticator configuration set via the --authentication-config flag is now dynamically reloaded as the file changes on disk. (#123525, @enj)
  • The StructuredAuthenticationConfiguration feature is now beta and enabled. (#123719, @enj)
  • The kube_codegen tool now ignores the vendor folder during code generation. (#122729, @jparrill)
  • The kubernetes repo now uses Go workspaces. This should not impact end users at all, but does have impact for developers of downstream projects. Switching to workspaces caused some breaking changes in the flags to the various k8s.io/code-generator tools. Downstream consumers should look at staging/src/k8s.io/code-generator/kube_codegen.sh to see the changes. (#123529, @thockin)
  • Updated an audit annotation key used by the …/serviceaccounts/<name>/token resource handler. The annotation used to persist the issued credential identifier is now authentication.kubernetes.io/issued-credential-id. (#123098, @munnerz) [SIG Auth]
  • Users are now allowed to mutate FSGroupPolicy and PodInfoOnMount in CSIDriver.Spec. (#116209, @haoruan)
  • ValidatingAdmissionPolicy was promoted to GA and will be enabled by default. (#123405, @cici37)
  • When scheduling a mix of pods using ResourceClaims and others that don't, scheduling a pod with ResourceClaims has a lower impact on scheduling latency. (#121876, @pohly)
  • When working with client-go events, it's now recommended to use NewEventBroadcasterAdapterWithContext instead of NewEventBroadcasterAdapter if contextual logging support is needed. (#122142, @pohly)
Feature
  • Added Timezone column in the output of the 'kubectl get cronjob' command. (#122231, @ardaguclu)
  • Added WatchListClient feature gate to client-go. When enabled, it allows the client to receive a stream of individual items instead of chunking from the server. (#122571, @p0lyn0mial)
  • Added the apiserver_watch_cache_read_wait metric to measure the watch cache impact on request latency. (#123190, @padlar)
  • Allowed scheduling framework plugins that implement io.Closer to be gracefully closed. (#122498, @Gekko0114)
  • Bumped cAdvisor to v0.49.0. (#123599, @bobbypage)
  • Changed --nodeport-addresses behavior to default to "primary node IP(s) only" rather than "all node IPs". (#122724, @nayihz)
  • In the Pod API, setting the alpha procMount field to Unmasked in a container now required setting spec.hostUsers=false as well. (#123520, @haircommander)
  • Informers now supports adding Indexers after the informer starts. (#117046, @howardjohn)
  • Printed more information when kubectl describe a VolumeAttributesClass. (#122640, @carlory)
  • Promoted the CRDValidationRatcheting feature gate to beta and made it enabled by default. (#121461, @alexzielenski)
  • Scheduler now skips the NodeAffinity Score plugin when it has nothing to do with a Pod. You might have noticed an increase in the metric plugin_execution_duration_seconds for extension_point=score and plugin=NodeAffinity, because the plugin only runs when it's relevant. (#117024, @sanposhiho)
  • Some interfaces' signatures in the scheduler were updated: - PluginsRunner: used NodeInfo in RunPreScorePlugins and RunScorePlugins. - PreScorePlugin: used NodeInfo in PreScore. - Extender: used NodeInfo in Filter and Prioritize. (#121954, @AxeZhan)
  • The watch cache now waits until it is at least as fresh as the given requestedWatchRV if sendInitialEvents was requested. (#122830, @p0lyn0mial)
  • Updated ImageGCMaxAge behavior in the kubelet to wait the MaxAge duration after the kubelet has restarted before garbage collecting. (#123343, @haircommander)
  • Updated distroless-iptables to v0.5.0, debian-base to bookworm-v1.0.1, and setcap to bookworm-v1.0.1. (#123170, @cpanato)
  • NewVolumeManagerReconstruction feature is now GA. (#123442, @jsafrane)
  • kubectl describe: Added Suspend to job and Node-Selectors and Tolerations to pod template output. (#122618, @ivanvc)
  • kubectl get job now displays the status for the listed jobs. (#123226, @ivanvc)
  • etcd: Built image for v3.5.11. (#122233, @mzaian)
  • kube-apiserver now reloads the --authorization-config file when it changes. Reloads increment the apiserver_authorization_config_controller_automatic_reload_last_timestamp_seconds timestamp metric, with status="success" for successful reloads and status="failed" for failed reloads. Failed reloads keep using the previously loaded authorization configuration. (#121946, @liggitt)
  • kube-apiserver now reported the following metrics for authorization webhook match conditions: - apiserver_authorization_match_condition_evaluation_errors_total counter metric labeled by authorizer type and name - apiserver_authorization_match_condition_exclusions_total counter metric labeled by authorizer type and name - apiserver_authorization_match_condition_evaluation_seconds histogram metric labeled by authorizer type and name. (#123611, @ritazh)
  • kube-apiserver: JWT authenticator now reports the following metrics: - apiserver_authentication_config_controller_automatic_reloads_total - apiserver_authentication_config_controller_automatic_reload_last_timestamp_seconds (#123793, @aramase)
  • kube-apiserver: The StructuredAuthorizationConfiguration feature gate has been promoted to beta and now allows the use of the --authorization-configuration flag. (#123641, @liggitt)
  • kube-scheduler implemented scheduling hints for the NodeUnschedulable plugin. The scheduling hints allowed the scheduler to only retry scheduling a Pod that had been previously rejected by the NodeSchedulable plugin if a new Node or a Node update had set .spec.unschedulable to false. (#122334, @carlory)
  • kubeadm: Added better handling of errors during unmount when calling kubeadm reset. When failing to unmount directories under /var/run/kubelet, kubeadm will now throw an error instead of showing a warning and continuing to clean up said directory. In such situations, it is better for you to inspect the problem and resolve it manually. Then, you can call kubeadm reset again to complete the cleanup. (#122530, @neolit123)
  • kubeadm: Added support for machine-readable output with -o yaml and -o json to the command kubeadm certs check-expiration. This change is introduced in a new API: kind: CertificateExpirationInfo apiVersion: output.kubeadm.k8s.io/v1alpha3 The existing non-structured formatting is preserved. The output API version v1alpha2 is now deprecated and will be removed in a future release. Please migrate to using v1alpha3. (#123372, @carlory)
  • kubeadm: added the WaitForAllControlPlaneComponents feature gate. It could be used to tell kubeadm to wait for all control plane components to be ready when running "kubeadm init" or "kubeadm join --control-plane". Previously, kubeadm only waited for the kube-apiserver. The "kubeadm join" workflow now includes a new experimental phase called "wait-control-plane". This phase was marked as non-experimental when WaitForAllControlPlaneComponents became GA. Accordingly, a "kubeadm init" phase "wait-control-plane" was also available once WaitForAllControlPlaneComponents became GA. These phases could be skipped if the user preferred not to wait for the control plane components. (#123341, @neolit123)
  • kubectl port-forward over websockets (tunneling SPDY) can now be enabled using an Alpha feature flag environment variable: KUBECTL_PORT_FORWARD_WEBSOCKETS=true. The API Server being communicated to must also have an Alpha feature flag enabled: PortForwardWebsockets. (#123413, @seans3)
  • A new flag called custom has been introduced in kubectl debug, allowing users to customize pre-defined profiles. (#120346, @ardaguclu)
  • A new kubelet metric image_pull_duration_seconds was added. The metric tracks the duration (in seconds) it takes for an image to be pulled, including the time spent in the waiting queue of image puller. The metric is broken down by bucketed image size. (#121719, @ruiwen-zhao)
  • A new metric lifecycle_handler_sleep_terminated_total is added to record how many times LifecycleHandler sleep got unexpectedly terminated. (#122456, @AxeZhan)
  • Added client-go support for upgrading subresource fields from client-side to server-side management. (#123484, @erikgb)
  • Added exec-interactive-mode and exec-provide-cluster-info flags in kubectl config set-credentials command. (#122023, @ardaguclu)
  • Added process_start_time_seconds to /metrics/slis endpoint of all components. (#122750, @richabanker)
  • Added a "reason" field to the "image_garbage_collected_total" metric, allowing administrators to differentiate between images that were collected for reasons "age" vs "space". (#123345, @haircommander)
  • Added a new metric apiserver_encryption_config_controller_automatic_reloads_total to measure the total number of API server encryption configuration reload successes and failures. This metric now contains the status label with a value that is either success or failure. Deprecated the metrics apiserver_encryption_config_controller_automatic_reload_success_total and apiserver_encryption_config_controller_automatic_reload_failure_total. Please use apiserver_encryption_config_controller_automatic_reloads_total instead. (#123179, @aramase)
  • Added feature gate MutatingAdmissionPolicy for enabling mutation policy in admission chain. (#123425, @cici37)
  • Added kubelet metrics to track the memory manager allocation and pinning. (#121778, @Tal-or)
  • Added the access_mode label to volume_manager_selinux_* metrics. (#123667, @jsafrane)
  • Enhanced cloud provider integrations to support optional, per-Node custom labels that can be supplied and applied to Nodes by the node controller. These extra labels will only be applied where the cloud provider integration supports this feature. (#123223, @mmerkes)
  • Graduated "Forensic Container Checkpointing" (KEP #2008) from Alpha to Beta. (#123215, @adrianreber)
  • Graduated HorizontalPodAutoscaler support for per-container metrics to stable. (#123482, @sanposhiho)
  • Graduated support for passing dual-stack kubelet --node-ip values when using a cloud provider. The feature is now GA, and the CloudDualStackNodeIPs feature gate is always enabled. (#123134, @danwinship)
  • In the PriorityLevelConfiguration object, the nominalConcurrencyShares field now accepts a zero value in both the flowcontrol.apiserver.k8s.io/v1 and flowcontrol.apiserver.k8s.io/v1beta3 APIs. (#123001, @tkashem)
  • Introduced a feature gate mechanism to client-go. Depending on the actual implementation, users can control features via environmental variables or command line options. (#122555, @p0lyn0mial)
  • Introduced a new alpha feature gate, SELinuxMount, which can now be enabled to accelerate SELinux relabeling. (#123157, @jsafrane)
  • Kube-apiserver now reports latency metric for JWT authenticator authenticate token decisions in the apiserver_authentication_jwt_authenticator_latency_seconds metric, labeled by jwtIssuer hash and result. (#123225, @aramase)
  • Kube-apiserver now reports metrics for authorization decisions in the apiserver_authorization_decisions_total metric, labeled by authorizer type, name, and decision. (#123333, @liggitt)
  • Kube-apiserver: Authorization webhooks now report the following metrics:
    • apiserver_authorization_webhook_evaluations_total
    • apiserver_authorization_webhook_duration_seconds
    • apiserver_authorization_webhook_evaluations_fail_open_total (#123639, @liggitt)
  • Kube-controller-manager: increased the global level for broadcaster's logging to 3 so that users can ignore event messages by lowering the logging level. It reduces information noise. (#122293, @mengjiao-liu)
  • Kube-scheduler implemented scheduling hints for the NodeAffinity plugin. The scheduling hints allowed the scheduler to only retry scheduling a Pod that had been previously rejected by the NodeAffinity plugin if a new Node or a Node update matched the Pod's node affinity. (#122309, @carlory)
  • Kube-scheduler implemented scheduling hints for the NodeResourceFit plugin. The scheduling hints allowed the scheduler to only retry scheduling a Pod that had been previously rejected by the NodeResourceFit plugin if a new Node or a Node update matched the Pod's resource requirements or if an old pod update or delete matched the Pod's resource requirements. (#119177, @carlory)
  • Kube-scheduler now retries scheduling Pods rejected by the PodTopologySpread plugin when related Pods are created, deleted, updated, or when a node matches the specified topologyKey. (#122195, @nayihz)
  • Kubeadm now prints all the kubelets and nodes that need to be upgraded when running "upgrade plan". (#123578, @carlory)
  • Kubectl debug now includes the sysadmin profile. (#119200, @eiffel-fl)
  • Kubelet now supports configuring the IDs used to create user namespaces. (#123593, @giuseppe)
  • Kubernetes is now built with Go 1.21.6. (#122705, @cpanato)
  • Kubernetes is now built with Go 1.22.1. (#123750, @cpanato)
  • Kubernetes is now built with Go 1.22. (#123217, @cpanato)
  • Kubernetes is now built with go 1.22rc2. (#122889, @cpanato)
  • LoadBalancerIPMode feature is now marked as Beta. (#123418, @rikatz)
  • Node information is now embedded into Pod-bound service account tokens as additional metadata. The 'JTI' field is set in issued service account tokens, and this information is embedded as authentication.kubernetes.io/credential-id in the user's ExtraInfo. (#123135, @munnerz)
  • Node podresources API now includes init containers with containerRestartPolicy of Always when SidecarContainers feature is enabled. (#120718, @gjkim42)
  • Promoted KubeProxyDrainingTerminatingNodes to Beta. (#122914, @alexanderConstantinescu)
  • Promoted feature gate StableLoadBalancerNodeSet to GA . (#122961, @alexanderConstantinescu)
  • Promoted the ImageMaximumGCAge feature to beta. (#123424, @haircommander)
  • Promoted the status.hostIPs field for Pods to general availability. The PodHostIPs feature gate no longer has any effect, and the status.hostIPs field is always available within the Pod API. (#122870, @wzshiming)
  • RemoteCommand feature gates for kubectl exec, cp, and attach over WebSockets are now enabled by default (Beta):
    • Server-side feature gate: TranslateStreamCloseWebsocketRequests
    • Client-side (kubectl) feature gate: KUBECTL_REMOTE_COMMAND_WEBSOCKETS
    • To disable RemoteCommand over WebSockets for kubectl, the environment variable feature gate must be explicitly set to false: KUBECTL_REMOTE_COMMAND_WEBSOCKETS=false. (#123281, @seans3)
  • Revised node selection based on container image location. The kube-scheduler now considers whether nodes have the required images available for init containers, and for sidecar containers if the cluster has 'SidecarContainers' enabled. (#123366, @kerthcet)
  • Scheduler extender ignorable option now handles errors for both filter and bind phases. (#122503, @sunbinnnnn)
  • The NodeLogQuery feature has been promoted to beta. No functional changes have been made from the alpha version. (#123205, @aravindhp)
  • The apiserver.latency.k8s.io/decode-response-object annotation was added to the audit log to record the decoding time. (#121512, @HirazawaUi)
  • The drop-in kubelet configuration directory feature has been targeted for Beta. (#122907, @sohankunkerkar)
  • The kubelet now rejects creating the pod if hostUserns=false and the CRI runtime does not support user namespaces. (#123216, @giuseppe)
  • The scheduler now retries Pods, which are failed by nodevolumelimits due to not found PVCs, only when new PVCs are added. (#121952, @sanposhiho)
  • Updated kubedns and nodelocaldns to release version 1.22.28. (#121908, @mzaian)
  • Users can now traverse all the pods that are in the scheduler and waiting in the permit stage through method IterateOverWaitingPods. In other words, all waitingPods in scheduler can be obtained from any profiles. Before this commit, each profile could only obtain waitingPods within that profile (#122946, @NoicFank)
  • ValidatingAdmissionPolicy now exclude TokenReview, SelfSubjectReview, LocalSubjectAccessReview, and SubjectAccessReview from all versions of authentication.k8s.io and authorization.k8s.io group. (#123543, @jiahuif)
  • When a PreFilterResult filters out certain Nodes, the scheduling framework now categorizes them as rejected via UnschedulableAndUnresolvable. Consequently, these nodes are excluded from the candidates for the preemption process. Additionally, this update corrects how the scheduling framework handles the Unschedulable status from PreFilter. Previously, if PreFilter returned Unschedulable, it could lead to an unexpected abortion in the preemption process, which shouldn't occur in the default scheduler but might occur in schedulers with custom plugins. (#119779, @sanposhiho)
  • When the RetryGenerateName feature gate is enabled on the kube-apiserver, create requests using generateName are retried automatically by the apiserver when the generated name conflicts with an existing resource name, up to a max limit of 7 retries. This feature is in alpha. (#122887, @jpbetz)
  • ValidatingAdmissionPolicy now supports type checking policies that utilize variables. (#123083, @jiahuif)
Documentation
  • Added a new internal metric kubelet_first_network_pod_start_sli_duration_second in the kubelet that allowed developers to understand the source of the latency problems on node startups. (#121720, @aojea)
  • A deprecated flag --pod-max-in-unschedulable-pods-duration was initially planned to be removed in v1.26, but we had to change this plan. We found an issue in which Pods can be stuck in the unschedulable pod pool for 5 min, and using this flag is the only workaround for this issue. This issue only could happen if you use custom plugins or if you change plugin set being used in your scheduler via the scheduler config. (#122013, @sanposhiho)
  • Modified the error message of unmanagedFatal to enhance clarity while preserving grammatical consistency with unmanagedWarning. This improvement ensures a more understandable prompt for users. (#120159, @Ithrael)
Failing Test
  • Reverted the change to "support sharing waitingPods across different scheduler profiles". (#124001, @kerthcet)
Bug or Regression
  • Added imagefs.inodesfree to default EvictionHard settings. (#121834, @vaibhav2107)
  • Changed the API server so that for admission webhooks that have a URL matching the hostname 'localhost' or a loopback IP address, the connection supports HTTP/2 where it could be negotiated. (#122558, @linxiulei)
  • Fixed CEL estimated cost for expressions that perform operations on the result of map() operations (e.g., .map(...).exists(...) ) to have the correct estimated cost instead of an unbounded cost. (#123562, @jpbetz)
  • Fixed a 1.27.0+ regression in kubeadm: The kubelet patch configuration will not be uploaded into the kube-system/kubelet-config ConfigMap anymore. (#123093, @SataQiu)
  • Fixed a bug in ValidatingAdmissionPolicy that caused policies using CRD parameters to fail to synchronize. (#123003, @alexzielenski)
  • Fixed a non-recursive list returning "resource version too high" error when consistent listing from cache is enabled. (#123674, @serathius)
  • Fixed a regression in kube-proxy introduced in version 1.26.0+ to make externalIPs workwith externalTrafficPolicy: Local. (#121919, @uablrek)
  • Fixed a regression in migration of in-tree vSphere volumes to the CSI driver introduced in version 1.29.0+. (#122341, @jsafrane)
  • Fixed a regression since 1.24 in the scheduling framework when overriding MultiPoint plugins (e.g. default plugins). The incorrect loop logic might have led to a plugin being loaded multiple times, consequently preventing any Pod from being scheduled, which was unexpected. (#122068, @caohe)
  • Fixed an issue where AvailableBytes sometimes did not report correctly on WindowsNodes when the PodAndContainerStatsFromCRI feature was enabled. (#122846, @marosset)
  • Fixed an issue where mount points could become local without calling NodePublishVolume after node rebooting. (#119923, @cvvz)
  • Fixed cleanup of Pod volume mounts when a file was used as a subpath. (#123052, @jsafrane)
  • Fixed error handling in EnsureAdminClusterRoleBindingImpl. (#122893, @danwinship)
  • Fixed incorrect error logging for syncCronJob. (#122493, @mengjiao-liu)
  • Fixed the deprecated version for pod_scheduling_duration_seconds that caused the metric to be hidden by default in 1.29. (#123038, @alculquicondor)
  • Fixed the disruption controller's PDB status synchronization to maintain all PDB conditions during an update. (#122056, @dhenkel92)
  • Improved scheduler performance when no scoring plugins were defined. (#122058, @aleksandra-malinowska)
  • Improved scheduler performance when no scoring plugins were defined. (#122435, @aleksandra-malinowska)
  • Improved scheduler performance when no scoring plugins were defined. (#123384, @aleksandra-malinowska)
  • In kubeadm, in the new output API 'output.kubeadm.k8s.io/v1alpha3', the UpgradePlan structure that is used when calling 'kubeadm upgrade plan ... -o yaml|json' was modified to include a list of multiple available upgrades. (#123461, @carlory)
  • Made decoding etcd's response respect the timeout context. (#121815, @HirazawaUi)
  • Previously, the scheduling queue didn't notice any extenders' failures, potentially resulting in missed cluster events and Pods rejected by Extenders being stuck in the unschedulable pod pool for up to 5 minutes in the worst-case scenario. Now, the scheduling queue notices extenders' failures and requeues Pods rejected by Extenders appropriately. (#122022, @sanposhiho)
  • QueueingHint implementation for NodeAffinity was reverted because potential scenarios were found where events that make Pods schedulable could be missed. (#122285, @sanposhiho)
  • Removed the incorrect warning event FileSystemResizeFailed during pod creation if it uses a readonly volume and the capacity of the volume is greater than or equal to its requested storage. (#122508, @carlory)
  • Restored the --verify-only function in code generation wrappers. (#123261, @skitt)
  • Reverted the EventedPLEG feature (beta, but disabled by default) back to alpha due to a known issue. (#122697, @pacoxu)
  • Used errors.Is() to handle errors returned by LookPath(). (#122600, @lzhecheng)
  • kube-proxy: Fixed LoadBalancerSourceRanges not working for nftables mode. (#122614, @tnqn)
  • kubeadm: fixed a bug where "kubeadm upgrade plan -o yaml|json" included unneeded output and was missing component config information. (#123492, @carlory)
  • Added metric name along with the utilization information when running kubectl get hpa. (#122804, @sreeram-venkitesh)
  • Addressed an issue where a JWT authenticator set up via --authentication-config would encounter failures in verifying tokens not signed with RS256. (#123282, @enj)
  • DRA: ResourceClaim and PodSchedulingContext status updates no longer allow changing object meta data. (#123730, @pohly)
  • Enabled deletion of pods that use raw block volumes on node reboot. (#122211, @gnufied)
  • Etcd: Updated to v3.5.11. (#122393, @mzaian)
  • Fixed Pod stuck in Terminating because of GenerateUnmapVolumeFunc missing globalUnmapPath when kubelet tries to clean up all volumes that failed reconstruction. (#123032, @carlory)
  • Fixed Windows credential provider, cannot find binary. Windows credential provider binary path may have ".exe" suffix so it is better to use LookPath() to support it flexibly. (#120291, @lzhecheng)
  • Fixed kubectl explain to show enum for field types if they were defined. (#123023, @ah8ad3)
  • Fixed a bug in kubeadm where the --rootfs global flag didn't work with "kubeadm upgrade node" for control plane nodes. (#123077, @neolit123)
  • Fixed a bug that an init container with containerRestartPolicy with Always cannot update its state from terminated to non-terminated for the pod with restartPolicy with Never or OnFailure. (#123323, @gjkim42)
  • Fixed a bug where kubectl drain would consider a pod as having been deleted if an error occurs while calling the API. (#122574, @brianpursley)
  • Fixed a potential data race in DRA with no known real-world implications. (#123222, @pohly)
  • Fixed a race condition in the iptables mode of kube-proxy in 1.27 and later that could result in some updates getting lost (e.g., when a service gets a new endpoint, the rules for the new endpoint might not be added until much later). (#122204, @danwinship)
  • Fixed a regression in "kubeadm init" where a user-specified --kubeconfig file was being ignored. (#122735, @avorima)
  • Fixed a regression in kubectl version 1.29.0 where the --attach flag was not honored. (#122447, @ardaguclu)
  • Fixed an error when trying to expand a volume that does not require node expansion. (#123055, @gnufied)
  • Fixed an issue calculating total CPU usage reported for Windows nodes. (#122999, @marosset)
  • Fixed an issue to ignore unnecessary node events and improve daemonset controller performance. (#121669, @xigang)
  • Fixed an issue where the configmap, secret, projected, and downwardAPI volume types didn't create user-visible files after a kubelet restart. This fix ensures data persistence and accessibility after restarts. (#122807, @carlory)
  • Fixed bug where health check could pass while APIServices are missing from aggregated discovery. (#122883, @Jefftree)
  • Fixed bug where providing a FieldPath to a CRD Validation Rule would erroneously affect the reported field path of other unrelated CRD Validation Rules on the same schema. (#123475, @alexzielenski)
  • Fixed enabling consistent list from watch cache that used to work for resourceVersion=0 (#123676, @serathius)
  • Fixed node lifecycle controller panic when conditionType ready is been patch nil by mistake. (#122874, @fusida)
  • Fixed panic of Evented PLEG during kubelet start-up. (#122475, @pacoxu)
  • Fixed resource deletion failure caused by quota calculation error when InPlacePodVerticalScaling is turned on. (#122701, @carlory)
  • For statically provisioned PVs, if their volume source is of CSI type or they have a migrated annotation, when they are deleted, the PersistentVolume controller won't change their phase to the Failed state. With this patch, the external provisioner can remove the finalizer in the next reconcile loop. Unfortunately, if a previously existing PV has the Failed state, this patch won't take effect. Users are required to remove the finalizer manually. (#122030, @carlory)
  • Improved the efficiency of NodeAdded QueueingHint by registering UpdateNodeTaint event for plugins that have NodeAdded event but don't have UpdateNodeTaint event. This ensures better requeuing efficiency and prevents Pods from being stuck in the unschedulable pod pool. (#122292, @sanposhiho)
  • JWTs used in service account and OIDC authentication are now strictly parsed to confirm that they use compact serialization. Other encodings were not previously accepted, but would result in different unspecific errors. (#123540, @enj)
  • Kube-apiserver: Fixed a 1.27+ regression in watch stability by serving watch requests without a resourceVersion from the watch cache by default, as in <1.27 (disabling the change in PR 115096 by default). This mitigates the impact of an etcd watch bug (https://github.com/etcd-io/etcd/pull/17555). If the 1.27 change in PR 115096 to serve these requests from underlying storage is still desired despite the impact on watch stability, it can be re-enabled with a WatchFromStorageWithoutResourceVersion feature gate. (#123935, @serathius)
  • Kubeadm: avoided uploading a defaulted flag value "--authorization-mode=Node,RBAC" for the kube-apiserver in the ClusterConfiguration stored in the "kube-system/kubeadm-config" ConfigMap. "Node,RBAC" are already the kubeadm defaults for this flag, so this action is redundant. (#123555, @neolit123)
  • Kubeadm: fixed a bug during kubeadm upgrade, where it is not possible to mount a new device and create a symbolic link for /etc/kubernetes (or a sub-directory) so that kubeadm stores its information on the mounted device. (#123406, @SataQiu)
  • Kubeadm: the kubelet-finalize phase of kubeadm init no longer requires the kubelet kubeconfig to have a specific authinfo. (#123171, @vrutkovs)
  • OpenAPI V2 will no longer publish aggregated API server OpenAPI for group versions that do not match the APIService specified group version. (#123570, @Jefftree)
  • Patched a leak of a discovery document that would occur when an Aggregated APIService changed its Spec.Service field and did not change it back. (#123517, @Jefftree)
  • Prevented watch cache starvation by moving its watch to separate RPC. Added a SeparateCacheWatchRPC feature flag to disable this behavior. (#123532, @serathius)
  • Reverted the QueueingHint implementation for NodeUnschedulable due to potential scenarios where events that make Pods schedulable could be missed. (#122288, @sanposhiho)
  • The PersistentVolume controller no longer automatically assigns a default StorageClass to Persistent Volume Claims (PVCs) with an empty storageClassName. (#122704, @carlory)
  • The initialization of nodes using external cloud providers now waits for the providerID value to be available before untainting it. This ensures that nodes are not declared Ready without necessary information such as the providerID and zone labels, which are required for integrations like load balancers to function correctly. Cloud providers that do not implement the GetInstanceProviderID method will not require the providerID to be set and will not fail to initialize the node for backward compatibility. (#123713, @aojea)
  • Updated google.golang.org/protobuf to v1.33.0 to resolve CVE-2024-24786. (#123758, @liggitt)
  • Updated the sample-apiserver manifest example to include correct RBAC configurations. (#123479, @Jefftree)
  • When initializing nodes using external cloud-providers, the process now waits for the providerID value to be available before declaring the node ready. This ensures that nodes are not marked as Ready prematurely due to communication errors with the cloud-provider. The providerID and zone labels are necessary for integrations such as load balancers to function correctly. Users can choose to opt out of this behavior by setting the feature flag OptionalProviderID in the cloud-controller-manager. (#123331, @aojea)
  • When using kubectl logs <pod-name> and the pod is not found, the error message now includes the namespace. Previously, the message would be "Error from server (NotFound): pods "my-pod-name" not found". Now, it reflects the namespace in the message as follows: "Error from server (NotFound): pods "my-pod-name" not found in namespace "default"". (#120111, @newtondev)
  • When using a claim with immediate allocation and a pod referencing that claim couldn't get scheduled, the scheduler incorrectly may have tried to deallocate that claim. (#122415, @pohly)
  • [kubeadam][structured authz] avoided setting default --authorization-mode when --authorization-config is provided (#123654, @LiorLieberman)
  • ValidateVolumeAttributesClassUpdate also validates new VolumeAttributesClass object. (#122449, @carlory)
Other (Cleanup or Flake)
  • Accepted zero as a default value for kubectl create token duration. (#123565, @ah8ad3)
  • Cleanup: removed getStorageAccountName warning messages. (#121983, @andyzhangx)
  • Client-go: Optimized leaders renewing leases by updating leader lock optimistically without getting the record from the API server first. Also, a new metric leader_election_slowpath_total was added to allow users to monitor how many leader elections are updated non-optimistically. (#122069, @linxiulei)
  • Locked the GA feature-gate ConsistentHTTPGetHandlers to default. (#122578, @carlory)
  • Migrated client-go/metadata to contextual logging. (#122225, @ricardoapl)
  • Removed the GA feature gate RemoveSelfLink. (#122468, @carlory)
  • Removed the generally available feature gate ExpandedDNSConfig. (#122086, @bzsuni)
  • Removed the generally available feature gate KubeletPodResourcesGetAllocatable. (#122138, @ii2day)
  • Removed the generally available feature gate KubeletPodResources. (#122139, @bzsuni)
  • Removed the generally available feature gate MinimizeIPTablesRestore. (#122136, @ty-dc)
  • The GA feature-gate APISelfSubjectReview has been removed, and the feature is unconditionally enabled. (#122032, @carlory)
  • Updated etcd to version 3.5.12. (#123150, @bzsuni)
  • Updated cri-tools to v1.29.0. (#122271, @saschagrunert)
  • Upgraded metrics server to v0.7.0. (#123504, @pacoxu)
  • kubeadm completion error message now displayed supported shell types when an invalid shell was specified. (#122477, @SataQiu)
  • kubeadm: ensured that a variety of API server requests were retried during "init", "join", "upgrade", "reset" workflows. Prior to this change, some API server requests, such as creating or updating ConfigMaps, were "one-shot" - i.e., they could fail if the API server dropped connectivity for a very short period of time. (#123271, @neolit123)
  • kubeadm: improved the overall logic, error handling, and output messages when waiting for the kubelet and API server /healthz endpoints to return OK. The kubelet and API server checks no longer ran in parallel, but one after another (in serial). (#121958, @neolit123)
  • Added an optimization to reduce stack memory usage for watch requests. It can be disabled with the feature gate: APIServingWithRoutine=false (#120902, @linxiulei)
  • Added warning for PV on reclaim policy when it is Recycle. (#122339, @carlory)
  • Deprecated the azureFile in-tree storage plugin. (#122576, @carlory)
  • Etcd image v3.5.12 has been built. (#123069, @bzsuni)
  • Fixed a bug in scheduler requeueing where registered wildcard cluster event sources didn't work. (#123117, @kerthcet)
  • Kubeadm: the bridge-nf-call-iptables=1 and bridge-nf-call-ip6tables=1 preflight checks are removed since not all the network implementations require this setting, network plugins are responsible for setting this correctly depending on whether or not they connect containers to Linux bridges or use some other mechanism. (#123464, @SataQiu)
  • Kubeadm: used ttlSecondsAfterFinished to automatically clean up the upgrade-health-check Job that runs during upgrade preflighting. (#122079, @carlory)
  • Migrated the kube-proxy to use contextual logging. (#122197, @fatsheep9146)
  • Promoted feature-gate LegacyServiceAccountTokenCleanUp to GA and locked it to default. (#122635, @carlory)
  • Removed GA featuregate about ExperimentalHostUserNamespaceDefaultingGate in 1.30. (#122088, @bzsuni)
  • Removed the GA feature gate for IPTablesOwnershipCleanup in version 1.30. (#122137, @bzsuni)
  • Removed the generally available feature gate ProxyTerminatingEndpoints. (#122134, @ty-dc)
  • The --cidr-allocator-type option set to CloudAllocator for kube-controller-manager will be deprecated and removed in a future release. Users are advised to transition to and explore the available options provided by their external cloud provider. (#123011, @dims)
  • The feature gate LegacyServiceAccountTokenTracking (GA since 1.28) is now removed because the feature is unconditionally enabled. (#122409, @Rei1010)
  • The in-tree cloud provider for Azure has now been removed. Please use the external cloud provider and CSI driver from https://github.com/kubernetes/cloud-provider-azure instead. (#122857, @nilo19)
  • The in-tree cloud provider for vSphere has been deprecated and removed. Users are advised to utilize the external cloud provider and CSI driver available at https://github.com/kubernetes/cloud-provider-vsphere. (#122937, @dims)
  • Updated kube-dns to v1.22.27. (#121736, @ty-dc)
  • Updated cni-plugins to version v1.4.0. (#122178, @saschagrunert)
  • Updated kubedns and nodelocaldns to version v1.23.0. (#123310, @bzsuni)
  • kube-proxy nftables mode is now compatible with kernel 5.4. (#122296, @tnqn)
  • Renamed Label cluster to storage_cluster_id for apiserver_storage_size_bytes metric (#124283, dims)
  • Bumped the stability level of apiserver_storage_size_bytes to STABLE (#123342, @logicalhan)
Uncategorized
  • Fixed an issue where kubectl apply could panic when imported as a library. (#122346, @Jefftree)
Dependencies
Added
Changed
Removed

Contributors, the CHANGELOG-1.30.md has been bootstrapped with v1.30.0 release notes and you may edit now as needed.


Published by your Kubernetes Release Managers.
Reply all
Reply to author
Forward
0 new messages