Kubernetes v1.37.0-alpha.2 is live!

21 views

Skip to first unread message

Jenny Shu

unread,

Jun 24, 2026, 10:51:29 PM (7 days ago) Jun 24

to kubernetes-announce, dev

Kubernetes Community,

Kubernetes v1.37.0-alpha.2 has been built and pushed using Golang version 1.26.4.

The release notes have been updated in CHANGELOG-1.37.md, with a pointer to them on GitHub:

v1.37.0-alpha.2

Downloads for v1.37.0-alpha.2

Source Code

filename	sha512 hash
kubernetes.tar.gz	834ebfaf9f6e895d3a74d31e0c046a19a9c499a0ce0f7dcbb4695fc85f8e5ab63fe8afff2639c152f4aace8d4ec72b9e849f926300a8a0f11548298a394e175d
kubernetes-src.tar.gz	2d6fc271a71bc6105428fa29837163fe5089f9dad28add0393c43d8d693be36cdbc179919f5cec21758e260edf85601420228e6105b0906a29b86567a2a3d61d

Client Binaries

filename	sha512 hash
kubernetes-client-darwin-amd64.tar.gz	86d617eca510a1de567120eadb2f6a45a38506abb495325b213766c25168f8dfb108a04c7a24d74b7252d110e02770f68e681f61c27cdcd2a53e849cdf472398
kubernetes-client-darwin-arm64.tar.gz	16fd148c6bb335f5507a329ece30762cd624ff8e4cf776ff84110ced1be2a478e4cc412b66624a982ca1e701271eaa2cfc11674b6c56b95124dd6fc21868e157
kubernetes-client-linux-386.tar.gz	1a734b4e839c25bc2b50cd764af075aabbabdf054855ed2762ca2eb1648f11fe047382ea888817cb27854d9cfed320896bb2251e16a68d1ea6d2add6c85c239e
kubernetes-client-linux-amd64.tar.gz	f619cb6ae2b2907134f9624521169996f8501c20f2d430bd2909e422e4501442d4840714673954902f11ef51867ab688ce79979e62c74542d19e377d7835e21a
kubernetes-client-linux-arm.tar.gz	70b83e9e72c4f75ac1e0925229e1bb55e358a7e14833a2c9878d09228a3b94ced4eed62b90ac6a257fb1e93ae3019da4a86219a4489b9a937c023b04bf1bc092
kubernetes-client-linux-arm64.tar.gz	a60bd2e8ff138b9be87054fce9ba43a5ae721d2c7922bad97a34ba2610e157746c376bc267ee62fb2fefed9e446dcc517c601efa5231eb965865c55f7e5fba5a
kubernetes-client-linux-ppc64le.tar.gz	ca201806431e2ffc7bddb355c4f194d738222f53288ead2a008d16c44ef494d5ee58da529f578b921ee13ca239527454b8556ab28b9dce1511dd852801a5e2d9
kubernetes-client-linux-s390x.tar.gz	a87ac2544a04c9cd7de5d80e4e9e0f933dcb05deee6fa0734999a6befacbf01e58eb060c249528e9ab08c6e45ddf8dbf637d1f293670c6ff33366f5255e6ca46
kubernetes-client-windows-386.tar.gz	f2d124c66d95496e55829b314ba29eba6fa494e44004a05272c7b6138dd98082695557d666d9466501390c6895b8eb374bcacffe71449e908a30efbc69a3816c
kubernetes-client-windows-amd64.tar.gz	1d6d31f249a283f576d4a75b95b802069b0d338b9e54a6d983eb7a595a5f6b110f38bcdb60ef90a030e96ab06e6d8666632f9cd720ed22bb07208d9041682310
kubernetes-client-windows-arm64.tar.gz	f4231567a6f99f94de243b42386c76cc577ae465de6550fc4f0889c9c1c8c195dfabd2c88351875bb8e90146df9958776ba86eeed0a3650efb46e343991f90a5

Server Binaries

filename	sha512 hash
kubernetes-server-linux-amd64.tar.gz	d13945b5e4fbcba87abbde01649f0f75bbd5121e1eadfd74a2f8f03c370ba68accd2d46694b9a94077e07e1655562b53025d43d33aa11e9e3714759fd8b80e65
kubernetes-server-linux-arm64.tar.gz	210ecd08861ad00a5c712128ca03eda5d23bf7570d026f65cabd5161f296006b47f3c56eb8ce703d0064f27c7e6be4fe2749d87011c9ce5d15c5071de714eabc
kubernetes-server-linux-ppc64le.tar.gz	c71e9f31155b4c16e63c5d90dbd75b48e80a3960b25cede95ff4aaedf1abfbd707f8b9a7cf39925b1134a4958887362b1147886a3690aae577521edd9ef06ecf
kubernetes-server-linux-s390x.tar.gz	55e5661fdb982c7922534d39e79c5cf6d15fe158fd97009ec1a9fd7c320eb2f513e3071464bc22731a87b6d6023becda47887370233dd9ec5fa259f5c8b2c82e

Node Binaries

filename	sha512 hash
kubernetes-node-linux-amd64.tar.gz	36327dca5cfcbcd934cff975066e98d4a503da63f8889028f367a12519dad0a3cd8d5268fb7d07ba2ccc4a6e55ccc822d04fe2059bcc6117d2872441f822ce5b
kubernetes-node-linux-arm64.tar.gz	dda000d9f40e21f48d3d52c728e26eb9b7677dccaf6c596869e2ddbdc436ce98429e37676fe7792e7b6246d725210d8fe79307bc159e40d42763c25ec533fb99
kubernetes-node-linux-ppc64le.tar.gz	2a4f6e5f2e90f4bc99f39f7bac1d8ebcd52e3a606e614c6ebf44c1306c30a0fd1076be8b3ab7852dba2e227c3b18288c843265110cd8728058151277cee5086b
kubernetes-node-linux-s390x.tar.gz	f82fd679ce1a80dfec4b91856fdac268dd8cc135e4bb7eb9734c5a437d856721314c3c2b6d4bf4e3a419eff6c3e8ddccf65d209fd60d1b96bf45c9162c8924d1
kubernetes-node-windows-amd64.tar.gz	439bf15f4e0bc4fbea622747a83ed6ecb02d931b84b4ab4abe729334038237db0029d4a9cc262494945c869533b5b1afeae14a324a0be6cc49423859646d2165

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:v1.37.0-alpha.2	amd64, arm64, ppc64le, s390x
registry.k8s.io/kube-apiserver:v1.37.0-alpha.2	amd64, arm64, ppc64le, s390x
registry.k8s.io/kube-controller-manager:v1.37.0-alpha.2	amd64, arm64, ppc64le, s390x
registry.k8s.io/kube-proxy:v1.37.0-alpha.2	amd64, arm64, ppc64le, s390x
registry.k8s.io/kube-scheduler:v1.37.0-alpha.2	amd64, arm64, ppc64le, s390x
registry.k8s.io/kubectl:v1.37.0-alpha.2	amd64, arm64, ppc64le, s390x

Changelog since v1.37.0-alpha.1

Changes by Kind

Deprecation

Kubeadm: if the user is providing a KubeProxyConfiguration with an empty value for the 'mode' field or if KubeProxyConfiguration is not provided, explicitly set the mode to 'iptables'. In 1.37 kube-proxy will start throwing a warning if the user has not explicitly set the field. This is part of the plan to switch the default mode to 'nftables' in a future release. (#139777, @neolit123) [SIG Cluster Lifecycle]

API Change

Cloud-provider: the NodeSyncPeriod field was moved from KubeCloudSharedConfiguration to CloudControllerManagerConfiguration.NodeLifecycleController.NodeMonitorPeriod (#137964, @niewysoki) [SIG API Machinery, Apps, Cloud Provider, Instrumentation and Node]
Fix CEL estimated cost of metadata.name and metadata.generateName for CRDs to match what the 253 limit that is validated by default unless the CRD author adds sets validations on the metadata fields. (#139573, @jpbetz) [SIG API Machinery]
MinCount can be modify after setting in PodGroup and PodGroupTemplate, modifying template is not influencing already existing podgroups. (#139279, @antekjb) [SIG API Machinery, Scheduling and Testing]
Rename podGroup condition from PodGroupScheduled to PodGroupInitiallyScheduled, to express clearly that this condition is set after the PodGroup first becomes scheduled successfully, and may not reflect the latest state of the PodGroup. (#139743, @antekjb) [SIG API Machinery, Scheduling and Testing]
The GangScheduling and WorkloadAwarePreemption feature gates were removed, and GenericWorkload is used instead to enable all core workload-aware scheduling functionalities altogether. (#139520, @macsko) [SIG API Machinery, Node, Scheduling and Testing]

Feature

Add informer metrics to apiserver (#139968, @michaelasp) [SIG API Machinery and Testing]
Added CBOR as a supported encoding for discovery endpoints and structured errors when the CBORServingAndStorage feature gate is enabled. (#139632, @benluddy) [SIG API Machinery and Testing]
Added a feature gate for Composite Pod Group. (#139407, @jdzikowski) [SIG Scheduling]
Bump coredns to 1.14.4 (#139735, @yashsingh74) [SIG Cloud Provider and Cluster Lifecycle]
Client-go: allow passing a custom GenerateKey field in the Config structure of the certificate manager. (#138999, @neolit123) [SIG API Machinery and Auth]
HPA: Enable scaling to and from zero by default. (#139648, @johanneswuerbach) [SIG Apps, Autoscaling and Testing]
Kube-apiserver, when using --enable-aggregator-routing=true, now load-balances requests evenly across all admission webhook endpoints. This ensures connection caching does not route all concurrent requests to a single backend endpoint. Cluster administrators can temporarily opt out of this new behavior using the WebhookRoundTripLoadBalancing feature gate (Beta, default true). (#139237, @aojea) [SIG API Machinery and Testing]
Optimized kube-scheduler performance for Pods with PersistentVolumeClaim mounts by processing only delta counts between scheduling cycles. (#139238, @yue9944882) [SIG Scheduling and Testing]
Relaxed DNS names for Services is now GA. (#139282, @adrianmoisey) [SIG Apps and Network]
This PR introduces the WatchListCompression feature gate (Beta, enabled by default). When enabled WatchList responses are compressed with gzip for clients that include Accept-Encoding: gzip. Regular Watch requests are unaffected. (#139308, @p0lyn0mial) [SIG API Machinery]

Failing Test

Fixed a bug when the DRADeviceTaintRules feature is enabled that caused kube-scheduler to panic when DeviceTaintRules exist and ResourceSlices are changed or to ignore new changes to DeviceTaintRules. (#139651, @nojnhuh) [SIG Node and Testing]

Bug or Regression

Add a way to wait on run finishing when closing an event handler. (#139755, @michaelasp) [SIG API Machinery]
An assumed pod is correctly removed from podGroupStates in cache, when it gets updated with deletion timestamp. (#138445, @iomarsayed) [SIG Scheduling]
Cri-api: Reverts to pre-1.34 JSON encoding of the KeyValue value field (#139964, @liggitt) [SIG Node]
Fix duplicated configs reported in resourceclaim status (#139732, @LionelJouin) [SIG Node]
Fix(kubelet): pods with subPath mounts stuck in error loop after FUSE/GlusterFS network filesystem disruption. (#139275, @yuehaii) [SIG Node, Storage and Testing]
Fixed a Windows kube-proxy issue where transient HNS downtime during restart/recovery could cause incorrect LoadBalancer state reconciliation, resulting in duplicate LoadBalancer creation failures with "Cannot create a file when that file already exists. (0xb7)" errors. (#139503, @princepereira) [SIG Network and Windows]
Fixed a bug where the kubelet node shutdown manager could leak dbus connections on repeated failures, eventually leading to thread exhaustion and crash. (#137141, @harche) [SIG Node]
Fixed a regression where the Job controller could attempt to report status.active as 0 while replacement Pod creation was deferred due to pod-failure backoff, causing the Job status update to be rejected by the apiserver. This could delay flushing uncounted terminated Pods, finalizer removal, and Job status updates, leaving Pods stuck Terminating and the Job with stale status until the backoff elapsed. (#139457, @akhilsingh-git) [SIG Apps]
Fixed a scheduler bug where clearing NominatedNodeName left pods tracked under an empty node key in the nominator. (#139904, @pacoxu) [SIG Scheduling]
Fixed inconsistent ephemeral-storage format between capacity and allocatable in node status by using DecimalSI format for ephemeral-storage capacity. (#137652, @0xMH) [SIG Node]
Kube-proxy now removes stale conntrack entries when a UDP service no longer has any serving endpoints (e.g. scaled down to zero), preventing previously established one-way UDP flows from being blackholed to deleted pod IPs indefinitely. (#139629, @Bafff) [SIG Network]
Kubeadm: Improved resilience of kubeadm etcd learner promotion. kubeadm now correctly handles cases where learner promotion succeeds on the etcd side but a transient client-side error is returned, preventing unnecessary etcd-join failures. (#139842, @jihyun-huh) [SIG Cluster Lifecycle]
Kubeadm: during "kubeadm join", use the KubernetesAPICall timeout (default 1 minute) when fetching the kubeadm-config ConfigMap, instead of the short 350ms retry used for optional component configs. A new shortConfigMapGet parameter is added to FetchInitConfigurationFromCluster so that callers like "kubeadm reset" can still use the short retry. (#139667, @damdo) [SIG Cluster Lifecycle]
The PodReadyToStartContainers condition now includes a diagnostic message when Status is False, explaining why the pod sandbox is not ready (e.g., "pod sandbox has no IP address", "no pod sandbox exists"). This improves debuggability for pods stuck in ContainerCreating state without requiring access to node logs. (#135300, @harche) [SIG Node]

Other (Cleanup or Flake)

Empty requests field in config status of resourceclaim when the config applies to all requests (#139731, @LionelJouin) [SIG Node]
Kubelet: the Topology Manager now returns a clearer error when the prefer-closest-numa-nodes policy option is enabled on a windows node that does not expose NUMA distance information, explaining that the option is not supported there. (#139760, @zylxjtu) [SIG Node]
The service/proxy subresource of the apiserver now uses the EndpointSlices of the service rather than the Endpoints. (This will have no effect unless you were previously using that API with a Service for which you constructed the Endpoints by hand and disabled EndpointSlice mirroring of them.) (#134860, @danwinship) [SIG API Machinery and Network]

Dependencies

Added

Nothing has changed.

Changed

Removed

Nothing has changed.

Contributors, the CHANGELOG-1.37.md has been bootstrapped with v1.37.0-alpha.2 release notes and you may edit now as needed.