Kubernetes Release Notes

Validated on 1 Jul 2025

July 2025

1 July

  • The DOKS database operator, which facilitated integration of DOKS clusters with managed databases, is deprecated. Your existing databases remain operational and the operator-related CRDs are not removed. You can use the DigitalOcean API, CLI, or control panel to manage the databases created with the operator.

June 2025

30 June

20 June

  • The DOKS Cluster Autoscaler now supports autoscaling across multiple node pools for a single workload. If the preferred node pool is at capacity, the autoscaler uses the next compatible option in sequence — ensuring scaling continues smoothly without disruption. This feature is in private preview. For more information, see Flexible Node Pool Selection.

19 June

  • We have reverted to non-VPC-native cluster networking when creating new clusters with Kubernetes version 1.33 and higher using the doctl command line interface or API. If your VPC-native cluster provisioned successfully, no action is needed. If the cluster appears to be stuck in the Provisioning state, please delete the cluster and create a new one.

    You can automatically configure VPC-native networking when using the control panel or manually configure it usingdoctl or the API. For more information, see Create a Cluster with VPC-native Networking.

16 June

  • We have enabled the Gateway API by default on clusters with VPC-native networking running on Kubernetes version 1.33 or higher. This feature is in public preview. You can use the Gateway API to manage ingress natively in a Kubernetes cluster, without the need for third-party ingress controllers.

  • Load balancer services now default to network load balancers (NLB) instead of regional load balancers on clusters running on Kubernetes version 1.33 or higher. Use the service.beta.kubernetes.io/do-loadbalancer-type annotation with a value of REGIONAL or REGIONAL_NETWORK to change the load balancer type.

  • We have enabled VPC-native cluster networking by default when you create new clusters using Kubernetes version 1.33 or higher using the doctl command line interface or API. For more information, see Create a Cluster with VPC-native Networking.

12 June

  • AMD Instinct MI300X GPUs are now available for DOKS clusters. These GPUs are available in a single GPU or 8 GPU configuration when creating new clusters or adding node pools to existing clusters. You can use these GPUs in the control panel and the API with the gpu-mi300x1-192gb and gpu-mi300x8-1536gb slug, respectively.

May 2025

7 May

  • NVIDIA RTX 4000 and RTX 6000 Ada Generation GPUs are now available for DOKS clusters. You can choose these GPUs when creating new clusters or adding node pools to existing clusters using the control panel and the API with the gpu-4000adax1-20gb and gpu-6000adax1-48gb slug respectively.

April 2025

30 April

March 2025

31 March

  • DOKS now supports up to 1000 worker nodes per cluster, allowing you to run larger workloads within a single cluster. You can create 1000-node clusters using the DigitalOcean API or CLI. To learn more, see Scale Up to 1000 Nodes Per Cluster.

27 March

  • NVIDIA L40S GPUs are now available for creating new clusters or adding node pools to existing clusters using the control panel and the API with the gpu-l40sx1-48gb slug.

  • Network load balancers are now in general availability. Network load balancers are regional load balancers that route traffic at the TCP/UDP transport level. Read How to Create Regional Load Balancers to get started.

24 March

  • You can now view the most recent reconciler checks in the DigitalOcean Control Panel. For more information, see Best Practices.

  • The DOKS Cluster Autoscaler can now scale nodes in a node pool down to zero. This feature is in private preview. See Scaling to Zero for recommendations to follow for scaling down to zero.

10 March

  • The routing agent is now in public preview for DOKS clusters. The agent lets you create and maintain IP routes on your Kubernetes worker nodes. You can enable or disable the routing agent only using the DigitalOcean CLI or API. For more information, see How to Use the Routing Agent.

February 2025

28 February

  • IPv6 support for regional external load balancers is now in general availability. You may now choose between IPv4-only or dual-stack IPv4 and IPv6 networking when you create standalone load balancers. All DOKS load balancers created after 28 February 2025 are dual-stack.

  • Network load balancers are now in public preview. Network load balancers are regional load balancers that route traffic at the TCP/UDP transport level. You must opt-in to the public preview to create network load balancers.

January 2025

7 January

  • You can scale a fixed node pool down to 0 nodes using the DigitalOcean Control Panel, CLI, and API. Your cluster must have another fixed node pool with at least 1 node or a GPU node pool with 0 node to do so. To learn more, see Add Node Pools.

December 2024

12 December

October 2024

29 October

15 October

1 October

September 2024

18 September

13 September

  • SnapShooter can now back up and restore cluster-internal resources such as databases on DigitalOcean Kubernetes clusters. The new feature automatically installs a backup agent into your cluster, allowing SnapShooter to back up a variety of database types running in DOKS.

    See How to Back Up Resources on DigitalOcean Kubernetes with SnapShooter for more details.

12 September

August 2024

30 August

29 August

  • We have increased the volume attach limit for DOKS nodes from 7 to 15.

July 2024

16 July

  • DOKS now supports the LoadBalancerSourceRanges field in the load balancer service configuration file. This field specifies a list of IP addresses from which traffic can pass to the load balancer.

    We have deprecated the service.beta.kubernetes.io/do-loadbalancer-allow-rules annotation in favor of the LoadBalancerSourceRanges field.

June 2024

27 June

24 June

  • DigitalOcean Load Balancers added to DOKS clusters now default to Kubernetes’ recommended health check configuration which facilitates worker node replacements with minimal request disruption. The new configuration automatically updates for all existing managed load balancers on DOKS 1.26 and later.

    We do not recommend configuring health checks manually. You can continue the previous behavior by setting the service.beta.kubernetes.io/do-loadbalancer-override-health-check annotation as described here.

13 June

  • We are moving the managed Cilium Operator component (cilium-operator) from the worker nodes to the control plane of DOKS clusters. This frees up resources on the worker nodes and improves autoscaling of the component.

April 2024

8 April

  • Tokens returned by the /kubeconfig and /credentials endpoints now have custom scopes to provide read-only access to Kubernetes resources. Within DOKS clusters, operations to access Kubernetes objects are still available based on team role (owner, biller, or member) as before.

March 2024

19 March

  • We have added synchronous validation of LoadBalancer service annotations. If you provide invalid values, DOKS returns an error, thus preventing misconfiguration of your load balancer.

January 2024

31 January

18 January

  • All currently supported DigitalOcean Kubernetes versions now have Cilium Hubble, Hubble Relay and Hubble UI enabled. For more information, see Use Cilium Hubble.

October 2023

17 October

  • The Kubernetes API endpoints /v2/kubernetes/clusters/<cluster ID>/kubeconfig and /v2/kubernetes/clusters/<cluster ID>/credentials now require API tokens to have write scope.

May 2023

16 May

  • Premium Intel CPUs are now available for CPU-Optimized Droplets in TOR1.

April 2023

18 April

  • Premium Intel CPUs are now available for CPU-Optimized Droplets in BLR1.

March 2023

24 March

  • We have extended the promotional period for CPU-Optimized Droplets with Premium Intel CPUs (no billing for outbound data transfer at speeds faster than 2 Gbps) from 30 April 2023 to 30 June 2023. Learn more about bandwidth billing.

23 March

  • Premium Intel CPUs are now available for CPU-Optimized Droplets in SFO2.

February 2023

22 February

  • Premium Intel CPUs are now available for CPU-Optimized Droplets. You can create CPU-Optimized Droplets with Premium Intel CPUs in NYC1, NYC3, FRA1, AMS3, SFO3, and SYD1.

    Compared to CPU-Optimized Droplets with Regular Intel CPUs, CPU-Optimized Droplets with Premium Intel CPUs have the latest hardware and five times more network throughput.

    Additionally, for a promotional period from 1 February through 30 April 2023, we will not bill for outbound data transfer at speeds faster than 2 Gbps for CPU-Optimized Droplets with Premium Intel CPUs. Learn more about bandwidth billing.

    You can use this plan for both standalone Droplets and Kubernetes nodes. You can also resize your existing Droplets to this node plan.

January 2023

24 January

  • We have deprecated our legacy load balancer scaling system in all datacenter regions. This includes the deprecation of the do-loadbalancer-size-slug annotation for DigitalOcean Kubernetes load balancers.

    Horizontal scaling is now available in all regions.

November 2022

16 November

  • DigitalOcean Kubernetes clusters originally created with version 1.20 or older have an outdated version of our control plane architecture, which does not allow you to enable high availability. However, you can now upgrade your control plane to our new version. This upgrade option is available for Kubernetes versions currently 1.22 and later.

    To check whether you can upgrade your cluster to the new control plane, see our guide.

  • You can now enable high availability on existing Kubernetes clusters. For detailed steps, see our guide.

October 2022

13 October

  • When creating a new Kubernetes cluster, you can add a free database operator (now in beta), which allows you to automatically link new databases to your cluster. For more details, see documentation on GitHub.

  • do-operator, our operator for managing and consuming DigitalOcean resources from a Kubernetes cluster, is now an open-source beta project.

August 2022

18 August

July 2022

1 July

  • DOKS clusters now accrue free bandwidth based on the worker pool’s largest sizes within 28 days of usage. Learn more about DOKS billing.

    Previously, you may have received slightly more free bandwidth on 30 and 31-day months. Individual worker nodes were billed per hour, up to a maximum of 744 hours per month (31 days * 24 hours). As a result, they could accrue extra bandwidth allowance beyond the advertised monthly allowance for the corresponding Droplet plan.

June 2022

27 June

  • To improve security, DigitalOcean no longer accepts TLS 1.0 and TLS 1.1 connections. This includes connections to www.digitalocean.com, cloud.digitalocean.com, and api.digitalocean.com.

23 June

March 2022

15 March

11 March

October 2021

12 October

  • Released v1.65.0 of doctl, the official DigitalOcean CLI. This release includes a number of new features:

    • The --ha flag was added to the kubernetes cluster create sub-command to optionally create a cluster configured with a highly-available control plane. This feature is in early availability
    • The kubernetes cluster sub-commands now include a “Support Features” field when displaying version options
    • The --disable-lets-encrypt-dns-records flag was added to the compute load-balancer create sub-command to optionally disable automatic DNS record creation for Let’s Encrypt certificates that are added to the load balancer

  • High-availability control plane is now in early availability in the following regions: ams3, nyc1, sfo3, and sgp1.

August 2021

3 August

March 2021

18 March

  • You can now do the following on Kubernetes clusters:

    • Use surge upgrade when upgrading an existing cluster. Surge upgrade is enabled by default when you create a new cluster.

    • Move a Kubernetes cluster and its associated resources, such as Droplets, load balancers and volumes, to a project using the DigitalOcean Control Panel or doctl command-line tool. You can also assign a project when you create a new cluster. If you do not specify a project, it gets assigned to the default project.

    • Delete resources, such as load balancers and volumes, associated with a Kubernetes cluster using the DigitalOcean Control Panel, API or the doctl command-line tool.

October 2020

27 October

6 October

April 2020

28 April

  • The DigitalOcean Virtual Private Cloud (VPC) service is now available for all customers. VPC replaces the private networking service. Existing private networks will continue to function as normal but with the enhanced security and features of the VPC service. See the description of VPC features for more information.

7 April

February 2020

5 February

  • v1.14.0 of the DigitalOcean Terraform Provider is now available. This release includes a bug fix for projects containing many resources and exposes the Droplet IDs for individual nodes in Kubernetes clusters.

4 February

  • Released v1.38.0 of doctl, the official DigitalOcean CLI. This release adds the ability to set Kubernetes node pool labels as well as support for deleting multiple Kubernetes clusters with a single command.

November 2019

19 November

October 2019

3 October

June 2019

17 June

  • 6-hour and 1-day alert policies for Droplets and Kubernetes worker nodes have been deprecated. No new alert policies with these intervals can be created. Existing alert policies using these intervals will remain in place until 1 August 2019, at which point they will be modified to reflect a 1-hour interval.

May 2019

21 May

3 May

April 2019

24 April

  • DOKS node pools can now be named at creation time.

  • DOKS master nodes now automatically rotate logs to avoid disk space issues.

March 2019

20 March

  • DOKS customers can now see the cost of their Kubernetes nodes and load balancers aggregated by cluster name within a Kubernetes clusters group on their invoice. Volumes and volume snapshots used in a DOKS cluster are not yet included in the cluster aggregation.

December 2018

11 December

  • The following updates were released for DigitalOcean Kubernetes:

    • Any user can opt into Kubernetes during early availability via the control panel.
    • Users can rename clusters.
    • Users can edit tags associated with clusters and worker pool.
    • A guided walkthrough helps users set up their kubectl and kubeconfig properly and provides example manifests to make it easier to get started.
    • The node pool settings to add, remove, and scale node pools has been moved to the ‘Nodes’ tab for a more streamlined experience.
    • The cluster provisioning status bar more accurately reflects the cluster creation progress.
    • Users can view the estimated monthly cost for their clusters.
    • Users can view a breakdown of total cluster capacity including CPU, Memory, Disk cluster wide.
    • Users can create and manage Kubernetes clusters, worker pools, and configuration using doctl.

7 December

  • The minimum size for a Kubernetes node was changed to the 2 GB Memory / 1 vCPU plan.

October 2018

1 October

We can't find any results for your search.

Try using different keywords or simplifying your search terms.