Today we are pleased to announce the release of Kubernetes 1.5. This release comes on the heels of KubeCon/CloudNativeCon. The conference was packed with users who had a great share of how they run their apps on Kubernetes. Many of you have expressed interest in running stateful applications in containers, and eventually have all applications running on Kubernetes. If you’ve been waiting to try to run a distributed database in Kubernetes, or looking for a way to secure the Application outage SLO (Service Level Indicator) for staid and stateless applications, this release has the solution for you.

StatefulSet and PodDisruptionBudget are now in beta. These features simplify the deployment and scaling of stateful applications and enable cluster operations such as node degradation without violating the application interrupt SLO.

You’ll also see various usability improvements in this release, starting with the frequently used Kubectl command line. For those who find it difficult to configure a multi-cluster federation, there is a new kubefed command to help. At the same time, one of the most popular high availability Master configuration scripts for configuring across multiple availability zones has been added to Kube-Up.

Did you know that the Kubernetes community is working on supporting Windows containers? If your team has. NET developers, take a look at the Windows container section of this release. The work is still in the early alpha stage, and we look forward to hearing from you.

Finally, for those interested in the Kubernetes kernel, 1.5 introduces the Container runtime interface, or CRI. It provides an internal interface that abstracts and separates the container runtime from Kubelet. This decoupled runtime design gives users the freedom to choose the runtime that best meets their needs. This release also introduced the containerized Node Conformance test, which verifies that a node’s software meets the minimum requirements for joining the Kubernetes cluster.

The latest features

StatefulSet Beta (formerly known as PetSet) allows workloads that require consistency identifiers or persistent storage to be created, scaled, deleted, or repaired on Kubernetes. You can use StatefulSet to simplify the deployment of any stateful service, tutorials and examples can be found in the code repository. To ensure that an identifier is never used by two pods at different times, The Kubernetes node controller no longer forces the removal of a Pod on an unresponsive node. Instead, it waits for the old Pod to die in one of the following ways: Automatically die when Kubelet reports and confirms that the old Pod has reached the end of its life; When the cluster administrator deletes the node, it automatically hangs. Or when the database administrator forcibly deletes the old Pod and confirms that the operation can continue. Users are now warned if they try to forcibly remove a Pod from the command line. For those migrating from PetSet to StatefulSet, Please follow the upgrade guide (http://kubernetes.io/docs/tasks/manage-stateful-set/upgrade-pet-set-to-stateful-set).

PodDisruptionBudget Beta is an API object that specifies the minimum number or percentage of copies of a Pod set that are running at any one time. With PodDisruptionBudget, application deployers can ensure that cluster operations that actively remove PODS will never take down too many pods at a time, resulting in data loss, service interruption, or unacceptable service degradation. In Kubernetes 1.5, the Kubectl drain command supports PodDisruptionBudget, which allows nodes to be cut in a safe manner during maintenance operations. This feature will also soon be available for node upgrades and cluster scaling (when nodes are removed). This is useful for quorum-based applications to ensure that the number of replicas running is never less than quorum sufficient, or for a WEB front-end application that the number of replicas providing load services is never less than a percentage.

Kubefed Alpha is a new command line tool to help manage federated clusters, making it easy to deploy a new federated control panel to perform cluster add and cluster delete operations in federated clusters. Another update to the cluster federation is the addition of ConfigMap Beta, DaemonSet Alpha, and deployments Alpha to the Fedoration API, which allows you to create, update, and delete these objects across a cluster using a single endpoint.

Highly available Master Alpha allows you to create or delete clusters with highly available (replicated) Masters using kube-up/ kube-Down scripts on GCE. It allows you to configure a high availability master distributed across availability zones, specify at least one copy of ETCD per availability zone, specify at least one API server per availability zone, and configure optional components such as Scheduler and Controller-manager distributed across availability zones.

Windows Server Container Alpha provides some initial support for Windows Server 2016 nodes and scheduling Windows Server Containers.

The Container Runtime Interface (CRI) Alpha introduces version V1 of the CRI API to support pluggable container runtimes. A pilot Docker-CRI integration is ready for testing and feedback.

Node conformance Testing Beta is a containerized testing framework that provides system validation and functional testing for nodes. This test verifies that the node meets Kubernetes minimum requirements; Nodes that pass the test are considered eligible to join Kubernetes. The node consistency test can be obtained from gcr. IO /google_containers/node-test:0.2 so that users can verify their node Settings.

Some of the highlights from our latest releases are worth noting. Full update list, please check the issue log (https://github.com/kubernetes/kubernetes/blob/master/CHANGELOG.md#v150).

access

Kubernetes 1.5 by making (https://github.com/kubernetes/kubernetes/releases/tag/v1.5.0) to download or via the get. K8s. IO. To get started Kubernetes, try the latest interactive tutorial (http://kubernetes.io/docs/tutorials/kubernetes-basics/). Don’t forget to try version 1.5 before the Christmas holidays.

The user USES the

It’s been a year and a half since GA launched, and Kubernetes’ user adoption continues to exceed expectations. The organizations running production workloads on Kubernetes already include the world’s largest companies, young startups, and everyone else before them. Because Kubernetes is open and can run anywhere, we’ve seen it used on a variety of platforms: Pokemon Go (Google Cloud), Ticketmaster (AWS), SAP (OpenStack), Box (physical machine), and a mash-up of the above. Here are some selected user examples:

  • Yahoo! JAPAN – Built an automated toolchain that simplifies the process from code submission to deployment, while running OpenStack on Kubernetes.

  • Walmart — Will use Kubernetes and OneOps to manage its powerful distribution centers to help teams speed up delivery, increase system online time, and optimize asset utilization.

  • Monzo, a European startup that is building a mobile-first bank, is using Kubernetes to support its core business platform that can cope with extreme performance and consistency requirements.

Kubernetes ecosystem

The Kuberete ecosystem is growing rapidly, including Microsoft’s support for Kubernetes in Azure Container Service, VMWare’s integration with Kubernetes in its Photon Platform, And Canonical’s commercial support for Kubernetes. This is in addition to more than thirty technology and service partners that already provide commercial services to Kubernetes users.

CNCF recently announced the KMSP (Kubernetes Managed Service Provider) program, which provides a pre-qualified platform for service providers who can help companies successfully adopt Kubernetes. After deepening the knowledge and understanding of Kubernetes, the Linux Foundation will work with CNCF to develop and operate a Kubernetes training and certification program — the first course designed is Kubernetes Basics (http://sep9.cn/1rcery).

Community speed

In the last three months we’ve seen over 100 new contributors join the project and push around 5,000 submissions, making the project reach a new milestone of 1000+ contributors and 4000+ submissions for the core project. Such a strong push comes from the adoption of open design, openness to new ideas, and the open community’s ability to embrace both new and existing contributors. Thanks to the 1.5 release team — Saad Ali from Google, Davanum Srinivas from Mirantis, and Caleb Miles from CoreOS — for their hard work in bringing 1.5 to life.