The most recent variation of Kubernetes, 1.20, introduces the pursuing main adjustments:
  • The Docker runtime is being deprecated. Having said that, this doesn’t indicate Docker photographs or Dockerfiles do not perform in Kubernetes anymore. It just means Kubernetes will now use its have Container Runtime Interface (CRI) product or service to execute containers as an alternative of the Docker runtime. For most users this will have no sizeable impact—e.g., any present Docker photographs will perform fine. But some issues may final result when dealing with runtime useful resource limits, logging configurations, or how GPUs and other special hardware interact with the runtime (some thing to note for individuals employing Kubernetes for device mastering). The earlier link delivers information on how to migrate workloads, if needed, and what issues to be informed of.
  • Quantity snapshot operations are now stable. This enables volume snapshots—images of the state of a storage volume—to be utilised in output. Kubernetes apps that rely on highly precise state, this sort of as photographs of databases files, will be a lot easier to develop and sustain with this element active.
  • Kubectl Debug is now in beta, allowing prevalent debug workflows to be performed from inside the kubectl command-line surroundings. 
  • API Precedence and Fairness (APF) is now enabled by default, though even now in beta. Incoming requests to kube-apiserver can be sorted by priority stages, so that the administrator can specify which requests must be content most straight away.
  • Approach PID Limiting is now in basic availability. This element makes certain that pods can’t exhaust the amount of course of action IDs out there on a Linux host, or interfere with other pods by employing up also quite a few processes.

Kubernetes 1.seventeen, produced in December 2019, introduced the pursuing crucial new features and revisions: 

  • Quantity snapshots, introduced in alpha in Kubernetes 1.12, are now promoted to beta. This element enables a volume in a cluster to be snapshotted at a provided minute in time. Snapshots can be utilised to provision a new volume with information from the snapshot, or to roll back an present volume to an before snapshotted variation. Quantity snapshots make it feasible to conduct elaborate information-versioned or code-versioning functions within a cluster that weren’t previously feasible.
  • More of the “in-tree” (included by default) storage plug-ins are now being moved to the Container Storage Interface (CSI) infrastructure. This means fewer direct dependencies on individuals motorists for the core variation of Kubernetes. Having said that, a cluster has to be explicitly updated to assist migrating the in-tree storage plug-ins, but a productive migration should not have any ill results for a cluster.
  • The cloud company labels element, originally introduced in beta back in Kubernetes 1.two, is now normally out there. Nodes and volumes are labeled based on the cloud company where by the Kubernetes cluster runs, as a way to explain to the relaxation of Kubernetes how individuals nodes and volumes must be dealt with (e.g., by the scheduler). If you are employing the before beta variations of the labels your self, you must improve them to their new counterparts to prevent difficulties.

Exactly where to obtain Kubernetes

You can obtain the Kubernetes supply code from the releases web page of its official GitHub repository. Kubernetes is also out there by way of the improve course of action supplied by the several sellers that provide Kubernetes distributions.

What is new in Kubernetes 1.sixteen

Kubernetes 1.sixteen, produced in September 2019, has the pursuing new and revised features:

  • Personalized useful resource definitions (CRDs), the extended-advised mechanism for extending Kubernetes features introduced in Kubernetes, are now formally a normally out there element. CRDs have previously been broadly utilised by third functions. With the go to GA, quite a few optional-but-advised behaviors are now necessary by default to retain the APIs stable.
  • Many adjustments have been produced to how volumes are dealt with. Main among them is going the volume resizing API, found in the Container Storage Interface (CSI), to beta.
  • Kubeadm now has alpha assist for signing up for Windows worker nodes to an present cluster. The extended-phrase target here is to make Windows and Linux nodes both equally initially-class citizens in a cluster, as an alternative of owning only a partial established of behaviors for Windows.
  • CSI plug-in assist is now out there in alpha for Windows nodes, so individuals systems can start off employing the exact assortment of storage plug-ins as Linux nodes.
  • A new element, Endpoint Slices, enables for greater scaling of clusters and far more overall flexibility in handling network addresses. Endpoint Slices are now out there as an alpha test element.
  • The way metrics are dealt with proceeds a main overhaul with Kubernetes 1.sixteen. Some metrics are being renamed or deprecated to deliver them far more in line with Prometheus. The approach is to take out all deprecated metrics by Kubernetes 1.seventeen.
  • At last, Kubernetes 1.16 removes a amount of deprecated API variations. 

What is new in Kubernetes 1.15

Kubernetes 1.15, produced in late June 2019, delivers the pursuing new features and improvements:

  • More features (at the moment in alpha and beta) for Personalized Source Definitions, or CRDs. CRDs in Kubernetes are the basis of its extensibility know-how, allowing Kubernetes cases to be customized without slipping out of conformance with upstream Kubernetes benchmarks. The new features include the ability to convert CRDs between variations (some thing extended out there for native assets), OpenAPI publishing for CRDs, default values for fields in OpenAPI-validated schemas for CRDs, and far more.
  • Native superior availability (HA) in Kubernetes is now in beta. Placing up a cluster for HA even now necessitates organizing and forethought, but the extended-phrase target is to make HA feasible without any third-party software program.
  • More plug-ins that handle volumes have been migrated to use the Container Storage Interface (CSI), a dependable way to handle storage for hosted containers. Among the new features introduced in alpha for CSI are volume cloning, so that new persistent volumes can be based on an present just one.

Other adjustments in Kubernetes 1.15 include:

  • Certificate management now automatically rotates certificates prior to expiration.
  • A new framework for plug-ins that conduct scheduling functions has entered alpha.

What is new in Kubernetes 1.14

Variation 1.14 of Kubernetes, produced in March 2019, has the pursuing adjustments:

  • Microsoft Windows Server 2019 is now formally supported as a system for working both equally Kubernetes worker nodes and container scheduling. This means complete Kubernetes clusters can operate on Windows completely, rather than owning a mix of Windows and Linux systems.
  • The plugin mechanism for Kubectl, the default Kubernetes command-line tool, is now a stable element, letting developers apply their have Kubectl subcommands as standalone binaries.
  • Persistent community volumes are now a stable element. This lets locally hooked up storage be utilised by Kubernetes for persistent volumes. Apart from giving superior performance than employing network-hooked up storage, it also makes it a lot easier (and probably more cost-effective) to stand up a cluster.
  • Approach ID limiting for Linux hosts is now a beta element. This prevents any just one pod from employing up also quite a few course of action IDs and therefore leading to useful resource exhaustion on the host.

What is new in Kubernetes 1.13

Variation 1.13 of Kubernetes was produced in December 2018, with the pursuing new and upgraded features:

  • Kubeadm, a tool intended to make it a lot easier to established up a Kubernetes cluster, is finally out there as a entirely supported element. It walks an admin by the basic principles of location up nodes for output, signing up for them to the cluster, and making use of greatest practices along the way. It also delivers a way for infrastructure-orchestration equipment (Puppet, Chef, Salt, and many others.) to automate cluster setup.

  • The Container Storage Interface, or CSI, is now also out there as a supported element. CSI enables extensions for Kubernetes’s volume layer, so that storage plugins can perform with Kubernetes without owning to be produced portion of Kubernetes’s core code.

  • Kubernetes now employs CoreDNS as its default DNS server. CoreDNS is effective as a drop-in replacement for other DNS servers, but was designed to integrate with Kubernetes by way of plug-ins and integration with Kubernetes features this sort of as Prometheus checking metrics.

What is new in Kubernetes 1.12

Launched in late September 2018, Kubernetes 1.12 provides to basic availability the Kubelet TLS Bootstrap. The Kubelet TLS Bootstrap enables a Kubelet, or the major agent that runs on each Kubernetes node, to sign up for a TLS-secured cluster automatically, by requesting a TLS customer certification by an API. By automating this course of action, Kubernetes enables clusters to be configured with increased safety by default.

Also new in Kubernetes 1.12 is assist for Microsoft Azure’s digital device scale sets (VMSS), a way to established up a team of VMs that automatically ramp up or down on schedule or to fulfill need. Kubernetes’s cluster-autoscaling element now is effective with VMSS.

Other new features in Kubernetes 1.12:

  • Snapshot and restore features for volumes (alpha).
  • Personalized metrics for pod autoscaling (beta). This enables custom made standing conditions or other metrics to be utilised when scaling a pod—for occasion, if assets that are precise to a provided deployment of Kubernetes require to be tracked as portion of the application’s management tactic.
  • Vertical pod scaling (beta), which enables a pod’s useful resource limits to be diversified throughout its life time, as a way to superior handle pods that have a superior expense connected with disposing of them. This is a extended-standing product on quite a few want lists for Kubernetes, due to the fact it enables for methods to offer with pods whose behaviors are not easy to handle under the recent scheduling tactic.

What is new in Kubernetes 1.eleven

Launched in early July 2018, Kubernetes 1.eleven adds IPVS, or IP Digital Server, to delivers superior-performance cluster load balancing employing an in-kernel know-how that’s fewer complex than the iptables program usually utilised for this sort of issues. Eventually, Kubernetes will use IPVS as the default load balancer, but for now it is opt-in.

Personalized useful resource definitions, billed as a way to make custom made configuration adjustments to Kubernetes without breaking its standardizations, could now be versioned to allow for swish transitions from just one established of custom made assets to an additional over time. Also new are approaches to define “status” and “scale” subresources, which can integrate with checking and superior-availability frameworks in a cluster.

Other main adjustments include: