The first release of Kubernetes in 2020 is published. With v1.18 comes 38 enhancements. We picked some of the most interesting enhancements, hope they are also interesting to you.
Pod Troubleshooting with 'kubectl debug'
The alpha release of this new CLI resource is an addition to the 'kubectl exec' command and should support admins & developers in troubleshooting pods. The command will start an ephemeral container next to the pod which needs to be debugged and is attached to the console for easy access. You can find the design document here.
EndpointSlice API now GA
The GA Endpoint API is an replacement of the current core/v1 Endpoints API. The current Core/V1 Endpoints API comes with severe performance/scalability drawbacks affecting multiple components in the control-plane (apiserver, etcd, endpoints-controller, kube-proxy). The EndpointSlice API has the motivation to
- Support tens of thousands of backend endpoints in a single service on cluster with thousands of nodes.
- Move the API towards a general-purpose backend discovery API.
- Leave room for foreseeable extension like:
- Support multiple IPs per pod
- More endpoint states than Ready/NotReady
- Dynamic endpoint subsetting
Find the details here.
The community screemed for it, Kubernetes heared you! The Certificates API enables automation of x509 credential provisioning by providing a programmatic interface for clients of the Kubernetes API to request and obtain x509 certificates from a Certificate Authority (CA). The enhancement is described in depth in this request.
A typical successful issuance proceeds as follows.
Good news for Windows folks! containerd is now supported unter Windows
The containerd 1.3 release supports now Windows 2019 under Kubernetes 1.18. Here are the details. This is a great achievment as containenrd utilize the host container service (HCS v2) in Windows Server. This means a higher compatibility of the Kubernetes API and deeper control of containers.
Big Data and Data Processing will love it - the HugePage gets an Extension
While HugePage is since a while in stable, this feature is a big turn key for many data heavy projects. With HugePage you can reserve a blog of memory which is due to hardware restrictions faster accessable. Newly introduced by the team is the possibility to request multiple HugePages. In addition the Container isolation is now in place so no memory resource starvation can happen.
Maybe also interesting to you in terms of better ressource consumption/allocation: Node Topology Manager graduating to beta in 1.18
If you are interested to contribute to K8s, SIG Scaleability is currently looks for some hands! https://kubernetes.io/blog/2020/03/19/join-sig-scalability/