Week Ending June 30, 2019

Release Schedule

Next Deadline: 1.16 Release Cycle begins, July 1st

We’ve started 1.16, and a schedule for the release is up. Important dates:

• July 30th: Enhancements Freeze
• August 29th: Code Freeze
• Sept. 9th: Docs complete
• Sept. 16th: Release

So, time to start marking your enhancment proposals with milestone v1.16, and work on those KEPs.

Featured Merges

Runtimeclass admission #78484

Creates a new admission controller for RuntimeClass, primarily so that we can account for resources tied to the pod sandbox.

Volume Scheduling Limits #77595

SIG Storage has been working for a while to give you the ability to limit how many volumes of each type that can be deployed on a node. This PR implements most of that, in a way that’s the same for both CSI and in-tree providers.

Cleanup Week

The first week of a new release cycle is often a “cleanup week” where devs take care of a whole lot of pending refactoring that has been waiting since last release, or longer. This week is no exception:

• The alpha metadata.initializers field, deprecated since 1.13, got removed
• Remove the feature gates for CSI, since it’s been GA
• Clean up 5 different self-set Node labels, and prevent Nodes from changing labels they shouldn’t
• Refactor pkg/util/mount so that it can be included in external code, such as CSI driver repos
• Get the Conformance Test Walker to understand nested Describes so that it can handle complex tests correctly
• Replace the “pause” Pod with the platform-neutral agnhost Image so that our whole e2e test framework will work for Conformance
• Clean up the upgrade tests so that they don’t restart all pods on APIserver upgrade
• Add unit test coverage for APIextensions
• Refactor how bootstrap tokens are handled in Kubeadm

Other Merges

• The CLI stops printing duplicate error messages
• Azure Load Balancer understands an empty string as “default”
• kubelet retries Sandbox creation instead of hanging in ContainerCreating state
• Don’t panic the APIserver
• kubeadm reset removes etcd nodes correctly
• Support Vulkan GPU workloads
• Kase matters: s/KB/kB/ when dealing with the Linux kernel
• Account for Init containers when calculating QoS
• Allow kubectl logs --tail=-1
• Stop depending on the Docker SDK for kubeadm tests

Last Week In Kubernetes Development (LWKD) is a product of multiple contributors participating in Kubernetes SIG Contributor Experience. All original content is licensed Creative Commons Share-Alike, although linked content and images may be differently licensed. LWKD does collect some information on readers, see our privacy notice for details.

You may contribute to LWKD by submitting pull requests or issues on the LWKD github repo.