Week Ending November 15, 2020

[github-actions]

Developer News

It’s Kubecon Week! And there’s a ton of Kubernetes community sessions scheduled. See the end of LWKD for a full list.

Security Advisory: contributor Qin Ping found a way for unauthorized users to crash your backups via VolumeSnapshot. Fix this by either restricting access to the snapshot storage API, or by upgrading to snapshot-controller v3.0.2.

GitHub is having as-yet-unexplained fetch errors with some PRs. This is affecting builds and messed up the release of 1.18.11.

SIG-CLI has promoted Eddie Zaneski to Lead, and Tim Pepper has stepped down from SIG-Release to be replaced by Sascha Grunert and Dan Mangum.

Please share your feedback on how the Triage Workflow changes are working for you.

Release Schedule

Next Deadline: Docs ready by Nov. 23

We are in Code Freeze, so hopefully you got your PRs in – 45 enhancements made it. Aside from the conference, you should be focused entirely on fixing test failures and finishing up docs. Have those ready to review by Nov. 23, and to merge by (postponed) Dec. 2.

Due to the issue with Github, the v1.18.11 packages were unusable, and had to be replaced by v1.18.12 last Thursday. v1.17.14 and v1.19.4 were also released, on schedule.

Merges & Features

• Cronjobs v2; more about this next week
• Kubelet credential provider plugins in alpha
• CloudProvider-managed resources can have custom tags for doing provider-specific configuration
• Data protection trial feature: new --experimental-logging-sanitization flag
• New Graceful Node Shutdown feature “politely terminates” your pods before shutting down
• The /metrics/resources endpoint for pods, including CPU and Memory metrics, lets you check what your pods are actually consuming
• More New Metrics: stats by Volume plugin and controller, requestAbortsTotal, vSphere version, Node Authorizer load
• Get the ServiceAccount Tokens from the CSI driver
• Let the user prevent automatic allocation of NodePorts
• Add WindowsContainerResources to the CRI API
• Only dump stack traces at log level 4, not 2
• Expose hugepages info in the Downward API
• ServiceTopology feature gate is required to use service.spec.topologyKeys; who knew?
• NodeAffinity gets a shiny new AddedAffinity option
• Scheduler config gets logged at startup, and you can use DefaultPreemption with legacy scheduler policies
• New guidance on InstancesV2 and Zones for cloud providers
• Get preserve-unknown-fields to actually work in kubectl
• Final move of the kubectl bits to staging
• Default Accept-Encoding header removed from HTTP readiness probes
• Requests submitted to the APIServer with invalid timeouts will error HTTP 400
• Add slower/larger buckets to kubelet\_runtime\_operations\_seconds for monitoring slow pulls
• Now that we have APIserver Lease objects, better garbage-collect them
• Fix CVE-2020-8555 for Quobyte Volumes
• Node Local DNS gets a headless service

Promotions

• API Priority and Fairness to Beta
• fsGroup Change Policy to beta
• AppProtocol field to GA
• RuntimeClass to GA
• VolumeSnapshotDataSource to GA
• Pod Resources API to GA

Deprecated

• Make sure the APIserver stops serving deleted Alpha APIs
• EndPointSlice gets a NodeName field which will replace the topology field, the IP Address Type is removed (deprecated v1.17), and discovery.k8s.io/v1alpha1 API is deprecated, all in one PR, whew!
• kubectl service generators are gone

Version Updates

• snapshot-controller to 3.0.2 to fix CVE-2020-8569
• Snapshot APIs to v1

Kubecon Kubernetes Group Sessions

The following sessions are given by SIGs, WGs, or other community groups inside
kubernetes:

• Kubernetes peer group mentoring
• SIG-Multicluster
• SIG-Autoscaling
• Kubernetes Conformance
• SIG-Architecture
• SIG-CLI
• CoC Committee
• SIG-Docs
• SIG-Network
• SIG-Usability
• Release Engineering subproject
• SIG-Cluster-Lifecycle and the Cluster API
• SIG-Instrumentation
• IOT-Edge WG
• Community Marketing subproject
• SIG-Storage
• Naming WG
• Data Protection WG
• SIG-Scheduling