Kubernetes’ newly elected Steering Committee members are Davanum Srinivas, Jordan Liggitt, and Bob Killen. Thanks to all 312 contributors who voted.
The October Community Meeting is this week, featuring SIG-Instrumentation, Storage, and Service-Catalog.
The Reliability Working Group is tentatively approved, but needs work on their charter. If you care about reliability, find them on the #wg-reliablity Slack channel or kubernetes-wg-reliability group.
Next Deadline: Beta-0 released, Oct 27
Enhancements Freeze is now in effect, so if you’re still working on a feature to add, you have until Nov. 2 to file an exception and see if you can still get it accepted. Sooner is better.
Patch releases for 1.17, 1.18, and 1.19 are expected out this week; see below for some changes.
Patch Release Merges
These fixes have been backported to supported versions for this week’s patch releases.
- Don’t leak Docker secrets into the log
- Fix request duration metrics to capture latency
- Record kube-proxy network latency accurately
- Don’t crash the schedule cache when removing nodes messily
- Prevent pod-handling race condition in 1.19
- Dual-stack Nodes are now a thing, and pods on them inherit dual interfaces; fix how we handle dual-stack IPtables rules while we’re at it
- Add new
k8s.io/component-helpersrepository as the destination for shared scheduler code that’s being pushed out due to converting all schedulers to plugins
kubectl create ingresscommand
- Stop using systemd mount, and instead use core Linux mount, when mounting most storage types
- Let scheduler plugins call preemptHandle with embedded scoring
- Azure disks can be bigger than 4TB and won’t lose your data on unmount
- Cloud node controller copes with having a blank ProviderID
- Don’t log credentials even at Log Level 9
Original Source: http://lwkd.info/2020/20201013