Please also fill out the Kubernetes developer survey.
Next Deadline: Enhancements Freeze, Jan 28th
The 1.18 release cycle has begun! Time for SIGs to figure out what they’re likely to have for 1.18. Important dates:
- Enhancements Freeze: Jan 28
- Code Freeze: March 5
- Docs Due: March 16
- Release: March 24
The following patch releases, covering all supported versions, will be coming out tommorrow (Jan 14):
Version 1.14 is now out of support, so maybe look at upgrading.
For the 1.18 release, the team will be experimenting with sorting changes by Kind instead of by SIG, in response to user requests. Sascha Grunert has shared an example of 1.17’s notes sorted that way. If you have objections or concerns about this, comment on the PR before it merges on Thursday.
Several years ago, we set ginko.flakeAttempts to “2” in many e2e tests, even pushing an upstream change to ginko. This effectively allowed each test two tries to pass, as a hacky way to deal with network timeouts and other issues. The setting also has given us an artificial confidence in how reliable a lot of code is.
- Kube-proxy iptables mode supports dual-stack
- Kubeadm supports auto-retry for image pulls
- Partly revert the change delaying tracking of fields until they are applied
- Set a providerID for a cloud node even if it errored
VolumeBinderfield for the scheduler FrameworkHandler so that it works with volumes
- Azure VMSS clients changed their behavior
kubelet_pleg_last_seen_secondsmetric in case the pod event stream goes away
- Prevent pods from inaccurately remaining NotReady
- Autoset advertise address in fluentd-elasticsearch
- Improve the APIserver’s ability to support SNI across multiple IP addresses
- dns-horizontal nodes need to run on Linux in a mixed cluster
- Remove the following feature gates for now-GA features: PodPriority, TaintNodesByCondition, ScheduleDaemonSetPods, ResourceQuotaScopeSelectors; time to audit your YAML files again
Original Source: http://lwkd.info/2020/20200113