The community meeting started with a demo of the
IngressRoute CRD from Heptio Contour.
IngressRoutes allow a more formalized delegation of Ingress-style HTTP routing between multiple, independent teams. Contour also offers a fully-featured Ingress controller, with
IngressRoutes as an additional layer, available if you want it.
Aish Sundar gave an update from the 1.13 release team; beta0 is up and we are on track for code freeze on the 15th. CI signal is currently “yellow”, with the Go 1.11.2 upgrade possibly causing some flakes due to increased test runtime and the HPA tests still being on the fritz. The release notes draft should be going out the SIG leads some time today (the 12th).
Aaron Crickenberger brought us the Contributor Tip of the Week, digging in to the OWNERS file system and ticket approval commands. You find the very long version in a KubeCon EU slide deck, but the short version: OWNERS defines who is in charge of what repos/modules,
/lgtm means the code quality seems reasonable,
/approve means the change is wanted and well implemented.
For SIG updates we had Tim St. Clair for SIG Cluster Lifecycle showed off the progress made on Kubeadm and the ClusterAPI system. Chris Hoge from SIG OpenStack talked about the new out-of-tree OpenStack support and the impending deprecation of the in-tree drivers. Mo Khan from SIG Auth discussed future plans for the ServiceAccount token system, planned security improvements to node labeling, and continued forward motion on dynamic audit configuration.
There will be normal community meetings for the next two weeks (Nov 22 and Dec 6), but not the following three weeks (Dec 13, Dec 20, and Dec 27) due to KubeCon US.
Next Deadline: Code and Test Freeze, November 15 (EOD PST)
Code freeze is imminent so if you have things to merge, do it this week. Please do keep CI Signal in mind though. After code freeze is finalized, 1.13.0-beta1 will be released. Some of the release team is traveling for KubeCon China so responses may be delayed or in unusual timezones.
If you didn’t check it out last week, please do take a look at the new criteria for blocking tests. These will be used in the near future for deciding if things need to be removed from the 1.13 release.
The Kubernetes team has made the decision to fork the
glog library to address long-term issues that have frustrated us. The new library is
k8s.io/klog and is mostly compatible with
glog with the notable exception of requiring a manual call to
klog.InitFlags to set up the command-line flags before parsing. There is an example in the
klog repo of how to integrate
klog into existing
glog-using codebases, or check out this PR for a more complete conversion. Switching to
klog will probably be advisable for Kubernetes ecosystem projects which want to match core component behaviors.
Last week we had the API skeleton for this feature, and now the implementation! As before, this enables custom resources to define a webhook which will be called to upgrade the internal representation of an object between versions. This allows slowly switching off the controllers for older versions of resources and relying on data upgrades instead.
Another part one of a two-parter, this PR adds the APIs for dynamic audit plugins. This allows adding and removing audit webhook receivers without having to restart kube-apiserver or having low-level configuration access (such as in hosted environments).
Last Week In Kubernetes Development (LWKD) is a product of some members of the Kubernetes project, but is not an official publication of the Kubernetes project or the CNCF. 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.