Last Week in Kubernetes Development
Stay up-to-date on Kubernetes development in 15 minutes a week.
Week Ending April 7, 2019
Developer News
Need a lighter, smaller Kubernetes distribution for IoT or embedded use cases? Give K3S a try.
Release Schedule
Next Deadline: 1.15 Release Cycle begins April 8
The new release cycle has begun, and the release team is being staffed up. Release Lead Claire Laurence has set an 11-week target for this release, aiming for final on June 18th in order to avoid clashing with Kubecon Shanghai and the July 4th holiday.
The 1.14.1 patch release came out Monday with an assortment of bug fixes for the new version. Update your servers.
Featured PRs
#74877: add go module support, manage vendor directory using go modules
As a follow up to #75595 last week, more module-y goodness is here. Hopefully this will be faster and more stable! If you commonly manipulate the dependencies, make sure to check out the new hack/pin-dependency.sh script which helps in pinning a specific SHA for a given package. This also means that staging repos will contain a go.mod file as well.
#75953: Expose etcd client latency metrics
This PR re-adds the same etcd latency metrics as we had back in the etcd v2 codepaths. This can be useful both for debugging api-server issues and changes to etcd, as well as general performance monitoring.
test-infra#12038: Automatically push all variants of all test images
Now pushing all of the test images will no longer need to be done manually by Google folks! Previously a lot of the automation for our images required some level of manual action by a Google employee, now the bulk of the work can be handled by our CI system and Google Cloud Build.
website#13543: Init Spanish translation
On the heels of German last week, this week we have Spanish! If you’re interested in helping out with any website or documentation localization, now is a great time to get involved and help make our resources available to even more people.
Other Merges
- Double the throughput of PodAffinity by using atomic operations instead of explicit locking
- New
kubectl rollout restartcommand, does what you’d expect it to kubectl get -wuses Server-Side Printing, allowing it to fetch custom columns from CRDs- Add scheduler metrics for pending pods across the cluster
- All the Webhooks get Port configuration
- kube-proxy won’t auto-cleanup network rules every restart, because bugs and lag, per KEP; use the new
--cleanupswitch if you need it - Stackdriver logging has been disabled on GCE/Windows until it can be debugged
- You can surpress detailed output from the integration tests
- Fix bug in DaemonSetController with failed nodes
- Actually save failed test results from conformance tests
- kubeadm upgrading from 1.12 to 1.13 to 1.14 works now
- New
--image-repositoryflag for Kubeadm supports overriding image sources with a private repo - APIserver switches to standard stacktrace so that it’ll work with debugging tools
- Stop making StatefulSets resync every 30s
- Paginate syncing the Watch Cache with etcd
Deprecated
- The original alpha CSI CRD is removed from 1.15, on schedule
Version Updates
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.