Kubernetes 1.20 Release Cycle

[damemi]

The 1.20 release cycle is now available at https://github.com/kubernetes/sig-release/blob/master/releases/release-1.20/README.md. Here are the important dates:

1. Monday, September 14th: Week 1 - Release cycle begins (you are here)
2. Tuesday, Oct 6th: Week 4 - Enhancements Freeze
3. Thursday, Nov 12th: Week 9 - Code Freeze
4. Monday, Nov 30th: Week 12 - Docs must be completed and reviewed
5. Tuesday, Dec 8th: Week 13 - Kubernetes v1.20.0 released

(From kubernetes-dev list)

So we will target our 1.20 release soon after Tuesday, Dec 8th. Please keep these dates in mind!

[damemi]

We can use this thread to discuss planned features, release notes, etc

[seanmalloy]

Here are my thoughts on what could be worked on for the next descheduler release.

Try To Get These Done

• new pod topology spread constraint strategy Policy for Balancing Pods across topology domains #146 WIP - New Strategy RemovePodsViolatingTopologySpreadConstraint #383
• RemoveDuplocates bug fix RemoveDuplicates should average out based on node count #263 Adding replica check when evicting managed pods from node #398
• Structured logging Add Support For Structured Logging #347
• Add Prometheus metrics Add Support For Prometheus Metrics #348

Nice To Have

• Filter by label Filter by label #195
• Allow evicting critical pods Permit descheduling of critical pods #378 [WIP] Add evict-system-critical-pods option #379
• Node selector enhancement Add a nodeSelector strategy #311

[ingvagabund]

Vendoring k8s.io/kube-scheduler/pkg/framework!!! ]:->

[damemi]

Vendoring k8s.io/kube-scheduler/pkg/framework!!! ]:->

Yeah, this would be a big one. It's still going to depend on opening up the merge queue to the new component-helpers repo in kubernetes/kubernetes#92507. Depending on when that happens in the release cycle we may not have time to finish migrating the framework and add it here.

I'd like to put more pressure on that, because I think it does fit in with the "stability" goals of 1.20 (reducing duplicate/internal helpers)

[damemi]

Today is the planned upstream kubernetes 1.20 release date 🎉 Meaning we should be able to merge a final 1.20 rebase today or tomorrow.

I think it would be best to cut our v0.20 release by the end of the week. With holidays coming up, many people (including myself) will be taking time off over the next couple weeks. @seanmalloy @ingvagabund is there anything out of the above goals that we should hold the release on? All that's missing is the removeduplicates fix and prometheus metrics. There's also a couple PRs open. What do you think?

[seanmalloy]

@seanmalloy @ingvagabund is there anything out of the above goals that we should hold the release on?

@damemi I'm not aware of anything that would hold up the release of descheduler v0.20.0 by Friday 12/11. I'll try to submit a PR to update/add all of the references(i.e. README) required for the v0.20.0 release.

I think we just need to volunteer to put the release notes together.

[seanmalloy]

I think this is all that is left. This list is not in any particular order.

• Merge Update Docs and Manifests for v0.20.0 #462
• Merge Delete everything but index.yaml from gh-pages #460
• Merge Update descheduler CI for k8s 1.20 kubernetes/test-infra#20166
• Bump the deps in go.mod to the GA v1.20 tags
• Create release notes
• Execute release process

[damemi]

@seanmalloy @ingvagabund here's what I put together for release notes:

Container image for this release is available at: `k8s.gcr.io/descheduler/descheduler:v0.20.0`

The Helm chart for this release is published [here](https://github.com/kubernetes-sigs/descheduler/releases/tag/descheduler-helm-chart-0.20.0).

## New Features :rainbow:
* #393 PodLifeTime: allow custom podStatusPhases @lixiang233 
* #406 Add namespace filtering to RemoveDuplicates @damemi
* #412 Add logging-format flag @farah 
* #421 Support for remote resources in kustomize @m3y 
* #418 charts/descheduler: add PodSecurityPolicy support @invidian 
* #413 Add PodTopologySpread strategy @damemi
* #436 Rename helm chart @stevehipwell 
* #444 Update helm chart to allow setting startingDeadlineSeconds @seanmalloy 
* #464 bump k8s dependencies to 1.20.0 @damemi 
* #449 Enable multi-arch container image builds @seanmalloy 
* #453 allow setting options successfulJobsHistoryLimit and failedJobsHistoryLimit for cronjob @dieterdemeyer 

## Bug Fixes :bug:
* #395 Move IsEvictable check in PodAntiAffinity
* #427 Fix version output in automated container builds @seanmalloy 
* #443 Set container resources in helm chart @seanmalloy 
* #442 Set container resources in YAML manifests @seanmalloy 
* #454 Fix broken namespace logic in TopologySpreadConstraints @damemi 
* #455 Helm: allow the topologySpreadConstraint strategy to list namespaces @AmoVanB 

## Others :running:
* #376 Change klog to use structured logging @farah 
* #392 Add link to Helm hub @seanmalloy 
* #394 Convert logs to use structured logging @farah 
* #384 Refactor low node utilization @ingvagabund 
* #401 Update e2e tests to use kind v0.9.0
* #407 Flip Info/Infof/Error to InfoS/ErrorS @ingvagabund 
* #410 Add use case for lowNodeUtilization @lixiang233 
* #411 Convert last log message to structured logging @seanmalloy 
* #414 Update event logging to use structured logging @seanmalloy 
* #416 Bump to go 1.15.2 @seanmalloy 
* #417 Helm chart readme fix @plutzilla 
* #429 fix typo @invidian 
* #430 Add hack/verify-spelling.sh script @damemi 
* #428 Start using helpers from k8s.io/component-helpers @damemi
* #438 Add missing parameter in README @lixiang233 
* #441 LowNodeUtilization:show node usage in percentage as well @ingvagabund 
* #433 Fix trivial typo in helm README @ankon 
* #446 LowNodeUtilization: express usagePercentage multiplied by 100
* #448 Add more topology spread logs @damemi 
* #452 bump to go 1.15.5 @seanmalloy 
* #462 Update docs and manifests for v0.20.0 @seanmalloy 
* #465 Skip topology calculations if domains are already balanced @damemi 

We could also merge #463 if you approve @seanmalloy. Then we'll start the image promotion PR and I'll follow up with the branch/tag/release process

[seanmalloy]

@damemi #463 just merged. Also, the release nodes look good. I believe. you can go ahead and cut the release. Let me know if you need any assistance. Thanks!

[damemi]

Thanks @seanmalloy, I've pushed the release branch and the helm chart-releaser action has run successfully.

All that's left is image promotion and publishing the release. Can you open the image-promotion PR?

[damemi]

@seanmalloy I went ahead and created the image promotion PR here, ptal: kubernetes/k8s.io#1477

Want to make sure I got the multi-arch part correct

[damemi]

The release is published! https://github.com/kubernetes-sigs/descheduler/releases/tag/v0.20.0

I will send out the announcement email later today. Thank you to everyone who contributed!
/close

[k8s-ci-robot]

@damemi: Closing this issue.

In response to this:

The release is published! https://github.com/kubernetes-sigs/descheduler/releases/tag/v0.20.0

I will send out the announcement email later today. Thank you to everyone who contributed!
/close

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.