a.length-e.length}for(var a,t,s,n,r=[],d=[],i=[],_=[],o=0;o<7;o++)a=c([2e3,1]).day(o),t=he(this.weekdaysMin(a,"")),s=he(this.weekdaysShort(a,"")),n=he(this.weekdays(a,"")),r.push(t),d.push(s),i.push(n),_.push(t),_.push(s),_.push(n);r.sort(e),d.sort(e),i.sort(e),_.sort(e),this._weekdaysRegex=new RegExp("^("+_.join("|")+")","i"),this._weekdaysShortRegex=this._weekdaysRegex,this._weekdaysMinRegex=this._weekdaysRegex,this._weekdaysStrictRegex=new RegExp("^("+i.join("|")+")","i"),this._weekdaysShortStrictRegex=new...

- Fixes a problem with 63-second or 1-second connection delays with some VXLAN-based
  network plugins which was first widely noticed in 1.16 (though some users saw it
  earlier than that, possibly only with specific network plugins). If you were previously
  using ethtool to disable checksum offload on your primary network interface, you should

*   deviceplugin has been extended to more gracefully handle the full device plugin lifecycle, including: ([#55088](https://github.com/kubernetes/kubernetes/pull/55088),[ @jiayingz](https://github.com/jiayingz))
    *   Kubelet now uses an explicit cm.GetDevicePluginResourceCapacity() function that makes it possible to more accurately determine what resources are inactive and return a more accurate view of available resources.

  - [Changelog since v1.18.17](#changelog-since-v11817)
  - [Important Security Information](#important-security-information)
    - [CVE-2021-25735: Validating Admission Webhook does not observe some previous fields](#cve-2021-25735-validating-admission-webhook-does-not-observe-some-previous-fields)
  - [Changes by Kind](#changes-by-kind-2)
    - [API Change](#api-change-1)
    - [Feature](#feature)
    - [Bug or Regression](#bug-or-regression-2)

#### Extensibility

- The community has been asking for continuing support of extensibility, so this cycle features more work around CRDs and API Machinery. Most of the enhancements in this cycle were from SIG API Machinery and related areas.

### Extensibility around core Kubernetes APIs

- Fixed a situation when, sometimes, the scheduler incorrectly placed a pod in the `unschedulable` queue instead of the `backoff` queue. This happened when some plugin previously declared the pod as `unschedulable` and then in a later attempt encounters some other error. Scheduling of that pod then got delayed by up to five minutes, after which periodic flushing moved the pod back into the `active` queue. ([#120334](https://github.com/kubernetes/kubernetes/pull/120334),...

* If Openstack LoadBalancer is not defined in cloud config, the loadbalancer is not initialized any more in openstack. All setups must have some setting under that section ([#65781](https://github.com/kubernetes/kubernetes/pull/65781), [@zetaab](https://github.com/zetaab))

For more information, see [#53288](https://issue.k8s.io/53288).

    - [**Cluster Federation**](#cluster-federation)
      - [Placement Policy](#placement-policy)
      - [Cluster Selection](#cluster-selection)
    - [**Instrumentation**](#instrumentation)
      - [Core Metrics API](#core-metrics-api)
    - [**Internationalization**](#internationalization)
    - [**kubectl (CLI)**](#kubectl-cli-1)
    - [**Networking**](#networking)
      - [Network Policy](#network-policy)

- The Scheduler did not run the plugin Score method when its PreScore method returned a Skip status. In other words, the PreScore/Score plugin could return a Skip status in PreScore if the plugin did nothing in Score for that Pod. ([#115652](https://github.com/kubernetes/kubernetes/pull/115652), [@AxeZhan](https://github.com/AxeZhan))