// Run shape inference function for newly added node.
      status->status = desc->graph->refiner.AddNode(ret);
    }
    if (status->status.ok()) {
      // Add the node to the name-to-node mapping.
      desc->graph->name_map[ret->name()] = ret;
    } else if (ret != nullptr) {
      desc->graph->graph.RemoveNode(ret);
      ret = nullptr;
    }
  }

  delete desc;

- Port mapping now allows the same `containerPort` of different containers to different `hostPort` without naming the mapping explicitly. ([#94494](https://github.com/kubernetes/kubernetes/pull/94494), [@SergeyKanzhelev](https://github.com/SergeyKanzhelev))

The Kubernetes in-tree storage plugin to Container Storage Interface (CSI) migration infrastructure is now beta in Kubernetes v1.17. CSI migration was introduced as alpha in Kubernetes v1.14.

## Known Issues
- volumeDevices mapping ignored when container is privileged

of the deprecated flags '--experimental-initial-corrupt-check' and '--experimental-watch-progress-notify-interval' if the etcd version is < 3.6.0. In this version of kubeadm, etcd < 3.6.0 is no longer supported in terms of the k8s / etcd version mapping. These deprecated flags have been replaced by '--feature-gates=InitialCorruptCheck=true' and '--watch-progress-notify-interval'. ([#135701](https://github.com/kubernetes/kubernetes/pull/135701), [@neolit123](https://github.com/neolit123)) [SIG Cluster...

Initially, the bound type parameters are known but the map is empty.
During type unification, if a new type argument <code>A</code> is inferred,
the respective mapping <code>P ➞ A</code> from type parameter to argument
is added to the map.
Conversely, when comparing types, a known type argument
(a type argument for which a map entry already exists)

* `nodeSelector.matchFields` (node's `metadata.node`) is now supported in scheduler. ([#62453](https://github.com/kubernetes/kubernetes/pull/62453), [@k82cn](https://github.com/k82cn))
* Added a parametrizable priority function mapping requested/capacity ratio to priority. This function is disabled by default and can be enabled via the scheduler policy config file.
([#63929](https://github.com/kubernetes/kubernetes/pull/63929), [@losipiuk](https://github.com/losipiuk))

  - `requests.deviceclass.resource.k8s.io/<deviceclass>` is charged based on the worst-case number of devices requested.
  - Device classes mapping to an extended resource now consume `requests.<extended resource name>`. ([#134210](https://github.com/kubernetes/kubernetes/pull/134210), [@yliaog](https://github.com/yliaog)) [SIG API Machinery, Apps, Node, Scheduling and Testing]

* Fixed regression in initial kubectl exec terminal dimensions ([#51127](https://github.com/kubernetes/kubernetes/pull/51127), [@chen-anders](https://github.com/chen-anders))
* Enforcement of fsGroup; enable ScaleIO multiple-instance volume mapping; default PVC capacity; alignment of PVC, PV, and volume names for dynamic provisioning ([#48999](https://github.com/kubernetes/kubernetes/pull/48999), [@vladimirvivien](https://github.com/vladimirvivien))



# v1.7.4

* Enforcement of fsGroup; enable ScaleIO multiple-instance volume mapping; default PVC capacity; alignment of PVC, PV, and volume names for dynamic provisioning ([#48999](https://github.com/kubernetes/kubernetes/pull/48999), [@vladimirvivien](https://github.com/vladimirvivien))

- Fix removing pods from podTopologyHints mapping ([#101892](https://github.com/kubernetes/kubernetes/pull/101892), [@aheng-ch](https://github.com/aheng-ch)) [SIG Node]