add default for ProcMountType

[jessfraz]

What this PR does / why we need it: This adds a default for the ProcMount feature and fixes #69647

kube-apiserver: fixes `procMount` field incorrectly being marked as required in openapi schema

[jessfraz]

we should also cherry-pick this on a release

[dims]

/assign @liggitt

Jordan, Fix for a backwards compatibility problem. Old yamls fail against 1.12.x

[jessfraz]

/priority critical-urgent

[jessfraz]

/kind bug

[liggitt]

The new field must be made optional/omitempty in types.go… setting a server side default won't fix the client side attempt to require the field in validation for old clients

[liggitt]

I didn't expect checking feature gates in defaulting… is this feature enabled by default?

[MarcPow]

Hmmm. I don't understand this either. I'm the original reporter (and a bit of a newb), but I don't believe this gate is enabled.

azureuser@k8s-master-71973844-0:~$ ps aux | grep feature-gate
root 9035 3.7 1.9 671344 155776 ? Ssl Oct10 54:10 /usr/local/bin/kubelet --enable-server --node-labels=kubernetes.io/role=master,kubernetes.azure.com/cluster=ClarityOrDie18RG --v=2 --volume-plugin-dir=/etc/kubernetes/volumeplugins --address=0.0.0.0 --allow-privileged=true --anonymous-auth=false --authorization-mode=Webhook --azure-container-registry-config=/etc/kubernetes/azure.json --cgroups-per-qos=true --client-ca-file=/etc/kubernetes/certs/ca.crt --cloud-config=/etc/kubernetes/azure.json --cloud-provider=azure --cluster-dns=10.0.0.10 --cluster-domain=cluster.local --enforce-node-allocatable=pods --event-qps=0 --eviction-hard=memory.available<100Mi,nodefs.available<10%,nodefs.inodesFree<5% --feature-gates=PodPriority=true --image-gc-high-threshold=85 --image-gc-low-threshold=80 --image-pull-progress-deadline=30m --keep-terminated-pod-volumes=false --kubeconfig=/var/lib/kubelet/kubeconfig --max-pods=30 --network-plugin=cni --node-status-update-frequency=10s --non-masquerade-cidr=10.240.0.0/12 --pod-infra-container-image=k8s.gcr.io/pause-amd64:3.1 --pod-manifest-path=/etc/kubernetes/manifests --pod-max-pids=100
root 11244 0.1 1.1 408860 92912 ? Ssl Oct10 2:49 /hyperkube proxy --kubeconfig=/var/lib/kubelet/kubeconfig --cluster-cidr=10.240.0.0/12 --feature-gates=ExperimentalCriticalPodAnnotation=true
root 18830 2.8 1.7 587656 145652 ? Ssl Oct11 6:29 /hyperkube controller-manager --allocate-node-cidrs=false --cloud-config=/etc/kubernetes/azure.json --cloud-provider=azure --cluster-cidr=10.240.0.0/12 --cluster-name=ClarityOrDie18 --cluster-signing-cert-file=/etc/kubernetes/certs/ca.crt --cluster-signing-key-file=/etc/kubernetes/certs/ca.key --configure-cloud-routes=false --feature-gates=ServiceNodeExclusion=true --kubeconfig=/var/lib/kubelet/kubeconfig --leader-elect=true --node-monitor-grace-period=40s --pod-eviction-timeout=5m0s --profiling=false --root-ca-file=/etc/kubernetes/certs/ca.crt --route-reconciliation-period=10s --service-account-private-key-file=/etc/kubernetes/certs/apiserver.key --terminated-pod-gc-threshold=5000 --use-service-account-credentials=true --v=2

[jessfraz]

it is not enabled by default, but there was already the feature gate check for a volume feature in that same file so I figured it was normal, if not i can remove

[jessfraz]

The new field must be made optional/omitempty in types.go… setting a server side default won't fix the client side attempt to require the field in validation for old clients

AFAICT it is set as optional am i missing something https://github.com/kubernetes/kubernetes/blob/master/pkg/apis/core/types.go#L4634

[liggitt]

kubernetes/staging/src/k8s.io/api/core/v1/types.go

Lines 5217 to 5222 in b5a5010

	// procMount denotes the type of proc mount to use for the containers.
	// The default is DefaultProcMount which uses the container runtime defaults for
	// readonly paths and masked paths.
	// This requires the ProcMountType feature flag to be enabled.
	// +optional
	ProcMount *ProcMountType `json:"procMount,omitEmpty" protobuf:"bytes,9,opt,name=procMount"`

s/omitEmpty/omitempty/, I think. It still shows up as required in the openapi spec

[liggitt]

/lgtm
/approve

can you cherry-pick this to release-1.12 once tests are green?

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: jessfraz, liggitt

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• api/OWNERS [liggitt]
• docs/api-reference/OWNERS [liggitt]
• staging/src/k8s.io/api/OWNERS [liggitt]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[jessfraz]

will do!

…

On Fri, Oct 12, 2018 at 9:32 AM Jordan Liggitt ***@***.***> wrote: /lgtm /approve can you cherry-pick this to release-1.12 once tests are green? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#69694 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABYNbJvhoZZWQlI_s7TjbI-D9rzMlLVqks5ukMQNgaJpZM4XYObz> .

-- Jessie Frazelle 4096R / D4C4 DD60 0D66 F65A 8EFC 511E 18F3 685C 0022 BFF3 pgp.mit.edu <http://pgp.mit.edu/pks/lookup?op=get&search=0x18F3685C0022BFF3>

[jessfraz]

/test pull-kubernetes-kubemark-e2e-gce-big

[liggitt]

added linting test for this in kubernetes/kube-openapi#106