From 1ed554c2e6dc3b0ade296f39a84c7992cbf73f09 Mon Sep 17 00:00:00 2001 From: niam0522 Date: Tue, 26 Nov 2024 12:40:46 +0530 Subject: [PATCH] add Calico v3.29.1 and remove k8s v1.26 --- kubemarine/patches/software_upgrade.yaml | 12 +- .../calico-apiserver-v3.29.1-original.yaml | 298 + .../plugins/yaml/calico-v3.29.1-original.yaml | 6631 +++++++++++++++++ .../internal/kubernetes_images.yaml | 56 - .../compatibility/internal/packages.yaml | 24 - .../compatibility/internal/plugins.yaml | 52 +- .../compatibility/internal/thirdparties.yaml | 64 +- .../compatibility/kubernetes_versions.yaml | 34 +- 8 files changed, 6955 insertions(+), 216 deletions(-) create mode 100644 kubemarine/plugins/yaml/calico-apiserver-v3.29.1-original.yaml create mode 100644 kubemarine/plugins/yaml/calico-v3.29.1-original.yaml diff --git a/kubemarine/patches/software_upgrade.yaml b/kubemarine/patches/software_upgrade.yaml index 87ec8fbbf..f4b6236f6 100644 --- a/kubemarine/patches/software_upgrade.yaml +++ b/kubemarine/patches/software_upgrade.yaml @@ -8,7 +8,11 @@ # The order of upgrade of defined by the implementation. thirdparties: - calicoctl: [] + calicoctl: + - v1.28.12 + - v1.29.7 + - v1.30.3 + - v1.31.1 crictl: [] packages: containerd: @@ -28,7 +32,11 @@ packages: version_rhel9: false version_debian: false plugins: - calico: [] + calico: + - v1.28.12 + - v1.29.7 + - v1.30.3 + - v1.31.1 nginx-ingress-controller: [] kubernetes-dashboard: [] local-path-provisioner: [] diff --git a/kubemarine/plugins/yaml/calico-apiserver-v3.29.1-original.yaml b/kubemarine/plugins/yaml/calico-apiserver-v3.29.1-original.yaml new file mode 100644 index 000000000..a4a1f0d19 --- /dev/null +++ b/kubemarine/plugins/yaml/calico-apiserver-v3.29.1-original.yaml @@ -0,0 +1,298 @@ +# This is a tech-preview manifest which installs the Calico API server. Note that this manifest is liable to change +# or be removed in future releases without further warning. +# +# Namespace and namespace-scoped resources. +apiVersion: v1 +kind: Namespace +metadata: + labels: + name: calico-apiserver + name: calico-apiserver +spec: + +--- + +# Policy to ensure the API server isn't cut off. Can be modified, but ensure +# that the main API server is always able to reach the Calico API server. +kind: NetworkPolicy +apiVersion: networking.k8s.io/v1 +metadata: + name: allow-apiserver + namespace: calico-apiserver +spec: + podSelector: + matchLabels: + apiserver: "true" + ingress: + - ports: + - protocol: TCP + port: 5443 + +--- + +apiVersion: v1 +kind: Service +metadata: + name: calico-api + namespace: calico-apiserver +spec: + ports: + - name: apiserver + port: 443 + protocol: TCP + targetPort: 5443 + selector: + apiserver: "true" + type: ClusterIP + +--- + +apiVersion: apps/v1 +kind: Deployment +metadata: + labels: + apiserver: "true" + k8s-app: calico-apiserver + name: calico-apiserver + namespace: calico-apiserver +spec: + replicas: 1 + selector: + matchLabels: + apiserver: "true" + strategy: + type: Recreate + template: + metadata: + labels: + apiserver: "true" + k8s-app: calico-apiserver + name: calico-apiserver + namespace: calico-apiserver + spec: + containers: + - args: + - --secure-port=5443 + - -v=5 + env: + - name: DATASTORE_TYPE + value: kubernetes + image: calico/apiserver:v3.29.1 + name: calico-apiserver + readinessProbe: + httpGet: + path: /readyz + port: 5443 + scheme: HTTPS + timeoutSeconds: 5 + periodSeconds: 60 + securityContext: + privileged: false + runAsUser: 0 + volumeMounts: + - mountPath: /code/apiserver.local.config/certificates + name: calico-apiserver-certs + dnsPolicy: ClusterFirst + nodeSelector: + kubernetes.io/os: linux + restartPolicy: Always + serviceAccount: calico-apiserver + serviceAccountName: calico-apiserver + tolerations: + - effect: NoSchedule + key: node-role.kubernetes.io/master + - effect: NoSchedule + key: node-role.kubernetes.io/control-plane + volumes: + - name: calico-apiserver-certs + secret: + secretName: calico-apiserver-certs + +--- + +apiVersion: v1 +kind: ServiceAccount +metadata: + name: calico-apiserver + namespace: calico-apiserver + +--- + +# Cluster-scoped resources below here. +apiVersion: apiregistration.k8s.io/v1 +kind: APIService +metadata: + name: v3.projectcalico.org +spec: + group: projectcalico.org + groupPriorityMinimum: 1500 + service: + name: calico-api + namespace: calico-apiserver + port: 443 + version: v3 + versionPriority: 200 + +--- + +apiVersion: rbac.authorization.k8s.io/v1 +kind: ClusterRole +metadata: + name: calico-crds +rules: +- apiGroups: + - extensions + - networking.k8s.io + - "" + resources: + - networkpolicies + - nodes + - namespaces + - pods + - serviceaccounts + verbs: + - get + - list + - watch +- apiGroups: + - crd.projectcalico.org + resources: + - globalnetworkpolicies + - networkpolicies + - clusterinformations + - hostendpoints + - globalnetworksets + - networksets + - bgpconfigurations + - bgppeers + - bgpfilters + - felixconfigurations + - kubecontrollersconfigurations + - ippools + - ipreservations + - ipamblocks + - blockaffinities + - caliconodestatuses + - ipamconfigs + - tiers + verbs: + - get + - list + - watch + - create + - update + - delete +- apiGroups: + - policy + resourceNames: + - calico-apiserver + resources: + - podsecuritypolicies + verbs: + - use + +--- + +apiVersion: rbac.authorization.k8s.io/v1 +kind: ClusterRole +metadata: + name: calico-extension-apiserver-auth-access +rules: +- apiGroups: + - "" + resourceNames: + - extension-apiserver-authentication + resources: + - configmaps + verbs: + - list + - watch + - get +- apiGroups: + - rbac.authorization.k8s.io + resources: + - clusterroles + - clusterrolebindings + - roles + - rolebindings + verbs: + - get + - list + - watch + +--- + +apiVersion: rbac.authorization.k8s.io/v1 +kind: ClusterRole +metadata: + name: calico-webhook-reader +rules: +- apiGroups: + - admissionregistration.k8s.io + resources: + - mutatingwebhookconfigurations + - validatingwebhookconfigurations + verbs: + - get + - list + - watch + +--- + +apiVersion: rbac.authorization.k8s.io/v1 +kind: ClusterRoleBinding +metadata: + name: calico-apiserver-access-crds +roleRef: + apiGroup: rbac.authorization.k8s.io + kind: ClusterRole + name: calico-crds +subjects: +- kind: ServiceAccount + name: calico-apiserver + namespace: calico-apiserver + +--- + +apiVersion: rbac.authorization.k8s.io/v1 +kind: ClusterRoleBinding +metadata: + name: calico-apiserver-delegate-auth +roleRef: + apiGroup: rbac.authorization.k8s.io + kind: ClusterRole + name: system:auth-delegator +subjects: +- kind: ServiceAccount + name: calico-apiserver + namespace: calico-apiserver + +--- + +apiVersion: rbac.authorization.k8s.io/v1 +kind: ClusterRoleBinding +metadata: + name: calico-apiserver-webhook-reader +roleRef: + apiGroup: rbac.authorization.k8s.io + kind: ClusterRole + name: calico-webhook-reader +subjects: +- kind: ServiceAccount + name: calico-apiserver + namespace: calico-apiserver + +--- + +apiVersion: rbac.authorization.k8s.io/v1 +kind: ClusterRoleBinding +metadata: + name: calico-extension-apiserver-auth-access +roleRef: + apiGroup: rbac.authorization.k8s.io + kind: ClusterRole + name: calico-extension-apiserver-auth-access +subjects: +- kind: ServiceAccount + name: calico-apiserver + namespace: calico-apiserver diff --git a/kubemarine/plugins/yaml/calico-v3.29.1-original.yaml b/kubemarine/plugins/yaml/calico-v3.29.1-original.yaml new file mode 100644 index 000000000..b0dcf73f1 --- /dev/null +++ b/kubemarine/plugins/yaml/calico-v3.29.1-original.yaml @@ -0,0 +1,6631 @@ +--- +# Source: calico/templates/calico-kube-controllers.yaml +# This manifest creates a Pod Disruption Budget for Controller to allow K8s Cluster Autoscaler to evict + +apiVersion: policy/v1 +kind: PodDisruptionBudget +metadata: + name: calico-kube-controllers + namespace: kube-system + labels: + k8s-app: calico-kube-controllers +spec: + maxUnavailable: 1 + selector: + matchLabels: + k8s-app: calico-kube-controllers +--- +# Source: calico/templates/calico-typha.yaml +# This manifest creates a Pod Disruption Budget for Typha to allow K8s Cluster Autoscaler to evict + +apiVersion: policy/v1 +kind: PodDisruptionBudget +metadata: + name: calico-typha + namespace: kube-system + labels: + k8s-app: calico-typha +spec: + maxUnavailable: 1 + selector: + matchLabels: + k8s-app: calico-typha +--- +# Source: calico/templates/calico-kube-controllers.yaml +apiVersion: v1 +kind: ServiceAccount +metadata: + name: calico-kube-controllers + namespace: kube-system +--- +# Source: calico/templates/calico-node.yaml +apiVersion: v1 +kind: ServiceAccount +metadata: + name: calico-node + namespace: kube-system +--- +# Source: calico/templates/calico-node.yaml +apiVersion: v1 +kind: ServiceAccount +metadata: + name: calico-cni-plugin + namespace: kube-system +--- +# Source: calico/templates/calico-config.yaml +# This ConfigMap is used to configure a self-hosted Calico installation. +kind: ConfigMap +apiVersion: v1 +metadata: + name: calico-config + namespace: kube-system +data: + # You must set a non-zero value for Typha replicas below. + typha_service_name: "calico-typha" + # Configure the backend to use. + calico_backend: "bird" + + # Configure the MTU to use for workload interfaces and tunnels. + # By default, MTU is auto-detected, and explicitly setting this field should not be required. + # You can override auto-detection by providing a non-zero value. + veth_mtu: "0" + + # The CNI network configuration to install on each node. The special + # values in this config will be automatically populated. + cni_network_config: |- + { + "name": "k8s-pod-network", + "cniVersion": "0.3.1", + "plugins": [ + { + "type": "calico", + "log_level": "info", + "log_file_path": "/var/log/calico/cni/cni.log", + "datastore_type": "kubernetes", + "nodename": "__KUBERNETES_NODE_NAME__", + "mtu": __CNI_MTU__, + "ipam": { + "type": "calico-ipam" + }, + "policy": { + "type": "k8s" + }, + "kubernetes": { + "kubeconfig": "__KUBECONFIG_FILEPATH__" + } + }, + { + "type": "portmap", + "snat": true, + "capabilities": {"portMappings": true} + }, + { + "type": "bandwidth", + "capabilities": {"bandwidth": true} + } + ] + } +--- +# Source: calico/templates/kdd-crds.yaml +apiVersion: apiextensions.k8s.io/v1 +kind: CustomResourceDefinition +metadata: + name: bgpconfigurations.crd.projectcalico.org +spec: + group: crd.projectcalico.org + names: + kind: BGPConfiguration + listKind: BGPConfigurationList + plural: bgpconfigurations + singular: bgpconfiguration + preserveUnknownFields: false + scope: Cluster + versions: + - name: v1 + schema: + openAPIV3Schema: + description: BGPConfiguration contains the configuration for any BGP routing. + properties: + apiVersion: + description: 'APIVersion defines the versioned schema of this representation + of an object. Servers should convert recognized schemas to the latest + internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources' + type: string + kind: + description: 'Kind is a string value representing the REST resource this + object represents. Servers may infer this from the endpoint the client + submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds' + type: string + metadata: + type: object + spec: + description: BGPConfigurationSpec contains the values of the BGP configuration. + properties: + asNumber: + description: 'ASNumber is the default AS number used by a node. [Default: + 64512]' + format: int32 + type: integer + bindMode: + description: BindMode indicates whether to listen for BGP connections + on all addresses (None) or only on the node's canonical IP address + Node.Spec.BGP.IPvXAddress (NodeIP). Default behaviour is to listen + for BGP connections on all addresses. + type: string + communities: + description: Communities is a list of BGP community values and their + arbitrary names for tagging routes. + items: + description: Community contains standard or large community value + and its name. + properties: + name: + description: Name given to community value. + type: string + value: + description: Value must be of format `aa:nn` or `aa:nn:mm`. + For standard community use `aa:nn` format, where `aa` and + `nn` are 16 bit number. For large community use `aa:nn:mm` + format, where `aa`, `nn` and `mm` are 32 bit number. Where, + `aa` is an AS Number, `nn` and `mm` are per-AS identifier. + pattern: ^(\d+):(\d+)$|^(\d+):(\d+):(\d+)$ + type: string + type: object + type: array + ignoredInterfaces: + description: IgnoredInterfaces indicates the network interfaces that + needs to be excluded when reading device routes. + items: + type: string + type: array + listenPort: + description: ListenPort is the port where BGP protocol should listen. + Defaults to 179 + maximum: 65535 + minimum: 1 + type: integer + logSeverityScreen: + description: 'LogSeverityScreen is the log severity above which logs + are sent to the stdout. [Default: INFO]' + type: string + nodeMeshMaxRestartTime: + description: Time to allow for software restart for node-to-mesh peerings. When + specified, this is configured as the graceful restart timeout. When + not specified, the BIRD default of 120s is used. This field can + only be set on the default BGPConfiguration instance and requires + that NodeMesh is enabled + type: string + nodeMeshPassword: + description: Optional BGP password for full node-to-mesh peerings. + This field can only be set on the default BGPConfiguration instance + and requires that NodeMesh is enabled + properties: + secretKeyRef: + description: Selects a key of a secret in the node pod's namespace. + properties: + key: + description: The key of the secret to select from. Must be + a valid secret key. + type: string + name: + description: 'Name of the referent. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names + TODO: Add other useful fields. apiVersion, kind, uid?' + type: string + optional: + description: Specify whether the Secret or its key must be + defined + type: boolean + required: + - key + type: object + type: object + nodeToNodeMeshEnabled: + description: 'NodeToNodeMeshEnabled sets whether full node to node + BGP mesh is enabled. [Default: true]' + type: boolean + prefixAdvertisements: + description: PrefixAdvertisements contains per-prefix advertisement + configuration. + items: + description: PrefixAdvertisement configures advertisement properties + for the specified CIDR. + properties: + cidr: + description: CIDR for which properties should be advertised. + type: string + communities: + description: Communities can be list of either community names + already defined in `Specs.Communities` or community value + of format `aa:nn` or `aa:nn:mm`. For standard community use + `aa:nn` format, where `aa` and `nn` are 16 bit number. For + large community use `aa:nn:mm` format, where `aa`, `nn` and + `mm` are 32 bit number. Where,`aa` is an AS Number, `nn` and + `mm` are per-AS identifier. + items: + type: string + type: array + type: object + type: array + serviceClusterIPs: + description: ServiceClusterIPs are the CIDR blocks from which service + cluster IPs are allocated. If specified, Calico will advertise these + blocks, as well as any cluster IPs within them. + items: + description: ServiceClusterIPBlock represents a single allowed ClusterIP + CIDR block. + properties: + cidr: + type: string + type: object + type: array + serviceExternalIPs: + description: ServiceExternalIPs are the CIDR blocks for Kubernetes + Service External IPs. Kubernetes Service ExternalIPs will only be + advertised if they are within one of these blocks. + items: + description: ServiceExternalIPBlock represents a single allowed + External IP CIDR block. + properties: + cidr: + type: string + type: object + type: array + serviceLoadBalancerIPs: + description: ServiceLoadBalancerIPs are the CIDR blocks for Kubernetes + Service LoadBalancer IPs. Kubernetes Service status.LoadBalancer.Ingress + IPs will only be advertised if they are within one of these blocks. + items: + description: ServiceLoadBalancerIPBlock represents a single allowed + LoadBalancer IP CIDR block. + properties: + cidr: + type: string + type: object + type: array + type: object + type: object + served: true + storage: true +status: + acceptedNames: + kind: "" + plural: "" + conditions: [] + storedVersions: [] +--- +# Source: calico/templates/kdd-crds.yaml +apiVersion: apiextensions.k8s.io/v1 +kind: CustomResourceDefinition +metadata: + annotations: + controller-gen.kubebuilder.io/version: (devel) + creationTimestamp: null + name: bgpfilters.crd.projectcalico.org +spec: + group: crd.projectcalico.org + names: + kind: BGPFilter + listKind: BGPFilterList + plural: bgpfilters + singular: bgpfilter + scope: Cluster + versions: + - name: v1 + schema: + openAPIV3Schema: + properties: + apiVersion: + description: 'APIVersion defines the versioned schema of this representation + of an object. Servers should convert recognized schemas to the latest + internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources' + type: string + kind: + description: 'Kind is a string value representing the REST resource this + object represents. Servers may infer this from the endpoint the client + submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds' + type: string + metadata: + type: object + spec: + description: BGPFilterSpec contains the IPv4 and IPv6 filter rules of + the BGP Filter. + properties: + exportV4: + description: The ordered set of IPv4 BGPFilter rules acting on exporting + routes to a peer. + items: + description: BGPFilterRuleV4 defines a BGP filter rule consisting + a single IPv4 CIDR block and a filter action for this CIDR. + properties: + action: + type: string + cidr: + type: string + interface: + type: string + matchOperator: + type: string + prefixLength: + properties: + max: + format: int32 + maximum: 32 + minimum: 0 + type: integer + min: + format: int32 + maximum: 32 + minimum: 0 + type: integer + type: object + source: + type: string + required: + - action + type: object + type: array + exportV6: + description: The ordered set of IPv6 BGPFilter rules acting on exporting + routes to a peer. + items: + description: BGPFilterRuleV6 defines a BGP filter rule consisting + a single IPv6 CIDR block and a filter action for this CIDR. + properties: + action: + type: string + cidr: + type: string + interface: + type: string + matchOperator: + type: string + prefixLength: + properties: + max: + format: int32 + maximum: 128 + minimum: 0 + type: integer + min: + format: int32 + maximum: 128 + minimum: 0 + type: integer + type: object + source: + type: string + required: + - action + type: object + type: array + importV4: + description: The ordered set of IPv4 BGPFilter rules acting on importing + routes from a peer. + items: + description: BGPFilterRuleV4 defines a BGP filter rule consisting + a single IPv4 CIDR block and a filter action for this CIDR. + properties: + action: + type: string + cidr: + type: string + interface: + type: string + matchOperator: + type: string + prefixLength: + properties: + max: + format: int32 + maximum: 32 + minimum: 0 + type: integer + min: + format: int32 + maximum: 32 + minimum: 0 + type: integer + type: object + source: + type: string + required: + - action + type: object + type: array + importV6: + description: The ordered set of IPv6 BGPFilter rules acting on importing + routes from a peer. + items: + description: BGPFilterRuleV6 defines a BGP filter rule consisting + a single IPv6 CIDR block and a filter action for this CIDR. + properties: + action: + type: string + cidr: + type: string + interface: + type: string + matchOperator: + type: string + prefixLength: + properties: + max: + format: int32 + maximum: 128 + minimum: 0 + type: integer + min: + format: int32 + maximum: 128 + minimum: 0 + type: integer + type: object + source: + type: string + required: + - action + type: object + type: array + type: object + type: object + served: true + storage: true +status: + acceptedNames: + kind: "" + plural: "" + conditions: [] + storedVersions: [] +--- +# Source: calico/templates/kdd-crds.yaml +apiVersion: apiextensions.k8s.io/v1 +kind: CustomResourceDefinition +metadata: + name: bgppeers.crd.projectcalico.org +spec: + group: crd.projectcalico.org + names: + kind: BGPPeer + listKind: BGPPeerList + plural: bgppeers + singular: bgppeer + preserveUnknownFields: false + scope: Cluster + versions: + - name: v1 + schema: + openAPIV3Schema: + properties: + apiVersion: + description: 'APIVersion defines the versioned schema of this representation + of an object. Servers should convert recognized schemas to the latest + internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources' + type: string + kind: + description: 'Kind is a string value representing the REST resource this + object represents. Servers may infer this from the endpoint the client + submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds' + type: string + metadata: + type: object + spec: + description: BGPPeerSpec contains the specification for a BGPPeer resource. + properties: + asNumber: + description: The AS Number of the peer. + format: int32 + type: integer + filters: + description: The ordered set of BGPFilters applied on this BGP peer. + items: + type: string + type: array + keepOriginalNextHop: + description: Option to keep the original nexthop field when routes + are sent to a BGP Peer. Setting "true" configures the selected BGP + Peers node to use the "next hop keep;" instead of "next hop self;"(default) + in the specific branch of the Node on "bird.cfg". + type: boolean + maxRestartTime: + description: Time to allow for software restart. When specified, + this is configured as the graceful restart timeout. When not specified, + the BIRD default of 120s is used. + type: string + node: + description: The node name identifying the Calico node instance that + is targeted by this peer. If this is not set, and no nodeSelector + is specified, then this BGP peer selects all nodes in the cluster. + type: string + nodeSelector: + description: Selector for the nodes that should have this peering. When + this is set, the Node field must be empty. + type: string + numAllowedLocalASNumbers: + description: Maximum number of local AS numbers that are allowed in + the AS path for received routes. This removes BGP loop prevention + and should only be used if absolutely necessary. + format: int32 + type: integer + password: + description: Optional BGP password for the peerings generated by this + BGPPeer resource. + properties: + secretKeyRef: + description: Selects a key of a secret in the node pod's namespace. + properties: + key: + description: The key of the secret to select from. Must be + a valid secret key. + type: string + name: + description: 'Name of the referent. More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names + TODO: Add other useful fields. apiVersion, kind, uid?' + type: string + optional: + description: Specify whether the Secret or its key must be + defined + type: boolean + required: + - key + type: object + type: object + peerIP: + description: The IP address of the peer followed by an optional port + number to peer with. If port number is given, format should be `[]:port` + or `:` for IPv4. If optional port number is not set, + and this peer IP and ASNumber belongs to a calico/node with ListenPort + set in BGPConfiguration, then we use that port to peer. + type: string + peerSelector: + description: Selector for the remote nodes to peer with. When this + is set, the PeerIP and ASNumber fields must be empty. For each + peering between the local node and selected remote nodes, we configure + an IPv4 peering if both ends have NodeBGPSpec.IPv4Address specified, + and an IPv6 peering if both ends have NodeBGPSpec.IPv6Address specified. The + remote AS number comes from the remote node's NodeBGPSpec.ASNumber, + or the global default if that is not set. + type: string + reachableBy: + description: Add an exact, i.e. /32, static route toward peer IP in + order to prevent route flapping. ReachableBy contains the address + of the gateway which peer can be reached by. + type: string + sourceAddress: + description: Specifies whether and how to configure a source address + for the peerings generated by this BGPPeer resource. Default value + "UseNodeIP" means to configure the node IP as the source address. "None" + means not to configure a source address. + type: string + ttlSecurity: + description: TTLSecurity enables the generalized TTL security mechanism + (GTSM) which protects against spoofed packets by ignoring received + packets with a smaller than expected TTL value. The provided value + is the number of hops (edges) between the peers. + type: integer + type: object + type: object + served: true + storage: true +status: + acceptedNames: + kind: "" + plural: "" + conditions: [] + storedVersions: [] +--- +# Source: calico/templates/kdd-crds.yaml +apiVersion: apiextensions.k8s.io/v1 +kind: CustomResourceDefinition +metadata: + name: blockaffinities.crd.projectcalico.org +spec: + group: crd.projectcalico.org + names: + kind: BlockAffinity + listKind: BlockAffinityList + plural: blockaffinities + singular: blockaffinity + preserveUnknownFields: false + scope: Cluster + versions: + - name: v1 + schema: + openAPIV3Schema: + properties: + apiVersion: + description: 'APIVersion defines the versioned schema of this representation + of an object. Servers should convert recognized schemas to the latest + internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources' + type: string + kind: + description: 'Kind is a string value representing the REST resource this + object represents. Servers may infer this from the endpoint the client + submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds' + type: string + metadata: + type: object + spec: + description: BlockAffinitySpec contains the specification for a BlockAffinity + resource. + properties: + cidr: + type: string + deleted: + description: Deleted indicates that this block affinity is being deleted. + This field is a string for compatibility with older releases that + mistakenly treat this field as a string. + type: string + node: + type: string + state: + type: string + required: + - cidr + - deleted + - node + - state + type: object + type: object + served: true + storage: true +status: + acceptedNames: + kind: "" + plural: "" + conditions: [] + storedVersions: [] +--- +# Source: calico/templates/kdd-crds.yaml +apiVersion: apiextensions.k8s.io/v1 +kind: CustomResourceDefinition +metadata: + annotations: + controller-gen.kubebuilder.io/version: (devel) + creationTimestamp: null + name: caliconodestatuses.crd.projectcalico.org +spec: + group: crd.projectcalico.org + names: + kind: CalicoNodeStatus + listKind: CalicoNodeStatusList + plural: caliconodestatuses + singular: caliconodestatus + preserveUnknownFields: false + scope: Cluster + versions: + - name: v1 + schema: + openAPIV3Schema: + properties: + apiVersion: + description: 'APIVersion defines the versioned schema of this representation + of an object. Servers should convert recognized schemas to the latest + internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources' + type: string + kind: + description: 'Kind is a string value representing the REST resource this + object represents. Servers may infer this from the endpoint the client + submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds' + type: string + metadata: + type: object + spec: + description: CalicoNodeStatusSpec contains the specification for a CalicoNodeStatus + resource. + properties: + classes: + description: Classes declares the types of information to monitor + for this calico/node, and allows for selective status reporting + about certain subsets of information. + items: + type: string + type: array + node: + description: The node name identifies the Calico node instance for + node status. + type: string + updatePeriodSeconds: + description: UpdatePeriodSeconds is the period at which CalicoNodeStatus + should be updated. Set to 0 to disable CalicoNodeStatus refresh. + Maximum update period is one day. + format: int32 + type: integer + type: object + status: + description: CalicoNodeStatusStatus defines the observed state of CalicoNodeStatus. + No validation needed for status since it is updated by Calico. + properties: + agent: + description: Agent holds agent status on the node. + properties: + birdV4: + description: BIRDV4 represents the latest observed status of bird4. + properties: + lastBootTime: + description: LastBootTime holds the value of lastBootTime + from bird.ctl output. + type: string + lastReconfigurationTime: + description: LastReconfigurationTime holds the value of lastReconfigTime + from bird.ctl output. + type: string + routerID: + description: Router ID used by bird. + type: string + state: + description: The state of the BGP Daemon. + type: string + version: + description: Version of the BGP daemon + type: string + type: object + birdV6: + description: BIRDV6 represents the latest observed status of bird6. + properties: + lastBootTime: + description: LastBootTime holds the value of lastBootTime + from bird.ctl output. + type: string + lastReconfigurationTime: + description: LastReconfigurationTime holds the value of lastReconfigTime + from bird.ctl output. + type: string + routerID: + description: Router ID used by bird. + type: string + state: + description: The state of the BGP Daemon. + type: string + version: + description: Version of the BGP daemon + type: string + type: object + type: object + bgp: + description: BGP holds node BGP status. + properties: + numberEstablishedV4: + description: The total number of IPv4 established bgp sessions. + type: integer + numberEstablishedV6: + description: The total number of IPv6 established bgp sessions. + type: integer + numberNotEstablishedV4: + description: The total number of IPv4 non-established bgp sessions. + type: integer + numberNotEstablishedV6: + description: The total number of IPv6 non-established bgp sessions. + type: integer + peersV4: + description: PeersV4 represents IPv4 BGP peers status on the node. + items: + description: CalicoNodePeer contains the status of BGP peers + on the node. + properties: + peerIP: + description: IP address of the peer whose condition we are + reporting. + type: string + since: + description: Since the state or reason last changed. + type: string + state: + description: State is the BGP session state. + type: string + type: + description: Type indicates whether this peer is configured + via the node-to-node mesh, or via en explicit global or + per-node BGPPeer object. + type: string + type: object + type: array + peersV6: + description: PeersV6 represents IPv6 BGP peers status on the node. + items: + description: CalicoNodePeer contains the status of BGP peers + on the node. + properties: + peerIP: + description: IP address of the peer whose condition we are + reporting. + type: string + since: + description: Since the state or reason last changed. + type: string + state: + description: State is the BGP session state. + type: string + type: + description: Type indicates whether this peer is configured + via the node-to-node mesh, or via en explicit global or + per-node BGPPeer object. + type: string + type: object + type: array + required: + - numberEstablishedV4 + - numberEstablishedV6 + - numberNotEstablishedV4 + - numberNotEstablishedV6 + type: object + lastUpdated: + description: LastUpdated is a timestamp representing the server time + when CalicoNodeStatus object last updated. It is represented in + RFC3339 form and is in UTC. + format: date-time + nullable: true + type: string + routes: + description: Routes reports routes known to the Calico BGP daemon + on the node. + properties: + routesV4: + description: RoutesV4 represents IPv4 routes on the node. + items: + description: CalicoNodeRoute contains the status of BGP routes + on the node. + properties: + destination: + description: Destination of the route. + type: string + gateway: + description: Gateway for the destination. + type: string + interface: + description: Interface for the destination + type: string + learnedFrom: + description: LearnedFrom contains information regarding + where this route originated. + properties: + peerIP: + description: If sourceType is NodeMesh or BGPPeer, IP + address of the router that sent us this route. + type: string + sourceType: + description: Type of the source where a route is learned + from. + type: string + type: object + type: + description: Type indicates if the route is being used for + forwarding or not. + type: string + type: object + type: array + routesV6: + description: RoutesV6 represents IPv6 routes on the node. + items: + description: CalicoNodeRoute contains the status of BGP routes + on the node. + properties: + destination: + description: Destination of the route. + type: string + gateway: + description: Gateway for the destination. + type: string + interface: + description: Interface for the destination + type: string + learnedFrom: + description: LearnedFrom contains information regarding + where this route originated. + properties: + peerIP: + description: If sourceType is NodeMesh or BGPPeer, IP + address of the router that sent us this route. + type: string + sourceType: + description: Type of the source where a route is learned + from. + type: string + type: object + type: + description: Type indicates if the route is being used for + forwarding or not. + type: string + type: object + type: array + type: object + type: object + type: object + served: true + storage: true +status: + acceptedNames: + kind: "" + plural: "" + conditions: [] + storedVersions: [] +--- +# Source: calico/templates/kdd-crds.yaml +apiVersion: apiextensions.k8s.io/v1 +kind: CustomResourceDefinition +metadata: + name: clusterinformations.crd.projectcalico.org +spec: + group: crd.projectcalico.org + names: + kind: ClusterInformation + listKind: ClusterInformationList + plural: clusterinformations + singular: clusterinformation + preserveUnknownFields: false + scope: Cluster + versions: + - name: v1 + schema: + openAPIV3Schema: + description: ClusterInformation contains the cluster specific information. + properties: + apiVersion: + description: 'APIVersion defines the versioned schema of this representation + of an object. Servers should convert recognized schemas to the latest + internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources' + type: string + kind: + description: 'Kind is a string value representing the REST resource this + object represents. Servers may infer this from the endpoint the client + submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds' + type: string + metadata: + type: object + spec: + description: ClusterInformationSpec contains the values of describing + the cluster. + properties: + calicoVersion: + description: CalicoVersion is the version of Calico that the cluster + is running + type: string + clusterGUID: + description: ClusterGUID is the GUID of the cluster + type: string + clusterType: + description: ClusterType describes the type of the cluster + type: string + datastoreReady: + description: DatastoreReady is used during significant datastore migrations + to signal to components such as Felix that it should wait before + accessing the datastore. + type: boolean + variant: + description: Variant declares which variant of Calico should be active. + type: string + type: object + type: object + served: true + storage: true +status: + acceptedNames: + kind: "" + plural: "" + conditions: [] + storedVersions: [] +--- +# Source: calico/templates/kdd-crds.yaml +apiVersion: apiextensions.k8s.io/v1 +kind: CustomResourceDefinition +metadata: + name: felixconfigurations.crd.projectcalico.org +spec: + group: crd.projectcalico.org + names: + kind: FelixConfiguration + listKind: FelixConfigurationList + plural: felixconfigurations + singular: felixconfiguration + preserveUnknownFields: false + scope: Cluster + versions: + - name: v1 + schema: + openAPIV3Schema: + description: Felix Configuration contains the configuration for Felix. + properties: + apiVersion: + description: 'APIVersion defines the versioned schema of this representation + of an object. Servers should convert recognized schemas to the latest + internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources' + type: string + kind: + description: 'Kind is a string value representing the REST resource this + object represents. Servers may infer this from the endpoint the client + submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds' + type: string + metadata: + type: object + spec: + description: FelixConfigurationSpec contains the values of the Felix configuration. + properties: + allowIPIPPacketsFromWorkloads: + description: 'AllowIPIPPacketsFromWorkloads controls whether Felix + will add a rule to drop IPIP encapsulated traffic from workloads + [Default: false]' + type: boolean + allowVXLANPacketsFromWorkloads: + description: 'AllowVXLANPacketsFromWorkloads controls whether Felix + will add a rule to drop VXLAN encapsulated traffic from workloads + [Default: false]' + type: boolean + awsSrcDstCheck: + description: 'Set source-destination-check on AWS EC2 instances. Accepted + value must be one of "DoNothing", "Enable" or "Disable". [Default: + DoNothing]' + enum: + - DoNothing + - Enable + - Disable + type: string + bpfCTLBLogFilter: + description: 'BPFCTLBLogFilter specifies, what is logged by connect + time load balancer when BPFLogLevel is debug. Currently has to be + specified as ''all'' when BPFLogFilters is set to see CTLB logs. + [Default: unset - means logs are emitted when BPFLogLevel id debug + and BPFLogFilters not set.]' + type: string + bpfConnectTimeLoadBalancing: + description: 'BPFConnectTimeLoadBalancing when in BPF mode, controls + whether Felix installs the connect-time load balancer. The connect-time + load balancer is required for the host to be able to reach Kubernetes + services and it improves the performance of pod-to-service connections.When + set to TCP, connect time load balancing is available only for services + with TCP ports. [Default: TCP]' + enum: + - TCP + - Enabled + - Disabled + type: string + bpfConnectTimeLoadBalancingEnabled: + description: 'BPFConnectTimeLoadBalancingEnabled when in BPF mode, + controls whether Felix installs the connection-time load balancer. The + connect-time load balancer is required for the host to be able to + reach Kubernetes services and it improves the performance of pod-to-service + connections. The only reason to disable it is for debugging purposes. + This will be deprecated. Use BPFConnectTimeLoadBalancing [Default: + true]' + type: boolean + bpfDSROptoutCIDRs: + description: BPFDSROptoutCIDRs is a list of CIDRs which are excluded + from DSR. That is, clients in those CIDRs will accesses nodeports + as if BPFExternalServiceMode was set to Tunnel. + items: + type: string + type: array + bpfDataIfacePattern: + description: BPFDataIfacePattern is a regular expression that controls + which interfaces Felix should attach BPF programs to in order to + catch traffic to/from the network. This needs to match the interfaces + that Calico workload traffic flows over as well as any interfaces + that handle incoming traffic to nodeports and services from outside + the cluster. It should not match the workload interfaces (usually + named cali...). + type: string + bpfDisableGROForIfaces: + description: BPFDisableGROForIfaces is a regular expression that controls + which interfaces Felix should disable the Generic Receive Offload + [GRO] option. It should not match the workload interfaces (usually + named cali...). + type: string + bpfDisableUnprivileged: + description: 'BPFDisableUnprivileged, if enabled, Felix sets the kernel.unprivileged_bpf_disabled + sysctl to disable unprivileged use of BPF. This ensures that unprivileged + users cannot access Calico''s BPF maps and cannot insert their own + BPF programs to interfere with Calico''s. [Default: true]' + type: boolean + bpfEnabled: + description: 'BPFEnabled, if enabled Felix will use the BPF dataplane. + [Default: false]' + type: boolean + bpfEnforceRPF: + description: 'BPFEnforceRPF enforce strict RPF on all host interfaces + with BPF programs regardless of what is the per-interfaces or global + setting. Possible values are Disabled, Strict or Loose. [Default: + Loose]' + pattern: ^(?i)(Disabled|Strict|Loose)?$ + type: string + bpfExcludeCIDRsFromNAT: + description: BPFExcludeCIDRsFromNAT is a list of CIDRs that are to + be excluded from NAT resolution so that host can handle them. A + typical usecase is node local DNS cache. + items: + type: string + type: array + bpfExtToServiceConnmark: + description: 'BPFExtToServiceConnmark in BPF mode, control a 32bit + mark that is set on connections from an external client to a local + service. This mark allows us to control how packets of that connection + are routed within the host and how is routing interpreted by RPF + check. [Default: 0]' + type: integer + bpfExternalServiceMode: + description: 'BPFExternalServiceMode in BPF mode, controls how connections + from outside the cluster to services (node ports and cluster IPs) + are forwarded to remote workloads. If set to "Tunnel" then both + request and response traffic is tunneled to the remote node. If + set to "DSR", the request traffic is tunneled but the response traffic + is sent directly from the remote node. In "DSR" mode, the remote + node appears to use the IP of the ingress node; this requires a + permissive L2 network. [Default: Tunnel]' + pattern: ^(?i)(Tunnel|DSR)?$ + type: string + bpfForceTrackPacketsFromIfaces: + description: 'BPFForceTrackPacketsFromIfaces in BPF mode, forces traffic + from these interfaces to skip Calico''s iptables NOTRACK rule, allowing + traffic from those interfaces to be tracked by Linux conntrack. Should + only be used for interfaces that are not used for the Calico fabric. For + example, a docker bridge device for non-Calico-networked containers. + [Default: docker+]' + items: + type: string + type: array + bpfHostConntrackBypass: + description: 'BPFHostConntrackBypass Controls whether to bypass Linux + conntrack in BPF mode for workloads and services. [Default: true + - bypass Linux conntrack]' + type: boolean + bpfHostNetworkedNATWithoutCTLB: + description: 'BPFHostNetworkedNATWithoutCTLB when in BPF mode, controls + whether Felix does a NAT without CTLB. This along with BPFConnectTimeLoadBalancing + determines the CTLB behavior. [Default: Enabled]' + enum: + - Enabled + - Disabled + type: string + bpfKubeProxyEndpointSlicesEnabled: + description: BPFKubeProxyEndpointSlicesEnabled is deprecated and has + no effect. BPF kube-proxy always accepts endpoint slices. This option + will be removed in the next release. + type: boolean + bpfKubeProxyIptablesCleanupEnabled: + description: 'BPFKubeProxyIptablesCleanupEnabled, if enabled in BPF + mode, Felix will proactively clean up the upstream Kubernetes kube-proxy''s + iptables chains. Should only be enabled if kube-proxy is not running. [Default: + true]' + type: boolean + bpfKubeProxyMinSyncPeriod: + description: 'BPFKubeProxyMinSyncPeriod, in BPF mode, controls the + minimum time between updates to the dataplane for Felix''s embedded + kube-proxy. Lower values give reduced set-up latency. Higher values + reduce Felix CPU usage by batching up more work. [Default: 1s]' + pattern: ^([0-9]+(\\.[0-9]+)?(ms|s|m|h))*$ + type: string + bpfL3IfacePattern: + description: BPFL3IfacePattern is a regular expression that allows + to list tunnel devices like wireguard or vxlan (i.e., L3 devices) + in addition to BPFDataIfacePattern. That is, tunnel interfaces not + created by Calico, that Calico workload traffic flows over as well + as any interfaces that handle incoming traffic to nodeports and + services from outside the cluster. + type: string + bpfLogFilters: + additionalProperties: + type: string + description: "BPFLogFilters is a map of key=values where the value + is a pcap filter expression and the key is an interface name with + 'all' denoting all interfaces, 'weps' all workload endpoints and + 'heps' all host endpoints. \n When specified as an env var, it accepts + a comma-separated list of key=values. [Default: unset - means all + debug logs are emitted]" + type: object + bpfLogLevel: + description: 'BPFLogLevel controls the log level of the BPF programs + when in BPF dataplane mode. One of "Off", "Info", or "Debug". The + logs are emitted to the BPF trace pipe, accessible with the command + `tc exec bpf debug`. [Default: Off].' + pattern: ^(?i)(Off|Info|Debug)?$ + type: string + bpfMapSizeConntrack: + description: 'BPFMapSizeConntrack sets the size for the conntrack + map. This map must be large enough to hold an entry for each active + connection. Warning: changing the size of the conntrack map can + cause disruption.' + type: integer + bpfMapSizeIPSets: + description: BPFMapSizeIPSets sets the size for ipsets map. The IP + sets map must be large enough to hold an entry for each endpoint + matched by every selector in the source/destination matches in network + policy. Selectors such as "all()" can result in large numbers of + entries (one entry per endpoint in that case). + type: integer + bpfMapSizeIfState: + description: BPFMapSizeIfState sets the size for ifstate map. The + ifstate map must be large enough to hold an entry for each device + (host + workloads) on a host. + type: integer + bpfMapSizeNATAffinity: + type: integer + bpfMapSizeNATBackend: + description: BPFMapSizeNATBackend sets the size for nat back end map. + This is the total number of endpoints. This is mostly more than + the size of the number of services. + type: integer + bpfMapSizeNATFrontend: + description: BPFMapSizeNATFrontend sets the size for nat front end + map. FrontendMap should be large enough to hold an entry for each + nodeport, external IP and each port in each service. + type: integer + bpfMapSizeRoute: + description: BPFMapSizeRoute sets the size for the routes map. The + routes map should be large enough to hold one entry per workload + and a handful of entries per host (enough to cover its own IPs and + tunnel IPs). + type: integer + bpfPSNATPorts: + anyOf: + - type: integer + - type: string + description: 'BPFPSNATPorts sets the range from which we randomly + pick a port if there is a source port collision. This should be + within the ephemeral range as defined by RFC 6056 (1024–65535) and + preferably outside the ephemeral ranges used by common operating + systems. Linux uses 32768–60999, while others mostly use the IANA + defined range 49152–65535. It is not necessarily a problem if this + range overlaps with the operating systems. Both ends of the range + are inclusive. [Default: 20000:29999]' + pattern: ^.* + x-kubernetes-int-or-string: true + bpfPolicyDebugEnabled: + description: BPFPolicyDebugEnabled when true, Felix records detailed + information about the BPF policy programs, which can be examined + with the calico-bpf command-line tool. + type: boolean + bpfRedirectToPeer: + description: 'BPFRedirectToPeer controls which whether it is allowed + to forward straight to the peer side of the workload devices. It + is allowed for any host L2 devices by default (L2Only), but it breaks + TCP dump on the host side of workload device as it bypasses it on + ingress. Value of Enabled also allows redirection from L3 host devices + like IPIP tunnel or Wireguard directly to the peer side of the workload''s + device. This makes redirection faster, however, it breaks tools + like tcpdump on the peer side. Use Enabled with caution. [Default: + L2Only]' + type: string + chainInsertMode: + description: 'ChainInsertMode controls whether Felix hooks the kernel''s + top-level iptables chains by inserting a rule at the top of the + chain or by appending a rule at the bottom. insert is the safe default + since it prevents Calico''s rules from being bypassed. If you switch + to append mode, be sure that the other rules in the chains signal + acceptance by falling through to the Calico rules, otherwise the + Calico policy will be bypassed. [Default: insert]' + pattern: ^(?i)(insert|append)?$ + type: string + dataplaneDriver: + description: DataplaneDriver filename of the external dataplane driver + to use. Only used if UseInternalDataplaneDriver is set to false. + type: string + dataplaneWatchdogTimeout: + description: "DataplaneWatchdogTimeout is the readiness/liveness timeout + used for Felix's (internal) dataplane driver. Increase this value + if you experience spurious non-ready or non-live events when Felix + is under heavy load. Decrease the value to get felix to report non-live + or non-ready more quickly. [Default: 90s] \n Deprecated: replaced + by the generic HealthTimeoutOverrides." + type: string + debugDisableLogDropping: + type: boolean + debugHost: + description: DebugHost is the host IP or hostname to bind the debug + port to. Only used if DebugPort is set. [Default:localhost] + type: string + debugMemoryProfilePath: + type: string + debugPort: + description: DebugPort if set, enables Felix's debug HTTP port, which + allows memory and CPU profiles to be retrieved. The debug port + is not secure, it should not be exposed to the internet. + type: integer + debugSimulateCalcGraphHangAfter: + pattern: ^([0-9]+(\\.[0-9]+)?(ms|s|m|h))*$ + type: string + debugSimulateDataplaneApplyDelay: + pattern: ^([0-9]+(\\.[0-9]+)?(ms|s|m|h))*$ + type: string + debugSimulateDataplaneHangAfter: + pattern: ^([0-9]+(\\.[0-9]+)?(ms|s|m|h))*$ + type: string + defaultEndpointToHostAction: + description: 'DefaultEndpointToHostAction controls what happens to + traffic that goes from a workload endpoint to the host itself (after + the traffic hits the endpoint egress policy). By default Calico + blocks traffic from workload endpoints to the host itself with an + iptables "DROP" action. If you want to allow some or all traffic + from endpoint to host, set this parameter to RETURN or ACCEPT. Use + RETURN if you have your own rules in the iptables "INPUT" chain; + Calico will insert its rules at the top of that chain, then "RETURN" + packets to the "INPUT" chain once it has completed processing workload + endpoint egress policy. Use ACCEPT to unconditionally accept packets + from workloads after processing workload endpoint egress policy. + [Default: Drop]' + pattern: ^(?i)(Drop|Accept|Return)?$ + type: string + deviceRouteProtocol: + description: This defines the route protocol added to programmed device + routes, by default this will be RTPROT_BOOT when left blank. + type: integer + deviceRouteSourceAddress: + description: This is the IPv4 source address to use on programmed + device routes. By default the source address is left blank, leaving + the kernel to choose the source address used. + type: string + deviceRouteSourceAddressIPv6: + description: This is the IPv6 source address to use on programmed + device routes. By default the source address is left blank, leaving + the kernel to choose the source address used. + type: string + disableConntrackInvalidCheck: + type: boolean + endpointReportingDelay: + pattern: ^([0-9]+(\\.[0-9]+)?(ms|s|m|h))*$ + type: string + endpointReportingEnabled: + type: boolean + endpointStatusPathPrefix: + description: "EndpointStatusPathPrefix is the path to the directory + where endpoint status will be written. Endpoint status file reporting + is disabled if field is left empty. \n Chosen directory should match + the directory used by the CNI for PodStartupDelay. [Default: \"\"]" + type: string + externalNodesList: + description: ExternalNodesCIDRList is a list of CIDR's of external-non-calico-nodes + which may source tunnel traffic and have the tunneled traffic be + accepted at calico nodes. + items: + type: string + type: array + failsafeInboundHostPorts: + description: 'FailsafeInboundHostPorts is a list of PortProto struct + objects including UDP/TCP/SCTP ports and CIDRs that Felix will allow + incoming traffic to host endpoints on irrespective of the security + policy. This is useful to avoid accidentally cutting off a host + with incorrect configuration. For backwards compatibility, if the + protocol is not specified, it defaults to "tcp". If a CIDR is not + specified, it will allow traffic from all addresses. To disable + all inbound host ports, use the value "[]". The default value allows + ssh access, DHCP, BGP, etcd and the Kubernetes API. [Default: tcp:22, + udp:68, tcp:179, tcp:2379, tcp:2380, tcp:5473, tcp:6443, tcp:6666, + tcp:6667 ]' + items: + description: ProtoPort is combination of protocol, port, and CIDR. + Protocol and port must be specified. + properties: + net: + type: string + port: + type: integer + protocol: + type: string + required: + - port + - protocol + type: object + type: array + failsafeOutboundHostPorts: + description: 'FailsafeOutboundHostPorts is a list of List of PortProto + struct objects including UDP/TCP/SCTP ports and CIDRs that Felix + will allow outgoing traffic from host endpoints to irrespective + of the security policy. This is useful to avoid accidentally cutting + off a host with incorrect configuration. For backwards compatibility, + if the protocol is not specified, it defaults to "tcp". If a CIDR + is not specified, it will allow traffic from all addresses. To disable + all outbound host ports, use the value "[]". The default value opens + etcd''s standard ports to ensure that Felix does not get cut off + from etcd as well as allowing DHCP, DNS, BGP and the Kubernetes + API. [Default: udp:53, udp:67, tcp:179, tcp:2379, tcp:2380, tcp:5473, + tcp:6443, tcp:6666, tcp:6667 ]' + items: + description: ProtoPort is combination of protocol, port, and CIDR. + Protocol and port must be specified. + properties: + net: + type: string + port: + type: integer + protocol: + type: string + required: + - port + - protocol + type: object + type: array + featureDetectOverride: + description: FeatureDetectOverride is used to override feature detection + based on auto-detected platform capabilities. Values are specified + in a comma separated list with no spaces, example; "SNATFullyRandom=true,MASQFullyRandom=false,RestoreSupportsLock=". "true" + or "false" will force the feature, empty or omitted values are auto-detected. + pattern: ^([a-zA-Z0-9-_]+=(true|false|),)*([a-zA-Z0-9-_]+=(true|false|))?$ + type: string + featureGates: + description: FeatureGates is used to enable or disable tech-preview + Calico features. Values are specified in a comma separated list + with no spaces, example; "BPFConnectTimeLoadBalancingWorkaround=enabled,XyZ=false". + This is used to enable features that are not fully production ready. + pattern: ^([a-zA-Z0-9-_]+=([^=]+),)*([a-zA-Z0-9-_]+=([^=]+))?$ + type: string + floatingIPs: + description: FloatingIPs configures whether or not Felix will program + non-OpenStack floating IP addresses. (OpenStack-derived floating + IPs are always programmed, regardless of this setting.) + enum: + - Enabled + - Disabled + type: string + genericXDPEnabled: + description: 'GenericXDPEnabled enables Generic XDP so network cards + that don''t support XDP offload or driver modes can use XDP. This + is not recommended since it doesn''t provide better performance + than iptables. [Default: false]' + type: boolean + goGCThreshold: + description: "GoGCThreshold Sets the Go runtime's garbage collection + threshold. I.e. the percentage that the heap is allowed to grow + before garbage collection is triggered. In general, doubling the + value halves the CPU time spent doing GC, but it also doubles peak + GC memory overhead. A special value of -1 can be used to disable + GC entirely; this should only be used in conjunction with the GoMemoryLimitMB + setting. \n This setting is overridden by the GOGC environment variable. + \n [Default: 40]" + type: integer + goMaxProcs: + description: "GoMaxProcs sets the maximum number of CPUs that the + Go runtime will use concurrently. A value of -1 means \"use the + system default\"; typically the number of real CPUs on the system. + \n this setting is overridden by the GOMAXPROCS environment variable. + \n [Default: -1]" + type: integer + goMemoryLimitMB: + description: "GoMemoryLimitMB sets a (soft) memory limit for the Go + runtime in MB. The Go runtime will try to keep its memory usage + under the limit by triggering GC as needed. To avoid thrashing, + it will exceed the limit if GC starts to take more than 50% of the + process's CPU time. A value of -1 disables the memory limit. \n + Note that the memory limit, if used, must be considerably less than + any hard resource limit set at the container or pod level. This + is because felix is not the only process that must run in the container + or pod. \n This setting is overridden by the GOMEMLIMIT environment + variable. \n [Default: -1]" + type: integer + healthEnabled: + type: boolean + healthHost: + type: string + healthPort: + type: integer + healthTimeoutOverrides: + description: HealthTimeoutOverrides allows the internal watchdog timeouts + of individual subcomponents to be overridden. This is useful for + working around "false positive" liveness timeouts that can occur + in particularly stressful workloads or if CPU is constrained. For + a list of active subcomponents, see Felix's logs. + items: + properties: + name: + type: string + timeout: + type: string + required: + - name + - timeout + type: object + type: array + interfaceExclude: + description: 'InterfaceExclude is a comma-separated list of interfaces + that Felix should exclude when monitoring for host endpoints. The + default value ensures that Felix ignores Kubernetes'' IPVS dummy + interface, which is used internally by kube-proxy. If you want to + exclude multiple interface names using a single value, the list + supports regular expressions. For regular expressions you must wrap + the value with ''/''. For example having values ''/^kube/,veth1'' + will exclude all interfaces that begin with ''kube'' and also the + interface ''veth1''. [Default: kube-ipvs0]' + type: string + interfacePrefix: + description: 'InterfacePrefix is the interface name prefix that identifies + workload endpoints and so distinguishes them from host endpoint + interfaces. Note: in environments other than bare metal, the orchestrators + configure this appropriately. For example our Kubernetes and Docker + integrations set the ''cali'' value, and our OpenStack integration + sets the ''tap'' value. [Default: cali]' + type: string + interfaceRefreshInterval: + description: InterfaceRefreshInterval is the period at which Felix + rescans local interfaces to verify their state. The rescan can be + disabled by setting the interval to 0. + pattern: ^([0-9]+(\\.[0-9]+)?(ms|s|m|h))*$ + type: string + ipForwarding: + description: 'IPForwarding controls whether Felix sets the host sysctls + to enable IP forwarding. IP forwarding is required when using Calico + for workload networking. This should only be disabled on hosts + where Calico is used for host protection. [Default: Enabled]' + enum: + - Enabled + - Disabled + type: string + ipipEnabled: + description: 'IPIPEnabled overrides whether Felix should configure + an IPIP interface on the host. Optional as Felix determines this + based on the existing IP pools. [Default: nil (unset)]' + type: boolean + ipipMTU: + description: 'IPIPMTU is the MTU to set on the tunnel device. See + Configuring MTU [Default: 1440]' + type: integer + ipsetsRefreshInterval: + description: 'IpsetsRefreshInterval is the period at which Felix re-checks + all iptables state to ensure that no other process has accidentally + broken Calico''s rules. Set to 0 to disable iptables refresh. [Default: + 90s]' + pattern: ^([0-9]+(\\.[0-9]+)?(ms|s|m|h))*$ + type: string + iptablesBackend: + description: IptablesBackend specifies which backend of iptables will + be used. The default is Auto. + pattern: ^(?i)(Auto|FelixConfiguration|FelixConfigurationList|Legacy|NFT)?$ + type: string + iptablesFilterAllowAction: + pattern: ^(?i)(Accept|Return)?$ + type: string + iptablesFilterDenyAction: + description: IptablesFilterDenyAction controls what happens to traffic + that is denied by network policy. By default Calico blocks traffic + with an iptables "DROP" action. If you want to use "REJECT" action + instead you can configure it in here. + pattern: ^(?i)(Drop|Reject)?$ + type: string + iptablesLockFilePath: + description: 'IptablesLockFilePath is the location of the iptables + lock file. You may need to change this if the lock file is not in + its standard location (for example if you have mapped it into Felix''s + container at a different path). [Default: /run/xtables.lock]' + type: string + iptablesLockProbeInterval: + description: 'IptablesLockProbeInterval is the time that Felix will + wait between attempts to acquire the iptables lock if it is not + available. Lower values make Felix more responsive when the lock + is contended, but use more CPU. [Default: 50ms]' + pattern: ^([0-9]+(\\.[0-9]+)?(ms|s|m|h))*$ + type: string + iptablesLockTimeout: + description: 'IptablesLockTimeout is the time that Felix will wait + for the iptables lock, or 0, to disable. To use this feature, Felix + must share the iptables lock file with all other processes that + also take the lock. When running Felix inside a container, this + requires the /run directory of the host to be mounted into the calico/node + or calico/felix container. [Default: 0s disabled]' + pattern: ^([0-9]+(\\.[0-9]+)?(ms|s|m|h))*$ + type: string + iptablesMangleAllowAction: + pattern: ^(?i)(Accept|Return)?$ + type: string + iptablesMarkMask: + description: 'IptablesMarkMask is the mask that Felix selects its + IPTables Mark bits from. Should be a 32 bit hexadecimal number with + at least 8 bits set, none of which clash with any other mark bits + in use on the system. [Default: 0xff000000]' + format: int32 + type: integer + iptablesNATOutgoingInterfaceFilter: + type: string + iptablesPostWriteCheckInterval: + description: 'IptablesPostWriteCheckInterval is the period after Felix + has done a write to the dataplane that it schedules an extra read + back in order to check the write was not clobbered by another process. + This should only occur if another application on the system doesn''t + respect the iptables lock. [Default: 1s]' + pattern: ^([0-9]+(\\.[0-9]+)?(ms|s|m|h))*$ + type: string + iptablesRefreshInterval: + description: 'IptablesRefreshInterval is the period at which Felix + re-checks the IP sets in the dataplane to ensure that no other process + has accidentally broken Calico''s rules. Set to 0 to disable IP + sets refresh. Note: the default for this value is lower than the + other refresh intervals as a workaround for a Linux kernel bug that + was fixed in kernel version 4.11. If you are using v4.11 or greater + you may want to set this to, a higher value to reduce Felix CPU + usage. [Default: 10s]' + pattern: ^([0-9]+(\\.[0-9]+)?(ms|s|m|h))*$ + type: string + ipv6Support: + description: IPv6Support controls whether Felix enables support for + IPv6 (if supported by the in-use dataplane). + type: boolean + kubeNodePortRanges: + description: 'KubeNodePortRanges holds list of port ranges used for + service node ports. Only used if felix detects kube-proxy running + in ipvs mode. Felix uses these ranges to separate host and workload + traffic. [Default: 30000:32767].' + items: + anyOf: + - type: integer + - type: string + pattern: ^.* + x-kubernetes-int-or-string: true + type: array + logDebugFilenameRegex: + description: LogDebugFilenameRegex controls which source code files + have their Debug log output included in the logs. Only logs from + files with names that match the given regular expression are included. The + filter only applies to Debug level logs. + type: string + logFilePath: + description: 'LogFilePath is the full path to the Felix log. Set to + none to disable file logging. [Default: /var/log/calico/felix.log]' + type: string + logPrefix: + description: 'LogPrefix is the log prefix that Felix uses when rendering + LOG rules. [Default: calico-packet]' + type: string + logSeverityFile: + description: 'LogSeverityFile is the log severity above which logs + are sent to the log file. [Default: Info]' + pattern: ^(?i)(Debug|Info|Warning|Error|Fatal)?$ + type: string + logSeverityScreen: + description: 'LogSeverityScreen is the log severity above which logs + are sent to the stdout. [Default: Info]' + pattern: ^(?i)(Debug|Info|Warning|Error|Fatal)?$ + type: string + logSeveritySys: + description: 'LogSeveritySys is the log severity above which logs + are sent to the syslog. Set to None for no logging to syslog. [Default: + Info]' + pattern: ^(?i)(Debug|Info|Warning|Error|Fatal)?$ + type: string + maxIpsetSize: + description: MaxIpsetSize is the maximum number of IP addresses that + can be stored in an IP set. Not applicable if using the nftables + backend. + type: integer + metadataAddr: + description: 'MetadataAddr is the IP address or domain name of the + server that can answer VM queries for cloud-init metadata. In OpenStack, + this corresponds to the machine running nova-api (or in Ubuntu, + nova-api-metadata). A value of none (case-insensitive) means that + Felix should not set up any NAT rule for the metadata path. [Default: + 127.0.0.1]' + type: string + metadataPort: + description: 'MetadataPort is the port of the metadata server. This, + combined with global.MetadataAddr (if not ''None''), is used to + set up a NAT rule, from 169.254.169.254:80 to MetadataAddr:MetadataPort. + In most cases this should not need to be changed [Default: 8775].' + type: integer + mtuIfacePattern: + description: MTUIfacePattern is a regular expression that controls + which interfaces Felix should scan in order to calculate the host's + MTU. This should not match workload interfaces (usually named cali...). + type: string + natOutgoingAddress: + description: NATOutgoingAddress specifies an address to use when performing + source NAT for traffic in a natOutgoing pool that is leaving the + network. By default the address used is an address on the interface + the traffic is leaving on (ie it uses the iptables MASQUERADE target) + type: string + natPortRange: + anyOf: + - type: integer + - type: string + description: NATPortRange specifies the range of ports that is used + for port mapping when doing outgoing NAT. When unset the default + behavior of the network stack is used. + pattern: ^.* + x-kubernetes-int-or-string: true + netlinkTimeout: + pattern: ^([0-9]+(\\.[0-9]+)?(ms|s|m|h))*$ + type: string + nftablesFilterAllowAction: + pattern: ^(?i)(Accept|Return)?$ + type: string + nftablesFilterDenyAction: + description: FilterDenyAction controls what happens to traffic that + is denied by network policy. By default Calico blocks traffic with + a "drop" action. If you want to use a "reject" action instead you + can configure it here. + pattern: ^(?i)(Drop|Reject)?$ + type: string + nftablesMangleAllowAction: + pattern: ^(?i)(Accept|Return)?$ + type: string + nftablesMarkMask: + description: 'MarkMask is the mask that Felix selects its nftables + Mark bits from. Should be a 32 bit hexadecimal number with at least + 8 bits set, none of which clash with any other mark bits in use + on the system. [Default: 0xffff0000]' + format: int32 + type: integer + nftablesMode: + description: 'NFTablesMode configures nftables support in Felix. [Default: + Disabled]' + type: string + nftablesRefreshInterval: + description: 'NftablesRefreshInterval controls the interval at which + Felix periodically refreshes the nftables rules. [Default: 90s]' + type: string + openstackRegion: + description: 'OpenstackRegion is the name of the region that a particular + Felix belongs to. In a multi-region Calico/OpenStack deployment, + this must be configured somehow for each Felix (here in the datamodel, + or in felix.cfg or the environment on each compute node), and must + match the [calico] openstack_region value configured in neutron.conf + on each node. [Default: Empty]' + type: string + policySyncPathPrefix: + description: 'PolicySyncPathPrefix is used to by Felix to communicate + policy changes to external services, like Application layer policy. + [Default: Empty]' + type: string + prometheusGoMetricsEnabled: + description: 'PrometheusGoMetricsEnabled disables Go runtime metrics + collection, which the Prometheus client does by default, when set + to false. This reduces the number of metrics reported, reducing + Prometheus load. [Default: true]' + type: boolean + prometheusMetricsEnabled: + description: 'PrometheusMetricsEnabled enables the Prometheus metrics + server in Felix if set to true. [Default: false]' + type: boolean + prometheusMetricsHost: + description: 'PrometheusMetricsHost is the host that the Prometheus + metrics server should bind to. [Default: empty]' + type: string + prometheusMetricsPort: + description: 'PrometheusMetricsPort is the TCP port that the Prometheus + metrics server should bind to. [Default: 9091]' + type: integer + prometheusProcessMetricsEnabled: + description: 'PrometheusProcessMetricsEnabled disables process metrics + collection, which the Prometheus client does by default, when set + to false. This reduces the number of metrics reported, reducing + Prometheus load. [Default: true]' + type: boolean + prometheusWireGuardMetricsEnabled: + description: 'PrometheusWireGuardMetricsEnabled disables wireguard + metrics collection, which the Prometheus client does by default, + when set to false. This reduces the number of metrics reported, + reducing Prometheus load. [Default: true]' + type: boolean + removeExternalRoutes: + description: Whether or not to remove device routes that have not + been programmed by Felix. Disabling this will allow external applications + to also add device routes. This is enabled by default which means + we will remove externally added routes. + type: boolean + reportingInterval: + description: 'ReportingInterval is the interval at which Felix reports + its status into the datastore or 0 to disable. Must be non-zero + in OpenStack deployments. [Default: 30s]' + pattern: ^([0-9]+(\\.[0-9]+)?(ms|s|m|h))*$ + type: string + reportingTTL: + description: 'ReportingTTL is the time-to-live setting for process-wide + status reports. [Default: 90s]' + pattern: ^([0-9]+(\\.[0-9]+)?(ms|s|m|h))*$ + type: string + routeRefreshInterval: + description: 'RouteRefreshInterval is the period at which Felix re-checks + the routes in the dataplane to ensure that no other process has + accidentally broken Calico''s rules. Set to 0 to disable route refresh. + [Default: 90s]' + pattern: ^([0-9]+(\\.[0-9]+)?(ms|s|m|h))*$ + type: string + routeSource: + description: 'RouteSource configures where Felix gets its routing + information. - WorkloadIPs: use workload endpoints to construct + routes. - CalicoIPAM: the default - use IPAM data to construct routes.' + pattern: ^(?i)(WorkloadIPs|CalicoIPAM)?$ + type: string + routeSyncDisabled: + description: RouteSyncDisabled will disable all operations performed + on the route table. Set to true to run in network-policy mode only. + type: boolean + routeTableRange: + description: Deprecated in favor of RouteTableRanges. Calico programs + additional Linux route tables for various purposes. RouteTableRange + specifies the indices of the route tables that Calico should use. + properties: + max: + type: integer + min: + type: integer + required: + - max + - min + type: object + routeTableRanges: + description: Calico programs additional Linux route tables for various + purposes. RouteTableRanges specifies a set of table index ranges + that Calico should use. Deprecates`RouteTableRange`, overrides `RouteTableRange`. + items: + properties: + max: + type: integer + min: + type: integer + required: + - max + - min + type: object + type: array + serviceLoopPrevention: + description: 'When service IP advertisement is enabled, prevent routing + loops to service IPs that are not in use, by dropping or rejecting + packets that do not get DNAT''d by kube-proxy. Unless set to "Disabled", + in which case such routing loops continue to be allowed. [Default: + Drop]' + pattern: ^(?i)(Drop|Reject|Disabled)?$ + type: string + sidecarAccelerationEnabled: + description: 'SidecarAccelerationEnabled enables experimental sidecar + acceleration [Default: false]' + type: boolean + usageReportingEnabled: + description: 'UsageReportingEnabled reports anonymous Calico version + number and cluster size to projectcalico.org. Logs warnings returned + by the usage server. For example, if a significant security vulnerability + has been discovered in the version of Calico being used. [Default: + true]' + type: boolean + usageReportingInitialDelay: + description: 'UsageReportingInitialDelay controls the minimum delay + before Felix makes a report. [Default: 300s]' + pattern: ^([0-9]+(\\.[0-9]+)?(ms|s|m|h))*$ + type: string + usageReportingInterval: + description: 'UsageReportingInterval controls the interval at which + Felix makes reports. [Default: 86400s]' + pattern: ^([0-9]+(\\.[0-9]+)?(ms|s|m|h))*$ + type: string + useInternalDataplaneDriver: + description: UseInternalDataplaneDriver, if true, Felix will use its + internal dataplane programming logic. If false, it will launch + an external dataplane driver and communicate with it over protobuf. + type: boolean + vxlanEnabled: + description: 'VXLANEnabled overrides whether Felix should create the + VXLAN tunnel device for IPv4 VXLAN networking. Optional as Felix + determines this based on the existing IP pools. [Default: nil (unset)]' + type: boolean + vxlanMTU: + description: 'VXLANMTU is the MTU to set on the IPv4 VXLAN tunnel + device. See Configuring MTU [Default: 1410]' + type: integer + vxlanMTUV6: + description: 'VXLANMTUV6 is the MTU to set on the IPv6 VXLAN tunnel + device. See Configuring MTU [Default: 1390]' + type: integer + vxlanPort: + type: integer + vxlanVNI: + type: integer + windowsManageFirewallRules: + description: 'WindowsManageFirewallRules configures whether or not + Felix will program Windows Firewall rules. (to allow inbound access + to its own metrics ports) [Default: Disabled]' + enum: + - Enabled + - Disabled + type: string + wireguardEnabled: + description: 'WireguardEnabled controls whether Wireguard is enabled + for IPv4 (encapsulating IPv4 traffic over an IPv4 underlay network). + [Default: false]' + type: boolean + wireguardEnabledV6: + description: 'WireguardEnabledV6 controls whether Wireguard is enabled + for IPv6 (encapsulating IPv6 traffic over an IPv6 underlay network). + [Default: false]' + type: boolean + wireguardHostEncryptionEnabled: + description: 'WireguardHostEncryptionEnabled controls whether Wireguard + host-to-host encryption is enabled. [Default: false]' + type: boolean + wireguardInterfaceName: + description: 'WireguardInterfaceName specifies the name to use for + the IPv4 Wireguard interface. [Default: wireguard.cali]' + type: string + wireguardInterfaceNameV6: + description: 'WireguardInterfaceNameV6 specifies the name to use for + the IPv6 Wireguard interface. [Default: wg-v6.cali]' + type: string + wireguardKeepAlive: + description: 'WireguardKeepAlive controls Wireguard PersistentKeepalive + option. Set 0 to disable. [Default: 0]' + pattern: ^([0-9]+(\\.[0-9]+)?(ms|s|m|h))*$ + type: string + wireguardListeningPort: + description: 'WireguardListeningPort controls the listening port used + by IPv4 Wireguard. [Default: 51820]' + type: integer + wireguardListeningPortV6: + description: 'WireguardListeningPortV6 controls the listening port + used by IPv6 Wireguard. [Default: 51821]' + type: integer + wireguardMTU: + description: 'WireguardMTU controls the MTU on the IPv4 Wireguard + interface. See Configuring MTU [Default: 1440]' + type: integer + wireguardMTUV6: + description: 'WireguardMTUV6 controls the MTU on the IPv6 Wireguard + interface. See Configuring MTU [Default: 1420]' + type: integer + wireguardRoutingRulePriority: + description: 'WireguardRoutingRulePriority controls the priority value + to use for the Wireguard routing rule. [Default: 99]' + type: integer + wireguardThreadingEnabled: + description: 'WireguardThreadingEnabled controls whether Wireguard + has NAPI threading enabled. [Default: false]' + type: boolean + workloadSourceSpoofing: + description: WorkloadSourceSpoofing controls whether pods can use + the allowedSourcePrefixes annotation to send traffic with a source + IP address that is not theirs. This is disabled by default. When + set to "Any", pods can request any prefix. + pattern: ^(?i)(Disabled|Any)?$ + type: string + xdpEnabled: + description: 'XDPEnabled enables XDP acceleration for suitable untracked + incoming deny rules. [Default: true]' + type: boolean + xdpRefreshInterval: + description: 'XDPRefreshInterval is the period at which Felix re-checks + all XDP state to ensure that no other process has accidentally broken + Calico''s BPF maps or attached programs. Set to 0 to disable XDP + refresh. [Default: 90s]' + pattern: ^([0-9]+(\\.[0-9]+)?(ms|s|m|h))*$ + type: string + type: object + type: object + served: true + storage: true +status: + acceptedNames: + kind: "" + plural: "" + conditions: [] + storedVersions: [] +--- +# Source: calico/templates/kdd-crds.yaml +apiVersion: apiextensions.k8s.io/v1 +kind: CustomResourceDefinition +metadata: + name: globalnetworkpolicies.crd.projectcalico.org +spec: + group: crd.projectcalico.org + names: + kind: GlobalNetworkPolicy + listKind: GlobalNetworkPolicyList + plural: globalnetworkpolicies + singular: globalnetworkpolicy + preserveUnknownFields: false + scope: Cluster + versions: + - name: v1 + schema: + openAPIV3Schema: + properties: + apiVersion: + description: 'APIVersion defines the versioned schema of this representation + of an object. Servers should convert recognized schemas to the latest + internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources' + type: string + kind: + description: 'Kind is a string value representing the REST resource this + object represents. Servers may infer this from the endpoint the client + submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds' + type: string + metadata: + type: object + spec: + properties: + applyOnForward: + description: ApplyOnForward indicates to apply the rules in this policy + on forward traffic. + type: boolean + doNotTrack: + description: DoNotTrack indicates whether packets matched by the rules + in this policy should go through the data plane's connection tracking, + such as Linux conntrack. If True, the rules in this policy are + applied before any data plane connection tracking, and packets allowed + by this policy are marked as not to be tracked. + type: boolean + egress: + description: The ordered set of egress rules. Each rule contains + a set of packet match criteria and a corresponding action to apply. + items: + description: "A Rule encapsulates a set of match criteria and an + action. Both selector-based security Policy and security Profiles + reference rules - separated out as a list of rules for both ingress + and egress packet matching. \n Each positive match criteria has + a negated version, prefixed with \"Not\". All the match criteria + within a rule must be satisfied for a packet to match. A single + rule can contain the positive and negative version of a match + and both must be satisfied for the rule to match." + properties: + action: + type: string + destination: + description: Destination contains the match criteria that apply + to destination entity. + properties: + namespaceSelector: + description: "NamespaceSelector is an optional field that + contains a selector expression. Only traffic that originates + from (or terminates at) endpoints within the selected + namespaces will be matched. When both NamespaceSelector + and another selector are defined on the same rule, then + only workload endpoints that are matched by both selectors + will be selected by the rule. \n For NetworkPolicy, an + empty NamespaceSelector implies that the Selector is limited + to selecting only workload endpoints in the same namespace + as the NetworkPolicy. \n For NetworkPolicy, `global()` + NamespaceSelector implies that the Selector is limited + to selecting only GlobalNetworkSet or HostEndpoint. \n + For GlobalNetworkPolicy, an empty NamespaceSelector implies + the Selector applies to workload endpoints across all + namespaces." + type: string + nets: + description: Nets is an optional field that restricts the + rule to only apply to traffic that originates from (or + terminates at) IP addresses in any of the given subnets. + items: + type: string + type: array + notNets: + description: NotNets is the negated version of the Nets + field. + items: + type: string + type: array + notPorts: + description: NotPorts is the negated version of the Ports + field. Since only some protocols have ports, if any ports + are specified it requires the Protocol match in the Rule + to be set to "TCP" or "UDP". + items: + anyOf: + - type: integer + - type: string + pattern: ^.* + x-kubernetes-int-or-string: true + type: array + notSelector: + description: NotSelector is the negated version of the Selector + field. See Selector field for subtleties with negated + selectors. + type: string + ports: + description: "Ports is an optional field that restricts + the rule to only apply to traffic that has a source (destination) + port that matches one of these ranges/values. This value + is a list of integers or strings that represent ranges + of ports. \n Since only some protocols have ports, if + any ports are specified it requires the Protocol match + in the Rule to be set to \"TCP\" or \"UDP\"." + items: + anyOf: + - type: integer + - type: string + pattern: ^.* + x-kubernetes-int-or-string: true + type: array + selector: + description: "Selector is an optional field that contains + a selector expression (see Policy for sample syntax). + \ Only traffic that originates from (terminates at) endpoints + matching the selector will be matched. \n Note that: in + addition to the negated version of the Selector (see NotSelector + below), the selector expression syntax itself supports + negation. The two types of negation are subtly different. + One negates the set of matched endpoints, the other negates + the whole match: \n \tSelector = \"!has(my_label)\" matches + packets that are from other Calico-controlled \tendpoints + that do not have the label \"my_label\". \n \tNotSelector + = \"has(my_label)\" matches packets that are not from + Calico-controlled \tendpoints that do have the label \"my_label\". + \n The effect is that the latter will accept packets from + non-Calico sources whereas the former is limited to packets + from Calico-controlled endpoints." + type: string + serviceAccounts: + description: ServiceAccounts is an optional field that restricts + the rule to only apply to traffic that originates from + (or terminates at) a pod running as a matching service + account. + properties: + names: + description: Names is an optional field that restricts + the rule to only apply to traffic that originates + from (or terminates at) a pod running as a service + account whose name is in the list. + items: + type: string + type: array + selector: + description: Selector is an optional field that restricts + the rule to only apply to traffic that originates + from (or terminates at) a pod running as a service + account that matches the given label selector. If + both Names and Selector are specified then they are + AND'ed. + type: string + type: object + services: + description: "Services is an optional field that contains + options for matching Kubernetes Services. If specified, + only traffic that originates from or terminates at endpoints + within the selected service(s) will be matched, and only + to/from each endpoint's port. \n Services cannot be specified + on the same rule as Selector, NotSelector, NamespaceSelector, + Nets, NotNets or ServiceAccounts. \n Ports and NotPorts + can only be specified with Services on ingress rules." + properties: + name: + description: Name specifies the name of a Kubernetes + Service to match. + type: string + namespace: + description: Namespace specifies the namespace of the + given Service. If left empty, the rule will match + within this policy's namespace. + type: string + type: object + type: object + http: + description: HTTP contains match criteria that apply to HTTP + requests. + properties: + methods: + description: Methods is an optional field that restricts + the rule to apply only to HTTP requests that use one of + the listed HTTP Methods (e.g. GET, PUT, etc.) Multiple + methods are OR'd together. + items: + type: string + type: array + paths: + description: 'Paths is an optional field that restricts + the rule to apply to HTTP requests that use one of the + listed HTTP Paths. Multiple paths are OR''d together. + e.g: - exact: /foo - prefix: /bar NOTE: Each entry may + ONLY specify either a `exact` or a `prefix` match. The + validator will check for it.' + items: + description: 'HTTPPath specifies an HTTP path to match. + It may be either of the form: exact: : which matches + the path exactly or prefix: : which matches + the path prefix' + properties: + exact: + type: string + prefix: + type: string + type: object + type: array + type: object + icmp: + description: ICMP is an optional field that restricts the rule + to apply to a specific type and code of ICMP traffic. This + should only be specified if the Protocol field is set to "ICMP" + or "ICMPv6". + properties: + code: + description: Match on a specific ICMP code. If specified, + the Type value must also be specified. This is a technical + limitation imposed by the kernel's iptables firewall, + which Calico uses to enforce the rule. + type: integer + type: + description: Match on a specific ICMP type. For example + a value of 8 refers to ICMP Echo Request (i.e. pings). + type: integer + type: object + ipVersion: + description: IPVersion is an optional field that restricts the + rule to only match a specific IP version. + type: integer + metadata: + description: Metadata contains additional information for this + rule + properties: + annotations: + additionalProperties: + type: string + description: Annotations is a set of key value pairs that + give extra information about the rule + type: object + type: object + notICMP: + description: NotICMP is the negated version of the ICMP field. + properties: + code: + description: Match on a specific ICMP code. If specified, + the Type value must also be specified. This is a technical + limitation imposed by the kernel's iptables firewall, + which Calico uses to enforce the rule. + type: integer + type: + description: Match on a specific ICMP type. For example + a value of 8 refers to ICMP Echo Request (i.e. pings). + type: integer + type: object + notProtocol: + anyOf: + - type: integer + - type: string + description: NotProtocol is the negated version of the Protocol + field. + pattern: ^.* + x-kubernetes-int-or-string: true + protocol: + anyOf: + - type: integer + - type: string + description: "Protocol is an optional field that restricts the + rule to only apply to traffic of a specific IP protocol. Required + if any of the EntityRules contain Ports (because ports only + apply to certain protocols). \n Must be one of these string + values: \"TCP\", \"UDP\", \"ICMP\", \"ICMPv6\", \"SCTP\", + \"UDPLite\" or an integer in the range 1-255." + pattern: ^.* + x-kubernetes-int-or-string: true + source: + description: Source contains the match criteria that apply to + source entity. + properties: + namespaceSelector: + description: "NamespaceSelector is an optional field that + contains a selector expression. Only traffic that originates + from (or terminates at) endpoints within the selected + namespaces will be matched. When both NamespaceSelector + and another selector are defined on the same rule, then + only workload endpoints that are matched by both selectors + will be selected by the rule. \n For NetworkPolicy, an + empty NamespaceSelector implies that the Selector is limited + to selecting only workload endpoints in the same namespace + as the NetworkPolicy. \n For NetworkPolicy, `global()` + NamespaceSelector implies that the Selector is limited + to selecting only GlobalNetworkSet or HostEndpoint. \n + For GlobalNetworkPolicy, an empty NamespaceSelector implies + the Selector applies to workload endpoints across all + namespaces." + type: string + nets: + description: Nets is an optional field that restricts the + rule to only apply to traffic that originates from (or + terminates at) IP addresses in any of the given subnets. + items: + type: string + type: array + notNets: + description: NotNets is the negated version of the Nets + field. + items: + type: string + type: array + notPorts: + description: NotPorts is the negated version of the Ports + field. Since only some protocols have ports, if any ports + are specified it requires the Protocol match in the Rule + to be set to "TCP" or "UDP". + items: + anyOf: + - type: integer + - type: string + pattern: ^.* + x-kubernetes-int-or-string: true + type: array + notSelector: + description: NotSelector is the negated version of the Selector + field. See Selector field for subtleties with negated + selectors. + type: string + ports: + description: "Ports is an optional field that restricts + the rule to only apply to traffic that has a source (destination) + port that matches one of these ranges/values. This value + is a list of integers or strings that represent ranges + of ports. \n Since only some protocols have ports, if + any ports are specified it requires the Protocol match + in the Rule to be set to \"TCP\" or \"UDP\"." + items: + anyOf: + - type: integer + - type: string + pattern: ^.* + x-kubernetes-int-or-string: true + type: array + selector: + description: "Selector is an optional field that contains + a selector expression (see Policy for sample syntax). + \ Only traffic that originates from (terminates at) endpoints + matching the selector will be matched. \n Note that: in + addition to the negated version of the Selector (see NotSelector + below), the selector expression syntax itself supports + negation. The two types of negation are subtly different. + One negates the set of matched endpoints, the other negates + the whole match: \n \tSelector = \"!has(my_label)\" matches + packets that are from other Calico-controlled \tendpoints + that do not have the label \"my_label\". \n \tNotSelector + = \"has(my_label)\" matches packets that are not from + Calico-controlled \tendpoints that do have the label \"my_label\". + \n The effect is that the latter will accept packets from + non-Calico sources whereas the former is limited to packets + from Calico-controlled endpoints." + type: string + serviceAccounts: + description: ServiceAccounts is an optional field that restricts + the rule to only apply to traffic that originates from + (or terminates at) a pod running as a matching service + account. + properties: + names: + description: Names is an optional field that restricts + the rule to only apply to traffic that originates + from (or terminates at) a pod running as a service + account whose name is in the list. + items: + type: string + type: array + selector: + description: Selector is an optional field that restricts + the rule to only apply to traffic that originates + from (or terminates at) a pod running as a service + account that matches the given label selector. If + both Names and Selector are specified then they are + AND'ed. + type: string + type: object + services: + description: "Services is an optional field that contains + options for matching Kubernetes Services. If specified, + only traffic that originates from or terminates at endpoints + within the selected service(s) will be matched, and only + to/from each endpoint's port. \n Services cannot be specified + on the same rule as Selector, NotSelector, NamespaceSelector, + Nets, NotNets or ServiceAccounts. \n Ports and NotPorts + can only be specified with Services on ingress rules." + properties: + name: + description: Name specifies the name of a Kubernetes + Service to match. + type: string + namespace: + description: Namespace specifies the namespace of the + given Service. If left empty, the rule will match + within this policy's namespace. + type: string + type: object + type: object + required: + - action + type: object + type: array + ingress: + description: The ordered set of ingress rules. Each rule contains + a set of packet match criteria and a corresponding action to apply. + items: + description: "A Rule encapsulates a set of match criteria and an + action. Both selector-based security Policy and security Profiles + reference rules - separated out as a list of rules for both ingress + and egress packet matching. \n Each positive match criteria has + a negated version, prefixed with \"Not\". All the match criteria + within a rule must be satisfied for a packet to match. A single + rule can contain the positive and negative version of a match + and both must be satisfied for the rule to match." + properties: + action: + type: string + destination: + description: Destination contains the match criteria that apply + to destination entity. + properties: + namespaceSelector: + description: "NamespaceSelector is an optional field that + contains a selector expression. Only traffic that originates + from (or terminates at) endpoints within the selected + namespaces will be matched. When both NamespaceSelector + and another selector are defined on the same rule, then + only workload endpoints that are matched by both selectors + will be selected by the rule. \n For NetworkPolicy, an + empty NamespaceSelector implies that the Selector is limited + to selecting only workload endpoints in the same namespace + as the NetworkPolicy. \n For NetworkPolicy, `global()` + NamespaceSelector implies that the Selector is limited + to selecting only GlobalNetworkSet or HostEndpoint. \n + For GlobalNetworkPolicy, an empty NamespaceSelector implies + the Selector applies to workload endpoints across all + namespaces." + type: string + nets: + description: Nets is an optional field that restricts the + rule to only apply to traffic that originates from (or + terminates at) IP addresses in any of the given subnets. + items: + type: string + type: array + notNets: + description: NotNets is the negated version of the Nets + field. + items: + type: string + type: array + notPorts: + description: NotPorts is the negated version of the Ports + field. Since only some protocols have ports, if any ports + are specified it requires the Protocol match in the Rule + to be set to "TCP" or "UDP". + items: + anyOf: + - type: integer + - type: string + pattern: ^.* + x-kubernetes-int-or-string: true + type: array + notSelector: + description: NotSelector is the negated version of the Selector + field. See Selector field for subtleties with negated + selectors. + type: string + ports: + description: "Ports is an optional field that restricts + the rule to only apply to traffic that has a source (destination) + port that matches one of these ranges/values. This value + is a list of integers or strings that represent ranges + of ports. \n Since only some protocols have ports, if + any ports are specified it requires the Protocol match + in the Rule to be set to \"TCP\" or \"UDP\"." + items: + anyOf: + - type: integer + - type: string + pattern: ^.* + x-kubernetes-int-or-string: true + type: array + selector: + description: "Selector is an optional field that contains + a selector expression (see Policy for sample syntax). + \ Only traffic that originates from (terminates at) endpoints + matching the selector will be matched. \n Note that: in + addition to the negated version of the Selector (see NotSelector + below), the selector expression syntax itself supports + negation. The two types of negation are subtly different. + One negates the set of matched endpoints, the other negates + the whole match: \n \tSelector = \"!has(my_label)\" matches + packets that are from other Calico-controlled \tendpoints + that do not have the label \"my_label\". \n \tNotSelector + = \"has(my_label)\" matches packets that are not from + Calico-controlled \tendpoints that do have the label \"my_label\". + \n The effect is that the latter will accept packets from + non-Calico sources whereas the former is limited to packets + from Calico-controlled endpoints." + type: string + serviceAccounts: + description: ServiceAccounts is an optional field that restricts + the rule to only apply to traffic that originates from + (or terminates at) a pod running as a matching service + account. + properties: + names: + description: Names is an optional field that restricts + the rule to only apply to traffic that originates + from (or terminates at) a pod running as a service + account whose name is in the list. + items: + type: string + type: array + selector: + description: Selector is an optional field that restricts + the rule to only apply to traffic that originates + from (or terminates at) a pod running as a service + account that matches the given label selector. If + both Names and Selector are specified then they are + AND'ed. + type: string + type: object + services: + description: "Services is an optional field that contains + options for matching Kubernetes Services. If specified, + only traffic that originates from or terminates at endpoints + within the selected service(s) will be matched, and only + to/from each endpoint's port. \n Services cannot be specified + on the same rule as Selector, NotSelector, NamespaceSelector, + Nets, NotNets or ServiceAccounts. \n Ports and NotPorts + can only be specified with Services on ingress rules." + properties: + name: + description: Name specifies the name of a Kubernetes + Service to match. + type: string + namespace: + description: Namespace specifies the namespace of the + given Service. If left empty, the rule will match + within this policy's namespace. + type: string + type: object + type: object + http: + description: HTTP contains match criteria that apply to HTTP + requests. + properties: + methods: + description: Methods is an optional field that restricts + the rule to apply only to HTTP requests that use one of + the listed HTTP Methods (e.g. GET, PUT, etc.) Multiple + methods are OR'd together. + items: + type: string + type: array + paths: + description: 'Paths is an optional field that restricts + the rule to apply to HTTP requests that use one of the + listed HTTP Paths. Multiple paths are OR''d together. + e.g: - exact: /foo - prefix: /bar NOTE: Each entry may + ONLY specify either a `exact` or a `prefix` match. The + validator will check for it.' + items: + description: 'HTTPPath specifies an HTTP path to match. + It may be either of the form: exact: : which matches + the path exactly or prefix: : which matches + the path prefix' + properties: + exact: + type: string + prefix: + type: string + type: object + type: array + type: object + icmp: + description: ICMP is an optional field that restricts the rule + to apply to a specific type and code of ICMP traffic. This + should only be specified if the Protocol field is set to "ICMP" + or "ICMPv6". + properties: + code: + description: Match on a specific ICMP code. If specified, + the Type value must also be specified. This is a technical + limitation imposed by the kernel's iptables firewall, + which Calico uses to enforce the rule. + type: integer + type: + description: Match on a specific ICMP type. For example + a value of 8 refers to ICMP Echo Request (i.e. pings). + type: integer + type: object + ipVersion: + description: IPVersion is an optional field that restricts the + rule to only match a specific IP version. + type: integer + metadata: + description: Metadata contains additional information for this + rule + properties: + annotations: + additionalProperties: + type: string + description: Annotations is a set of key value pairs that + give extra information about the rule + type: object + type: object + notICMP: + description: NotICMP is the negated version of the ICMP field. + properties: + code: + description: Match on a specific ICMP code. If specified, + the Type value must also be specified. This is a technical + limitation imposed by the kernel's iptables firewall, + which Calico uses to enforce the rule. + type: integer + type: + description: Match on a specific ICMP type. For example + a value of 8 refers to ICMP Echo Request (i.e. pings). + type: integer + type: object + notProtocol: + anyOf: + - type: integer + - type: string + description: NotProtocol is the negated version of the Protocol + field. + pattern: ^.* + x-kubernetes-int-or-string: true + protocol: + anyOf: + - type: integer + - type: string + description: "Protocol is an optional field that restricts the + rule to only apply to traffic of a specific IP protocol. Required + if any of the EntityRules contain Ports (because ports only + apply to certain protocols). \n Must be one of these string + values: \"TCP\", \"UDP\", \"ICMP\", \"ICMPv6\", \"SCTP\", + \"UDPLite\" or an integer in the range 1-255." + pattern: ^.* + x-kubernetes-int-or-string: true + source: + description: Source contains the match criteria that apply to + source entity. + properties: + namespaceSelector: + description: "NamespaceSelector is an optional field that + contains a selector expression. Only traffic that originates + from (or terminates at) endpoints within the selected + namespaces will be matched. When both NamespaceSelector + and another selector are defined on the same rule, then + only workload endpoints that are matched by both selectors + will be selected by the rule. \n For NetworkPolicy, an + empty NamespaceSelector implies that the Selector is limited + to selecting only workload endpoints in the same namespace + as the NetworkPolicy. \n For NetworkPolicy, `global()` + NamespaceSelector implies that the Selector is limited + to selecting only GlobalNetworkSet or HostEndpoint. \n + For GlobalNetworkPolicy, an empty NamespaceSelector implies + the Selector applies to workload endpoints across all + namespaces." + type: string + nets: + description: Nets is an optional field that restricts the + rule to only apply to traffic that originates from (or + terminates at) IP addresses in any of the given subnets. + items: + type: string + type: array + notNets: + description: NotNets is the negated version of the Nets + field. + items: + type: string + type: array + notPorts: + description: NotPorts is the negated version of the Ports + field. Since only some protocols have ports, if any ports + are specified it requires the Protocol match in the Rule + to be set to "TCP" or "UDP". + items: + anyOf: + - type: integer + - type: string + pattern: ^.* + x-kubernetes-int-or-string: true + type: array + notSelector: + description: NotSelector is the negated version of the Selector + field. See Selector field for subtleties with negated + selectors. + type: string + ports: + description: "Ports is an optional field that restricts + the rule to only apply to traffic that has a source (destination) + port that matches one of these ranges/values. This value + is a list of integers or strings that represent ranges + of ports. \n Since only some protocols have ports, if + any ports are specified it requires the Protocol match + in the Rule to be set to \"TCP\" or \"UDP\"." + items: + anyOf: + - type: integer + - type: string + pattern: ^.* + x-kubernetes-int-or-string: true + type: array + selector: + description: "Selector is an optional field that contains + a selector expression (see Policy for sample syntax). + \ Only traffic that originates from (terminates at) endpoints + matching the selector will be matched. \n Note that: in + addition to the negated version of the Selector (see NotSelector + below), the selector expression syntax itself supports + negation. The two types of negation are subtly different. + One negates the set of matched endpoints, the other negates + the whole match: \n \tSelector = \"!has(my_label)\" matches + packets that are from other Calico-controlled \tendpoints + that do not have the label \"my_label\". \n \tNotSelector + = \"has(my_label)\" matches packets that are not from + Calico-controlled \tendpoints that do have the label \"my_label\". + \n The effect is that the latter will accept packets from + non-Calico sources whereas the former is limited to packets + from Calico-controlled endpoints." + type: string + serviceAccounts: + description: ServiceAccounts is an optional field that restricts + the rule to only apply to traffic that originates from + (or terminates at) a pod running as a matching service + account. + properties: + names: + description: Names is an optional field that restricts + the rule to only apply to traffic that originates + from (or terminates at) a pod running as a service + account whose name is in the list. + items: + type: string + type: array + selector: + description: Selector is an optional field that restricts + the rule to only apply to traffic that originates + from (or terminates at) a pod running as a service + account that matches the given label selector. If + both Names and Selector are specified then they are + AND'ed. + type: string + type: object + services: + description: "Services is an optional field that contains + options for matching Kubernetes Services. If specified, + only traffic that originates from or terminates at endpoints + within the selected service(s) will be matched, and only + to/from each endpoint's port. \n Services cannot be specified + on the same rule as Selector, NotSelector, NamespaceSelector, + Nets, NotNets or ServiceAccounts. \n Ports and NotPorts + can only be specified with Services on ingress rules." + properties: + name: + description: Name specifies the name of a Kubernetes + Service to match. + type: string + namespace: + description: Namespace specifies the namespace of the + given Service. If left empty, the rule will match + within this policy's namespace. + type: string + type: object + type: object + required: + - action + type: object + type: array + namespaceSelector: + description: NamespaceSelector is an optional field for an expression + used to select a pod based on namespaces. + type: string + order: + description: Order is an optional field that specifies the order in + which the policy is applied. Policies with higher "order" are applied + after those with lower order within the same tier. If the order + is omitted, it may be considered to be "infinite" - i.e. the policy + will be applied last. Policies with identical order will be applied + in alphanumerical order based on the Policy "Name" within the tier. + type: number + performanceHints: + description: "PerformanceHints contains a list of hints to Calico's + policy engine to help process the policy more efficiently. Hints + never change the enforcement behaviour of the policy. \n Currently, + the only available hint is \"AssumeNeededOnEveryNode\". When that + hint is set on a policy, Felix will act as if the policy matches + a local endpoint even if it does not. This is useful for \"preloading\" + any large static policies that are known to be used on every node. + If the policy is _not_ used on a particular node then the work done + to preload the policy (and to maintain it) is wasted." + items: + type: string + type: array + preDNAT: + description: PreDNAT indicates to apply the rules in this policy before + any DNAT. + type: boolean + selector: + description: "The selector is an expression used to pick out the endpoints + that the policy should be applied to. \n Selector expressions follow + this syntax: \n \tlabel == \"string_literal\" -> comparison, e.g. + my_label == \"foo bar\" \tlabel != \"string_literal\" -> not + equal; also matches if label is not present \tlabel in { \"a\", + \"b\", \"c\", ... } -> true if the value of label X is one of + \"a\", \"b\", \"c\" \tlabel not in { \"a\", \"b\", \"c\", ... } + \ -> true if the value of label X is not one of \"a\", \"b\", \"c\" + \thas(label_name) -> True if that label is present \t! expr -> + negation of expr \texpr && expr -> Short-circuit and \texpr || + expr -> Short-circuit or \t( expr ) -> parens for grouping \tall() + or the empty selector -> matches all endpoints. \n Label names are + allowed to contain alphanumerics, -, _ and /. String literals are + more permissive but they do not support escape characters. \n Examples + (with made-up labels): \n \ttype == \"webserver\" && deployment + == \"prod\" \ttype in {\"frontend\", \"backend\"} \tdeployment != + \"dev\" \t! has(label_name)" + type: string + serviceAccountSelector: + description: ServiceAccountSelector is an optional field for an expression + used to select a pod based on service accounts. + type: string + tier: + description: The name of the tier that this policy belongs to. If + this is omitted, the default tier (name is "default") is assumed. The + specified tier must exist in order to create security policies within + the tier, the "default" tier is created automatically if it does + not exist, this means for deployments requiring only a single Tier, + the tier name may be omitted on all policy management requests. + type: string + types: + description: "Types indicates whether this policy applies to ingress, + or to egress, or to both. When not explicitly specified (and so + the value on creation is empty or nil), Calico defaults Types according + to what Ingress and Egress rules are present in the policy. The + default is: \n - [ PolicyTypeIngress ], if there are no Egress rules + (including the case where there are also no Ingress rules) \n + - [ PolicyTypeEgress ], if there are Egress rules but no Ingress + rules \n - [ PolicyTypeIngress, PolicyTypeEgress ], if there are + both Ingress and Egress rules. \n When the policy is read back again, + Types will always be one of these values, never empty or nil." + items: + description: PolicyType enumerates the possible values of the PolicySpec + Types field. + type: string + type: array + type: object + type: object + served: true + storage: true +status: + acceptedNames: + kind: "" + plural: "" + conditions: [] + storedVersions: [] +--- +# Source: calico/templates/kdd-crds.yaml +apiVersion: apiextensions.k8s.io/v1 +kind: CustomResourceDefinition +metadata: + name: globalnetworksets.crd.projectcalico.org +spec: + group: crd.projectcalico.org + names: + kind: GlobalNetworkSet + listKind: GlobalNetworkSetList + plural: globalnetworksets + singular: globalnetworkset + preserveUnknownFields: false + scope: Cluster + versions: + - name: v1 + schema: + openAPIV3Schema: + description: GlobalNetworkSet contains a set of arbitrary IP sub-networks/CIDRs + that share labels to allow rules to refer to them via selectors. The labels + of GlobalNetworkSet are not namespaced. + properties: + apiVersion: + description: 'APIVersion defines the versioned schema of this representation + of an object. Servers should convert recognized schemas to the latest + internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources' + type: string + kind: + description: 'Kind is a string value representing the REST resource this + object represents. Servers may infer this from the endpoint the client + submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds' + type: string + metadata: + type: object + spec: + description: GlobalNetworkSetSpec contains the specification for a NetworkSet + resource. + properties: + nets: + description: The list of IP networks that belong to this set. + items: + type: string + type: array + type: object + type: object + served: true + storage: true +status: + acceptedNames: + kind: "" + plural: "" + conditions: [] + storedVersions: [] +--- +# Source: calico/templates/kdd-crds.yaml +apiVersion: apiextensions.k8s.io/v1 +kind: CustomResourceDefinition +metadata: + name: hostendpoints.crd.projectcalico.org +spec: + group: crd.projectcalico.org + names: + kind: HostEndpoint + listKind: HostEndpointList + plural: hostendpoints + singular: hostendpoint + preserveUnknownFields: false + scope: Cluster + versions: + - name: v1 + schema: + openAPIV3Schema: + properties: + apiVersion: + description: 'APIVersion defines the versioned schema of this representation + of an object. Servers should convert recognized schemas to the latest + internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources' + type: string + kind: + description: 'Kind is a string value representing the REST resource this + object represents. Servers may infer this from the endpoint the client + submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds' + type: string + metadata: + type: object + spec: + description: HostEndpointSpec contains the specification for a HostEndpoint + resource. + properties: + expectedIPs: + description: "The expected IP addresses (IPv4 and IPv6) of the endpoint. + If \"InterfaceName\" is not present, Calico will look for an interface + matching any of the IPs in the list and apply policy to that. Note: + \tWhen using the selector match criteria in an ingress or egress + security Policy \tor Profile, Calico converts the selector into + a set of IP addresses. For host \tendpoints, the ExpectedIPs field + is used for that purpose. (If only the interface \tname is specified, + Calico does not learn the IPs of the interface for use in match + \tcriteria.)" + items: + type: string + type: array + interfaceName: + description: "Either \"*\", or the name of a specific Linux interface + to apply policy to; or empty. \"*\" indicates that this HostEndpoint + governs all traffic to, from or through the default network namespace + of the host named by the \"Node\" field; entering and leaving that + namespace via any interface, including those from/to non-host-networked + local workloads. \n If InterfaceName is not \"*\", this HostEndpoint + only governs traffic that enters or leaves the host through the + specific interface named by InterfaceName, or - when InterfaceName + is empty - through the specific interface that has one of the IPs + in ExpectedIPs. Therefore, when InterfaceName is empty, at least + one expected IP must be specified. Only external interfaces (such + as \"eth0\") are supported here; it isn't possible for a HostEndpoint + to protect traffic through a specific local workload interface. + \n Note: Only some kinds of policy are implemented for \"*\" HostEndpoints; + initially just pre-DNAT policy. Please check Calico documentation + for the latest position." + type: string + node: + description: The node name identifying the Calico node instance. + type: string + ports: + description: Ports contains the endpoint's named ports, which may + be referenced in security policy rules. + items: + properties: + name: + type: string + port: + type: integer + protocol: + anyOf: + - type: integer + - type: string + pattern: ^.* + x-kubernetes-int-or-string: true + required: + - name + - port + - protocol + type: object + type: array + profiles: + description: A list of identifiers of security Profile objects that + apply to this endpoint. Each profile is applied in the order that + they appear in this list. Profile rules are applied after the selector-based + security policy. + items: + type: string + type: array + type: object + type: object + served: true + storage: true +status: + acceptedNames: + kind: "" + plural: "" + conditions: [] + storedVersions: [] +--- +# Source: calico/templates/kdd-crds.yaml +apiVersion: apiextensions.k8s.io/v1 +kind: CustomResourceDefinition +metadata: + name: ipamblocks.crd.projectcalico.org +spec: + group: crd.projectcalico.org + names: + kind: IPAMBlock + listKind: IPAMBlockList + plural: ipamblocks + singular: ipamblock + preserveUnknownFields: false + scope: Cluster + versions: + - name: v1 + schema: + openAPIV3Schema: + properties: + apiVersion: + description: 'APIVersion defines the versioned schema of this representation + of an object. Servers should convert recognized schemas to the latest + internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources' + type: string + kind: + description: 'Kind is a string value representing the REST resource this + object represents. Servers may infer this from the endpoint the client + submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds' + type: string + metadata: + type: object + spec: + description: IPAMBlockSpec contains the specification for an IPAMBlock + resource. + properties: + affinity: + description: Affinity of the block, if this block has one. If set, + it will be of the form "host:". If not set, this block + is not affine to a host. + type: string + allocations: + description: Array of allocations in-use within this block. nil entries + mean the allocation is free. For non-nil entries at index i, the + index is the ordinal of the allocation within this block and the + value is the index of the associated attributes in the Attributes + array. + items: + type: integer + # TODO: This nullable is manually added in. We should update controller-gen + # to handle []*int properly itself. + nullable: true + type: array + attributes: + description: Attributes is an array of arbitrary metadata associated + with allocations in the block. To find attributes for a given allocation, + use the value of the allocation's entry in the Allocations array + as the index of the element in this array. + items: + properties: + handle_id: + type: string + secondary: + additionalProperties: + type: string + type: object + type: object + type: array + cidr: + description: The block's CIDR. + type: string + deleted: + description: Deleted is an internal boolean used to workaround a limitation + in the Kubernetes API whereby deletion will not return a conflict + error if the block has been updated. It should not be set manually. + type: boolean + sequenceNumber: + default: 0 + description: We store a sequence number that is updated each time + the block is written. Each allocation will also store the sequence + number of the block at the time of its creation. When releasing + an IP, passing the sequence number associated with the allocation + allows us to protect against a race condition and ensure the IP + hasn't been released and re-allocated since the release request. + format: int64 + type: integer + sequenceNumberForAllocation: + additionalProperties: + format: int64 + type: integer + description: Map of allocated ordinal within the block to sequence + number of the block at the time of allocation. Kubernetes does not + allow numerical keys for maps, so the key is cast to a string. + type: object + strictAffinity: + description: StrictAffinity on the IPAMBlock is deprecated and no + longer used by the code. Use IPAMConfig StrictAffinity instead. + type: boolean + unallocated: + description: Unallocated is an ordered list of allocations which are + free in the block. + items: + type: integer + type: array + required: + - allocations + - attributes + - cidr + - strictAffinity + - unallocated + type: object + type: object + served: true + storage: true +status: + acceptedNames: + kind: "" + plural: "" + conditions: [] + storedVersions: [] +--- +# Source: calico/templates/kdd-crds.yaml +apiVersion: apiextensions.k8s.io/v1 +kind: CustomResourceDefinition +metadata: + name: ipamconfigs.crd.projectcalico.org +spec: + group: crd.projectcalico.org + names: + kind: IPAMConfig + listKind: IPAMConfigList + plural: ipamconfigs + singular: ipamconfig + preserveUnknownFields: false + scope: Cluster + versions: + - name: v1 + schema: + openAPIV3Schema: + properties: + apiVersion: + description: 'APIVersion defines the versioned schema of this representation + of an object. Servers should convert recognized schemas to the latest + internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources' + type: string + kind: + description: 'Kind is a string value representing the REST resource this + object represents. Servers may infer this from the endpoint the client + submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds' + type: string + metadata: + type: object + spec: + description: IPAMConfigSpec contains the specification for an IPAMConfig + resource. + properties: + autoAllocateBlocks: + type: boolean + maxBlocksPerHost: + description: MaxBlocksPerHost, if non-zero, is the max number of blocks + that can be affine to each host. + maximum: 2147483647 + minimum: 0 + type: integer + strictAffinity: + type: boolean + required: + - autoAllocateBlocks + - strictAffinity + type: object + type: object + served: true + storage: true +status: + acceptedNames: + kind: "" + plural: "" + conditions: [] + storedVersions: [] +--- +# Source: calico/templates/kdd-crds.yaml +apiVersion: apiextensions.k8s.io/v1 +kind: CustomResourceDefinition +metadata: + name: ipamhandles.crd.projectcalico.org +spec: + group: crd.projectcalico.org + names: + kind: IPAMHandle + listKind: IPAMHandleList + plural: ipamhandles + singular: ipamhandle + preserveUnknownFields: false + scope: Cluster + versions: + - name: v1 + schema: + openAPIV3Schema: + properties: + apiVersion: + description: 'APIVersion defines the versioned schema of this representation + of an object. Servers should convert recognized schemas to the latest + internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources' + type: string + kind: + description: 'Kind is a string value representing the REST resource this + object represents. Servers may infer this from the endpoint the client + submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds' + type: string + metadata: + type: object + spec: + description: IPAMHandleSpec contains the specification for an IPAMHandle + resource. + properties: + block: + additionalProperties: + type: integer + type: object + deleted: + type: boolean + handleID: + type: string + required: + - block + - handleID + type: object + type: object + served: true + storage: true +status: + acceptedNames: + kind: "" + plural: "" + conditions: [] + storedVersions: [] +--- +# Source: calico/templates/kdd-crds.yaml +apiVersion: apiextensions.k8s.io/v1 +kind: CustomResourceDefinition +metadata: + name: ippools.crd.projectcalico.org +spec: + group: crd.projectcalico.org + names: + kind: IPPool + listKind: IPPoolList + plural: ippools + singular: ippool + preserveUnknownFields: false + scope: Cluster + versions: + - name: v1 + schema: + openAPIV3Schema: + properties: + apiVersion: + description: 'APIVersion defines the versioned schema of this representation + of an object. Servers should convert recognized schemas to the latest + internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources' + type: string + kind: + description: 'Kind is a string value representing the REST resource this + object represents. Servers may infer this from the endpoint the client + submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds' + type: string + metadata: + type: object + spec: + description: IPPoolSpec contains the specification for an IPPool resource. + properties: + allowedUses: + description: AllowedUse controls what the IP pool will be used for. If + not specified or empty, defaults to ["Tunnel", "Workload"] for back-compatibility + items: + type: string + type: array + blockSize: + description: The block size to use for IP address assignments from + this pool. Defaults to 26 for IPv4 and 122 for IPv6. + type: integer + cidr: + description: The pool CIDR. + type: string + disableBGPExport: + description: 'Disable exporting routes from this IP Pool''s CIDR over + BGP. [Default: false]' + type: boolean + disabled: + description: When disabled is true, Calico IPAM will not assign addresses + from this pool. + type: boolean + ipip: + description: 'Deprecated: this field is only used for APIv1 backwards + compatibility. Setting this field is not allowed, this field is + for internal use only.' + properties: + enabled: + description: When enabled is true, ipip tunneling will be used + to deliver packets to destinations within this pool. + type: boolean + mode: + description: The IPIP mode. This can be one of "always" or "cross-subnet". A + mode of "always" will also use IPIP tunneling for routing to + destination IP addresses within this pool. A mode of "cross-subnet" + will only use IPIP tunneling when the destination node is on + a different subnet to the originating node. The default value + (if not specified) is "always". + type: string + type: object + ipipMode: + description: Contains configuration for IPIP tunneling for this pool. + If not specified, then this is defaulted to "Never" (i.e. IPIP tunneling + is disabled). + type: string + nat-outgoing: + description: 'Deprecated: this field is only used for APIv1 backwards + compatibility. Setting this field is not allowed, this field is + for internal use only.' + type: boolean + natOutgoing: + description: When natOutgoing is true, packets sent from Calico networked + containers in this pool to destinations outside of this pool will + be masqueraded. + type: boolean + nodeSelector: + description: Allows IPPool to allocate for a specific node by label + selector. + type: string + vxlanMode: + description: Contains configuration for VXLAN tunneling for this pool. + If not specified, then this is defaulted to "Never" (i.e. VXLAN + tunneling is disabled). + type: string + required: + - cidr + type: object + type: object + served: true + storage: true +status: + acceptedNames: + kind: "" + plural: "" + conditions: [] + storedVersions: [] +--- +# Source: calico/templates/kdd-crds.yaml +apiVersion: apiextensions.k8s.io/v1 +kind: CustomResourceDefinition +metadata: + annotations: + controller-gen.kubebuilder.io/version: (devel) + creationTimestamp: null + name: ipreservations.crd.projectcalico.org +spec: + group: crd.projectcalico.org + names: + kind: IPReservation + listKind: IPReservationList + plural: ipreservations + singular: ipreservation + preserveUnknownFields: false + scope: Cluster + versions: + - name: v1 + schema: + openAPIV3Schema: + properties: + apiVersion: + description: 'APIVersion defines the versioned schema of this representation + of an object. Servers should convert recognized schemas to the latest + internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources' + type: string + kind: + description: 'Kind is a string value representing the REST resource this + object represents. Servers may infer this from the endpoint the client + submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds' + type: string + metadata: + type: object + spec: + description: IPReservationSpec contains the specification for an IPReservation + resource. + properties: + reservedCIDRs: + description: ReservedCIDRs is a list of CIDRs and/or IP addresses + that Calico IPAM will exclude from new allocations. + items: + type: string + type: array + type: object + type: object + served: true + storage: true +status: + acceptedNames: + kind: "" + plural: "" + conditions: [] + storedVersions: [] +--- +# Source: calico/templates/kdd-crds.yaml +apiVersion: apiextensions.k8s.io/v1 +kind: CustomResourceDefinition +metadata: + name: kubecontrollersconfigurations.crd.projectcalico.org +spec: + group: crd.projectcalico.org + names: + kind: KubeControllersConfiguration + listKind: KubeControllersConfigurationList + plural: kubecontrollersconfigurations + singular: kubecontrollersconfiguration + preserveUnknownFields: false + scope: Cluster + versions: + - name: v1 + schema: + openAPIV3Schema: + properties: + apiVersion: + description: 'APIVersion defines the versioned schema of this representation + of an object. Servers should convert recognized schemas to the latest + internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources' + type: string + kind: + description: 'Kind is a string value representing the REST resource this + object represents. Servers may infer this from the endpoint the client + submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds' + type: string + metadata: + type: object + spec: + description: KubeControllersConfigurationSpec contains the values of the + Kubernetes controllers configuration. + properties: + controllers: + description: Controllers enables and configures individual Kubernetes + controllers + properties: + namespace: + description: Namespace enables and configures the namespace controller. + Enabled by default, set to nil to disable. + properties: + reconcilerPeriod: + description: 'ReconcilerPeriod is the period to perform reconciliation + with the Calico datastore. [Default: 5m]' + type: string + type: object + node: + description: Node enables and configures the node controller. + Enabled by default, set to nil to disable. + properties: + hostEndpoint: + description: HostEndpoint controls syncing nodes to host endpoints. + Disabled by default, set to nil to disable. + properties: + autoCreate: + description: 'AutoCreate enables automatic creation of + host endpoints for every node. [Default: Disabled]' + type: string + type: object + leakGracePeriod: + description: 'LeakGracePeriod is the period used by the controller + to determine if an IP address has been leaked. Set to 0 + to disable IP garbage collection. [Default: 15m]' + type: string + reconcilerPeriod: + description: 'ReconcilerPeriod is the period to perform reconciliation + with the Calico datastore. [Default: 5m]' + type: string + syncLabels: + description: 'SyncLabels controls whether to copy Kubernetes + node labels to Calico nodes. [Default: Enabled]' + type: string + type: object + policy: + description: Policy enables and configures the policy controller. + Enabled by default, set to nil to disable. + properties: + reconcilerPeriod: + description: 'ReconcilerPeriod is the period to perform reconciliation + with the Calico datastore. [Default: 5m]' + type: string + type: object + serviceAccount: + description: ServiceAccount enables and configures the service + account controller. Enabled by default, set to nil to disable. + properties: + reconcilerPeriod: + description: 'ReconcilerPeriod is the period to perform reconciliation + with the Calico datastore. [Default: 5m]' + type: string + type: object + workloadEndpoint: + description: WorkloadEndpoint enables and configures the workload + endpoint controller. Enabled by default, set to nil to disable. + properties: + reconcilerPeriod: + description: 'ReconcilerPeriod is the period to perform reconciliation + with the Calico datastore. [Default: 5m]' + type: string + type: object + type: object + debugProfilePort: + description: DebugProfilePort configures the port to serve memory + and cpu profiles on. If not specified, profiling is disabled. + format: int32 + type: integer + etcdV3CompactionPeriod: + description: 'EtcdV3CompactionPeriod is the period between etcdv3 + compaction requests. Set to 0 to disable. [Default: 10m]' + type: string + healthChecks: + description: 'HealthChecks enables or disables support for health + checks [Default: Enabled]' + type: string + logSeverityScreen: + description: 'LogSeverityScreen is the log severity above which logs + are sent to the stdout. [Default: Info]' + type: string + prometheusMetricsPort: + description: 'PrometheusMetricsPort is the TCP port that the Prometheus + metrics server should bind to. Set to 0 to disable. [Default: 9094]' + type: integer + required: + - controllers + type: object + status: + description: KubeControllersConfigurationStatus represents the status + of the configuration. It's useful for admins to be able to see the actual + config that was applied, which can be modified by environment variables + on the kube-controllers process. + properties: + environmentVars: + additionalProperties: + type: string + description: EnvironmentVars contains the environment variables on + the kube-controllers that influenced the RunningConfig. + type: object + runningConfig: + description: RunningConfig contains the effective config that is running + in the kube-controllers pod, after merging the API resource with + any environment variables. + properties: + controllers: + description: Controllers enables and configures individual Kubernetes + controllers + properties: + namespace: + description: Namespace enables and configures the namespace + controller. Enabled by default, set to nil to disable. + properties: + reconcilerPeriod: + description: 'ReconcilerPeriod is the period to perform + reconciliation with the Calico datastore. [Default: + 5m]' + type: string + type: object + node: + description: Node enables and configures the node controller. + Enabled by default, set to nil to disable. + properties: + hostEndpoint: + description: HostEndpoint controls syncing nodes to host + endpoints. Disabled by default, set to nil to disable. + properties: + autoCreate: + description: 'AutoCreate enables automatic creation + of host endpoints for every node. [Default: Disabled]' + type: string + type: object + leakGracePeriod: + description: 'LeakGracePeriod is the period used by the + controller to determine if an IP address has been leaked. + Set to 0 to disable IP garbage collection. [Default: + 15m]' + type: string + reconcilerPeriod: + description: 'ReconcilerPeriod is the period to perform + reconciliation with the Calico datastore. [Default: + 5m]' + type: string + syncLabels: + description: 'SyncLabels controls whether to copy Kubernetes + node labels to Calico nodes. [Default: Enabled]' + type: string + type: object + policy: + description: Policy enables and configures the policy controller. + Enabled by default, set to nil to disable. + properties: + reconcilerPeriod: + description: 'ReconcilerPeriod is the period to perform + reconciliation with the Calico datastore. [Default: + 5m]' + type: string + type: object + serviceAccount: + description: ServiceAccount enables and configures the service + account controller. Enabled by default, set to nil to disable. + properties: + reconcilerPeriod: + description: 'ReconcilerPeriod is the period to perform + reconciliation with the Calico datastore. [Default: + 5m]' + type: string + type: object + workloadEndpoint: + description: WorkloadEndpoint enables and configures the workload + endpoint controller. Enabled by default, set to nil to disable. + properties: + reconcilerPeriod: + description: 'ReconcilerPeriod is the period to perform + reconciliation with the Calico datastore. [Default: + 5m]' + type: string + type: object + type: object + debugProfilePort: + description: DebugProfilePort configures the port to serve memory + and cpu profiles on. If not specified, profiling is disabled. + format: int32 + type: integer + etcdV3CompactionPeriod: + description: 'EtcdV3CompactionPeriod is the period between etcdv3 + compaction requests. Set to 0 to disable. [Default: 10m]' + type: string + healthChecks: + description: 'HealthChecks enables or disables support for health + checks [Default: Enabled]' + type: string + logSeverityScreen: + description: 'LogSeverityScreen is the log severity above which + logs are sent to the stdout. [Default: Info]' + type: string + prometheusMetricsPort: + description: 'PrometheusMetricsPort is the TCP port that the Prometheus + metrics server should bind to. Set to 0 to disable. [Default: + 9094]' + type: integer + required: + - controllers + type: object + type: object + type: object + served: true + storage: true +status: + acceptedNames: + kind: "" + plural: "" + conditions: [] + storedVersions: [] +--- +# Source: calico/templates/kdd-crds.yaml +apiVersion: apiextensions.k8s.io/v1 +kind: CustomResourceDefinition +metadata: + name: networkpolicies.crd.projectcalico.org +spec: + group: crd.projectcalico.org + names: + kind: NetworkPolicy + listKind: NetworkPolicyList + plural: networkpolicies + singular: networkpolicy + preserveUnknownFields: false + scope: Namespaced + versions: + - name: v1 + schema: + openAPIV3Schema: + properties: + apiVersion: + description: 'APIVersion defines the versioned schema of this representation + of an object. Servers should convert recognized schemas to the latest + internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources' + type: string + kind: + description: 'Kind is a string value representing the REST resource this + object represents. Servers may infer this from the endpoint the client + submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds' + type: string + metadata: + type: object + spec: + properties: + egress: + description: The ordered set of egress rules. Each rule contains + a set of packet match criteria and a corresponding action to apply. + items: + description: "A Rule encapsulates a set of match criteria and an + action. Both selector-based security Policy and security Profiles + reference rules - separated out as a list of rules for both ingress + and egress packet matching. \n Each positive match criteria has + a negated version, prefixed with \"Not\". All the match criteria + within a rule must be satisfied for a packet to match. A single + rule can contain the positive and negative version of a match + and both must be satisfied for the rule to match." + properties: + action: + type: string + destination: + description: Destination contains the match criteria that apply + to destination entity. + properties: + namespaceSelector: + description: "NamespaceSelector is an optional field that + contains a selector expression. Only traffic that originates + from (or terminates at) endpoints within the selected + namespaces will be matched. When both NamespaceSelector + and another selector are defined on the same rule, then + only workload endpoints that are matched by both selectors + will be selected by the rule. \n For NetworkPolicy, an + empty NamespaceSelector implies that the Selector is limited + to selecting only workload endpoints in the same namespace + as the NetworkPolicy. \n For NetworkPolicy, `global()` + NamespaceSelector implies that the Selector is limited + to selecting only GlobalNetworkSet or HostEndpoint. \n + For GlobalNetworkPolicy, an empty NamespaceSelector implies + the Selector applies to workload endpoints across all + namespaces." + type: string + nets: + description: Nets is an optional field that restricts the + rule to only apply to traffic that originates from (or + terminates at) IP addresses in any of the given subnets. + items: + type: string + type: array + notNets: + description: NotNets is the negated version of the Nets + field. + items: + type: string + type: array + notPorts: + description: NotPorts is the negated version of the Ports + field. Since only some protocols have ports, if any ports + are specified it requires the Protocol match in the Rule + to be set to "TCP" or "UDP". + items: + anyOf: + - type: integer + - type: string + pattern: ^.* + x-kubernetes-int-or-string: true + type: array + notSelector: + description: NotSelector is the negated version of the Selector + field. See Selector field for subtleties with negated + selectors. + type: string + ports: + description: "Ports is an optional field that restricts + the rule to only apply to traffic that has a source (destination) + port that matches one of these ranges/values. This value + is a list of integers or strings that represent ranges + of ports. \n Since only some protocols have ports, if + any ports are specified it requires the Protocol match + in the Rule to be set to \"TCP\" or \"UDP\"." + items: + anyOf: + - type: integer + - type: string + pattern: ^.* + x-kubernetes-int-or-string: true + type: array + selector: + description: "Selector is an optional field that contains + a selector expression (see Policy for sample syntax). + \ Only traffic that originates from (terminates at) endpoints + matching the selector will be matched. \n Note that: in + addition to the negated version of the Selector (see NotSelector + below), the selector expression syntax itself supports + negation. The two types of negation are subtly different. + One negates the set of matched endpoints, the other negates + the whole match: \n \tSelector = \"!has(my_label)\" matches + packets that are from other Calico-controlled \tendpoints + that do not have the label \"my_label\". \n \tNotSelector + = \"has(my_label)\" matches packets that are not from + Calico-controlled \tendpoints that do have the label \"my_label\". + \n The effect is that the latter will accept packets from + non-Calico sources whereas the former is limited to packets + from Calico-controlled endpoints." + type: string + serviceAccounts: + description: ServiceAccounts is an optional field that restricts + the rule to only apply to traffic that originates from + (or terminates at) a pod running as a matching service + account. + properties: + names: + description: Names is an optional field that restricts + the rule to only apply to traffic that originates + from (or terminates at) a pod running as a service + account whose name is in the list. + items: + type: string + type: array + selector: + description: Selector is an optional field that restricts + the rule to only apply to traffic that originates + from (or terminates at) a pod running as a service + account that matches the given label selector. If + both Names and Selector are specified then they are + AND'ed. + type: string + type: object + services: + description: "Services is an optional field that contains + options for matching Kubernetes Services. If specified, + only traffic that originates from or terminates at endpoints + within the selected service(s) will be matched, and only + to/from each endpoint's port. \n Services cannot be specified + on the same rule as Selector, NotSelector, NamespaceSelector, + Nets, NotNets or ServiceAccounts. \n Ports and NotPorts + can only be specified with Services on ingress rules." + properties: + name: + description: Name specifies the name of a Kubernetes + Service to match. + type: string + namespace: + description: Namespace specifies the namespace of the + given Service. If left empty, the rule will match + within this policy's namespace. + type: string + type: object + type: object + http: + description: HTTP contains match criteria that apply to HTTP + requests. + properties: + methods: + description: Methods is an optional field that restricts + the rule to apply only to HTTP requests that use one of + the listed HTTP Methods (e.g. GET, PUT, etc.) Multiple + methods are OR'd together. + items: + type: string + type: array + paths: + description: 'Paths is an optional field that restricts + the rule to apply to HTTP requests that use one of the + listed HTTP Paths. Multiple paths are OR''d together. + e.g: - exact: /foo - prefix: /bar NOTE: Each entry may + ONLY specify either a `exact` or a `prefix` match. The + validator will check for it.' + items: + description: 'HTTPPath specifies an HTTP path to match. + It may be either of the form: exact: : which matches + the path exactly or prefix: : which matches + the path prefix' + properties: + exact: + type: string + prefix: + type: string + type: object + type: array + type: object + icmp: + description: ICMP is an optional field that restricts the rule + to apply to a specific type and code of ICMP traffic. This + should only be specified if the Protocol field is set to "ICMP" + or "ICMPv6". + properties: + code: + description: Match on a specific ICMP code. If specified, + the Type value must also be specified. This is a technical + limitation imposed by the kernel's iptables firewall, + which Calico uses to enforce the rule. + type: integer + type: + description: Match on a specific ICMP type. For example + a value of 8 refers to ICMP Echo Request (i.e. pings). + type: integer + type: object + ipVersion: + description: IPVersion is an optional field that restricts the + rule to only match a specific IP version. + type: integer + metadata: + description: Metadata contains additional information for this + rule + properties: + annotations: + additionalProperties: + type: string + description: Annotations is a set of key value pairs that + give extra information about the rule + type: object + type: object + notICMP: + description: NotICMP is the negated version of the ICMP field. + properties: + code: + description: Match on a specific ICMP code. If specified, + the Type value must also be specified. This is a technical + limitation imposed by the kernel's iptables firewall, + which Calico uses to enforce the rule. + type: integer + type: + description: Match on a specific ICMP type. For example + a value of 8 refers to ICMP Echo Request (i.e. pings). + type: integer + type: object + notProtocol: + anyOf: + - type: integer + - type: string + description: NotProtocol is the negated version of the Protocol + field. + pattern: ^.* + x-kubernetes-int-or-string: true + protocol: + anyOf: + - type: integer + - type: string + description: "Protocol is an optional field that restricts the + rule to only apply to traffic of a specific IP protocol. Required + if any of the EntityRules contain Ports (because ports only + apply to certain protocols). \n Must be one of these string + values: \"TCP\", \"UDP\", \"ICMP\", \"ICMPv6\", \"SCTP\", + \"UDPLite\" or an integer in the range 1-255." + pattern: ^.* + x-kubernetes-int-or-string: true + source: + description: Source contains the match criteria that apply to + source entity. + properties: + namespaceSelector: + description: "NamespaceSelector is an optional field that + contains a selector expression. Only traffic that originates + from (or terminates at) endpoints within the selected + namespaces will be matched. When both NamespaceSelector + and another selector are defined on the same rule, then + only workload endpoints that are matched by both selectors + will be selected by the rule. \n For NetworkPolicy, an + empty NamespaceSelector implies that the Selector is limited + to selecting only workload endpoints in the same namespace + as the NetworkPolicy. \n For NetworkPolicy, `global()` + NamespaceSelector implies that the Selector is limited + to selecting only GlobalNetworkSet or HostEndpoint. \n + For GlobalNetworkPolicy, an empty NamespaceSelector implies + the Selector applies to workload endpoints across all + namespaces." + type: string + nets: + description: Nets is an optional field that restricts the + rule to only apply to traffic that originates from (or + terminates at) IP addresses in any of the given subnets. + items: + type: string + type: array + notNets: + description: NotNets is the negated version of the Nets + field. + items: + type: string + type: array + notPorts: + description: NotPorts is the negated version of the Ports + field. Since only some protocols have ports, if any ports + are specified it requires the Protocol match in the Rule + to be set to "TCP" or "UDP". + items: + anyOf: + - type: integer + - type: string + pattern: ^.* + x-kubernetes-int-or-string: true + type: array + notSelector: + description: NotSelector is the negated version of the Selector + field. See Selector field for subtleties with negated + selectors. + type: string + ports: + description: "Ports is an optional field that restricts + the rule to only apply to traffic that has a source (destination) + port that matches one of these ranges/values. This value + is a list of integers or strings that represent ranges + of ports. \n Since only some protocols have ports, if + any ports are specified it requires the Protocol match + in the Rule to be set to \"TCP\" or \"UDP\"." + items: + anyOf: + - type: integer + - type: string + pattern: ^.* + x-kubernetes-int-or-string: true + type: array + selector: + description: "Selector is an optional field that contains + a selector expression (see Policy for sample syntax). + \ Only traffic that originates from (terminates at) endpoints + matching the selector will be matched. \n Note that: in + addition to the negated version of the Selector (see NotSelector + below), the selector expression syntax itself supports + negation. The two types of negation are subtly different. + One negates the set of matched endpoints, the other negates + the whole match: \n \tSelector = \"!has(my_label)\" matches + packets that are from other Calico-controlled \tendpoints + that do not have the label \"my_label\". \n \tNotSelector + = \"has(my_label)\" matches packets that are not from + Calico-controlled \tendpoints that do have the label \"my_label\". + \n The effect is that the latter will accept packets from + non-Calico sources whereas the former is limited to packets + from Calico-controlled endpoints." + type: string + serviceAccounts: + description: ServiceAccounts is an optional field that restricts + the rule to only apply to traffic that originates from + (or terminates at) a pod running as a matching service + account. + properties: + names: + description: Names is an optional field that restricts + the rule to only apply to traffic that originates + from (or terminates at) a pod running as a service + account whose name is in the list. + items: + type: string + type: array + selector: + description: Selector is an optional field that restricts + the rule to only apply to traffic that originates + from (or terminates at) a pod running as a service + account that matches the given label selector. If + both Names and Selector are specified then they are + AND'ed. + type: string + type: object + services: + description: "Services is an optional field that contains + options for matching Kubernetes Services. If specified, + only traffic that originates from or terminates at endpoints + within the selected service(s) will be matched, and only + to/from each endpoint's port. \n Services cannot be specified + on the same rule as Selector, NotSelector, NamespaceSelector, + Nets, NotNets or ServiceAccounts. \n Ports and NotPorts + can only be specified with Services on ingress rules." + properties: + name: + description: Name specifies the name of a Kubernetes + Service to match. + type: string + namespace: + description: Namespace specifies the namespace of the + given Service. If left empty, the rule will match + within this policy's namespace. + type: string + type: object + type: object + required: + - action + type: object + type: array + ingress: + description: The ordered set of ingress rules. Each rule contains + a set of packet match criteria and a corresponding action to apply. + items: + description: "A Rule encapsulates a set of match criteria and an + action. Both selector-based security Policy and security Profiles + reference rules - separated out as a list of rules for both ingress + and egress packet matching. \n Each positive match criteria has + a negated version, prefixed with \"Not\". All the match criteria + within a rule must be satisfied for a packet to match. A single + rule can contain the positive and negative version of a match + and both must be satisfied for the rule to match." + properties: + action: + type: string + destination: + description: Destination contains the match criteria that apply + to destination entity. + properties: + namespaceSelector: + description: "NamespaceSelector is an optional field that + contains a selector expression. Only traffic that originates + from (or terminates at) endpoints within the selected + namespaces will be matched. When both NamespaceSelector + and another selector are defined on the same rule, then + only workload endpoints that are matched by both selectors + will be selected by the rule. \n For NetworkPolicy, an + empty NamespaceSelector implies that the Selector is limited + to selecting only workload endpoints in the same namespace + as the NetworkPolicy. \n For NetworkPolicy, `global()` + NamespaceSelector implies that the Selector is limited + to selecting only GlobalNetworkSet or HostEndpoint. \n + For GlobalNetworkPolicy, an empty NamespaceSelector implies + the Selector applies to workload endpoints across all + namespaces." + type: string + nets: + description: Nets is an optional field that restricts the + rule to only apply to traffic that originates from (or + terminates at) IP addresses in any of the given subnets. + items: + type: string + type: array + notNets: + description: NotNets is the negated version of the Nets + field. + items: + type: string + type: array + notPorts: + description: NotPorts is the negated version of the Ports + field. Since only some protocols have ports, if any ports + are specified it requires the Protocol match in the Rule + to be set to "TCP" or "UDP". + items: + anyOf: + - type: integer + - type: string + pattern: ^.* + x-kubernetes-int-or-string: true + type: array + notSelector: + description: NotSelector is the negated version of the Selector + field. See Selector field for subtleties with negated + selectors. + type: string + ports: + description: "Ports is an optional field that restricts + the rule to only apply to traffic that has a source (destination) + port that matches one of these ranges/values. This value + is a list of integers or strings that represent ranges + of ports. \n Since only some protocols have ports, if + any ports are specified it requires the Protocol match + in the Rule to be set to \"TCP\" or \"UDP\"." + items: + anyOf: + - type: integer + - type: string + pattern: ^.* + x-kubernetes-int-or-string: true + type: array + selector: + description: "Selector is an optional field that contains + a selector expression (see Policy for sample syntax). + \ Only traffic that originates from (terminates at) endpoints + matching the selector will be matched. \n Note that: in + addition to the negated version of the Selector (see NotSelector + below), the selector expression syntax itself supports + negation. The two types of negation are subtly different. + One negates the set of matched endpoints, the other negates + the whole match: \n \tSelector = \"!has(my_label)\" matches + packets that are from other Calico-controlled \tendpoints + that do not have the label \"my_label\". \n \tNotSelector + = \"has(my_label)\" matches packets that are not from + Calico-controlled \tendpoints that do have the label \"my_label\". + \n The effect is that the latter will accept packets from + non-Calico sources whereas the former is limited to packets + from Calico-controlled endpoints." + type: string + serviceAccounts: + description: ServiceAccounts is an optional field that restricts + the rule to only apply to traffic that originates from + (or terminates at) a pod running as a matching service + account. + properties: + names: + description: Names is an optional field that restricts + the rule to only apply to traffic that originates + from (or terminates at) a pod running as a service + account whose name is in the list. + items: + type: string + type: array + selector: + description: Selector is an optional field that restricts + the rule to only apply to traffic that originates + from (or terminates at) a pod running as a service + account that matches the given label selector. If + both Names and Selector are specified then they are + AND'ed. + type: string + type: object + services: + description: "Services is an optional field that contains + options for matching Kubernetes Services. If specified, + only traffic that originates from or terminates at endpoints + within the selected service(s) will be matched, and only + to/from each endpoint's port. \n Services cannot be specified + on the same rule as Selector, NotSelector, NamespaceSelector, + Nets, NotNets or ServiceAccounts. \n Ports and NotPorts + can only be specified with Services on ingress rules." + properties: + name: + description: Name specifies the name of a Kubernetes + Service to match. + type: string + namespace: + description: Namespace specifies the namespace of the + given Service. If left empty, the rule will match + within this policy's namespace. + type: string + type: object + type: object + http: + description: HTTP contains match criteria that apply to HTTP + requests. + properties: + methods: + description: Methods is an optional field that restricts + the rule to apply only to HTTP requests that use one of + the listed HTTP Methods (e.g. GET, PUT, etc.) Multiple + methods are OR'd together. + items: + type: string + type: array + paths: + description: 'Paths is an optional field that restricts + the rule to apply to HTTP requests that use one of the + listed HTTP Paths. Multiple paths are OR''d together. + e.g: - exact: /foo - prefix: /bar NOTE: Each entry may + ONLY specify either a `exact` or a `prefix` match. The + validator will check for it.' + items: + description: 'HTTPPath specifies an HTTP path to match. + It may be either of the form: exact: : which matches + the path exactly or prefix: : which matches + the path prefix' + properties: + exact: + type: string + prefix: + type: string + type: object + type: array + type: object + icmp: + description: ICMP is an optional field that restricts the rule + to apply to a specific type and code of ICMP traffic. This + should only be specified if the Protocol field is set to "ICMP" + or "ICMPv6". + properties: + code: + description: Match on a specific ICMP code. If specified, + the Type value must also be specified. This is a technical + limitation imposed by the kernel's iptables firewall, + which Calico uses to enforce the rule. + type: integer + type: + description: Match on a specific ICMP type. For example + a value of 8 refers to ICMP Echo Request (i.e. pings). + type: integer + type: object + ipVersion: + description: IPVersion is an optional field that restricts the + rule to only match a specific IP version. + type: integer + metadata: + description: Metadata contains additional information for this + rule + properties: + annotations: + additionalProperties: + type: string + description: Annotations is a set of key value pairs that + give extra information about the rule + type: object + type: object + notICMP: + description: NotICMP is the negated version of the ICMP field. + properties: + code: + description: Match on a specific ICMP code. If specified, + the Type value must also be specified. This is a technical + limitation imposed by the kernel's iptables firewall, + which Calico uses to enforce the rule. + type: integer + type: + description: Match on a specific ICMP type. For example + a value of 8 refers to ICMP Echo Request (i.e. pings). + type: integer + type: object + notProtocol: + anyOf: + - type: integer + - type: string + description: NotProtocol is the negated version of the Protocol + field. + pattern: ^.* + x-kubernetes-int-or-string: true + protocol: + anyOf: + - type: integer + - type: string + description: "Protocol is an optional field that restricts the + rule to only apply to traffic of a specific IP protocol. Required + if any of the EntityRules contain Ports (because ports only + apply to certain protocols). \n Must be one of these string + values: \"TCP\", \"UDP\", \"ICMP\", \"ICMPv6\", \"SCTP\", + \"UDPLite\" or an integer in the range 1-255." + pattern: ^.* + x-kubernetes-int-or-string: true + source: + description: Source contains the match criteria that apply to + source entity. + properties: + namespaceSelector: + description: "NamespaceSelector is an optional field that + contains a selector expression. Only traffic that originates + from (or terminates at) endpoints within the selected + namespaces will be matched. When both NamespaceSelector + and another selector are defined on the same rule, then + only workload endpoints that are matched by both selectors + will be selected by the rule. \n For NetworkPolicy, an + empty NamespaceSelector implies that the Selector is limited + to selecting only workload endpoints in the same namespace + as the NetworkPolicy. \n For NetworkPolicy, `global()` + NamespaceSelector implies that the Selector is limited + to selecting only GlobalNetworkSet or HostEndpoint. \n + For GlobalNetworkPolicy, an empty NamespaceSelector implies + the Selector applies to workload endpoints across all + namespaces." + type: string + nets: + description: Nets is an optional field that restricts the + rule to only apply to traffic that originates from (or + terminates at) IP addresses in any of the given subnets. + items: + type: string + type: array + notNets: + description: NotNets is the negated version of the Nets + field. + items: + type: string + type: array + notPorts: + description: NotPorts is the negated version of the Ports + field. Since only some protocols have ports, if any ports + are specified it requires the Protocol match in the Rule + to be set to "TCP" or "UDP". + items: + anyOf: + - type: integer + - type: string + pattern: ^.* + x-kubernetes-int-or-string: true + type: array + notSelector: + description: NotSelector is the negated version of the Selector + field. See Selector field for subtleties with negated + selectors. + type: string + ports: + description: "Ports is an optional field that restricts + the rule to only apply to traffic that has a source (destination) + port that matches one of these ranges/values. This value + is a list of integers or strings that represent ranges + of ports. \n Since only some protocols have ports, if + any ports are specified it requires the Protocol match + in the Rule to be set to \"TCP\" or \"UDP\"." + items: + anyOf: + - type: integer + - type: string + pattern: ^.* + x-kubernetes-int-or-string: true + type: array + selector: + description: "Selector is an optional field that contains + a selector expression (see Policy for sample syntax). + \ Only traffic that originates from (terminates at) endpoints + matching the selector will be matched. \n Note that: in + addition to the negated version of the Selector (see NotSelector + below), the selector expression syntax itself supports + negation. The two types of negation are subtly different. + One negates the set of matched endpoints, the other negates + the whole match: \n \tSelector = \"!has(my_label)\" matches + packets that are from other Calico-controlled \tendpoints + that do not have the label \"my_label\". \n \tNotSelector + = \"has(my_label)\" matches packets that are not from + Calico-controlled \tendpoints that do have the label \"my_label\". + \n The effect is that the latter will accept packets from + non-Calico sources whereas the former is limited to packets + from Calico-controlled endpoints." + type: string + serviceAccounts: + description: ServiceAccounts is an optional field that restricts + the rule to only apply to traffic that originates from + (or terminates at) a pod running as a matching service + account. + properties: + names: + description: Names is an optional field that restricts + the rule to only apply to traffic that originates + from (or terminates at) a pod running as a service + account whose name is in the list. + items: + type: string + type: array + selector: + description: Selector is an optional field that restricts + the rule to only apply to traffic that originates + from (or terminates at) a pod running as a service + account that matches the given label selector. If + both Names and Selector are specified then they are + AND'ed. + type: string + type: object + services: + description: "Services is an optional field that contains + options for matching Kubernetes Services. If specified, + only traffic that originates from or terminates at endpoints + within the selected service(s) will be matched, and only + to/from each endpoint's port. \n Services cannot be specified + on the same rule as Selector, NotSelector, NamespaceSelector, + Nets, NotNets or ServiceAccounts. \n Ports and NotPorts + can only be specified with Services on ingress rules." + properties: + name: + description: Name specifies the name of a Kubernetes + Service to match. + type: string + namespace: + description: Namespace specifies the namespace of the + given Service. If left empty, the rule will match + within this policy's namespace. + type: string + type: object + type: object + required: + - action + type: object + type: array + order: + description: Order is an optional field that specifies the order in + which the policy is applied. Policies with higher "order" are applied + after those with lower order within the same tier. If the order + is omitted, it may be considered to be "infinite" - i.e. the policy + will be applied last. Policies with identical order will be applied + in alphanumerical order based on the Policy "Name" within the tier. + type: number + performanceHints: + description: "PerformanceHints contains a list of hints to Calico's + policy engine to help process the policy more efficiently. Hints + never change the enforcement behaviour of the policy. \n Currently, + the only available hint is \"AssumeNeededOnEveryNode\". When that + hint is set on a policy, Felix will act as if the policy matches + a local endpoint even if it does not. This is useful for \"preloading\" + any large static policies that are known to be used on every node. + If the policy is _not_ used on a particular node then the work done + to preload the policy (and to maintain it) is wasted." + items: + type: string + type: array + selector: + description: "The selector is an expression used to pick out the endpoints + that the policy should be applied to. \n Selector expressions follow + this syntax: \n \tlabel == \"string_literal\" -> comparison, e.g. + my_label == \"foo bar\" \tlabel != \"string_literal\" -> not + equal; also matches if label is not present \tlabel in { \"a\", + \"b\", \"c\", ... } -> true if the value of label X is one of + \"a\", \"b\", \"c\" \tlabel not in { \"a\", \"b\", \"c\", ... } + \ -> true if the value of label X is not one of \"a\", \"b\", \"c\" + \thas(label_name) -> True if that label is present \t! expr -> + negation of expr \texpr && expr -> Short-circuit and \texpr || + expr -> Short-circuit or \t( expr ) -> parens for grouping \tall() + or the empty selector -> matches all endpoints. \n Label names are + allowed to contain alphanumerics, -, _ and /. String literals are + more permissive but they do not support escape characters. \n Examples + (with made-up labels): \n \ttype == \"webserver\" && deployment + == \"prod\" \ttype in {\"frontend\", \"backend\"} \tdeployment != + \"dev\" \t! has(label_name)" + type: string + serviceAccountSelector: + description: ServiceAccountSelector is an optional field for an expression + used to select a pod based on service accounts. + type: string + tier: + description: The name of the tier that this policy belongs to. If + this is omitted, the default tier (name is "default") is assumed. The + specified tier must exist in order to create security policies within + the tier, the "default" tier is created automatically if it does + not exist, this means for deployments requiring only a single Tier, + the tier name may be omitted on all policy management requests. + type: string + types: + description: "Types indicates whether this policy applies to ingress, + or to egress, or to both. When not explicitly specified (and so + the value on creation is empty or nil), Calico defaults Types according + to what Ingress and Egress are present in the policy. The default + is: \n - [ PolicyTypeIngress ], if there are no Egress rules (including + the case where there are also no Ingress rules) \n - [ PolicyTypeEgress + ], if there are Egress rules but no Ingress rules \n - [ PolicyTypeIngress, + PolicyTypeEgress ], if there are both Ingress and Egress rules. + \n When the policy is read back again, Types will always be one + of these values, never empty or nil." + items: + description: PolicyType enumerates the possible values of the PolicySpec + Types field. + type: string + type: array + type: object + type: object + served: true + storage: true +status: + acceptedNames: + kind: "" + plural: "" + conditions: [] + storedVersions: [] +--- +# Source: calico/templates/kdd-crds.yaml +apiVersion: apiextensions.k8s.io/v1 +kind: CustomResourceDefinition +metadata: + name: networksets.crd.projectcalico.org +spec: + group: crd.projectcalico.org + names: + kind: NetworkSet + listKind: NetworkSetList + plural: networksets + singular: networkset + preserveUnknownFields: false + scope: Namespaced + versions: + - name: v1 + schema: + openAPIV3Schema: + description: NetworkSet is the Namespaced-equivalent of the GlobalNetworkSet. + properties: + apiVersion: + description: 'APIVersion defines the versioned schema of this representation + of an object. Servers should convert recognized schemas to the latest + internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources' + type: string + kind: + description: 'Kind is a string value representing the REST resource this + object represents. Servers may infer this from the endpoint the client + submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds' + type: string + metadata: + type: object + spec: + description: NetworkSetSpec contains the specification for a NetworkSet + resource. + properties: + nets: + description: The list of IP networks that belong to this set. + items: + type: string + type: array + type: object + type: object + served: true + storage: true +status: + acceptedNames: + kind: "" + plural: "" + conditions: [] + storedVersions: [] +--- +# Source: calico/templates/kdd-crds.yaml +apiVersion: apiextensions.k8s.io/v1 +kind: CustomResourceDefinition +metadata: + annotations: + controller-gen.kubebuilder.io/version: (devel) + creationTimestamp: null + name: tiers.crd.projectcalico.org +spec: + group: crd.projectcalico.org + names: + kind: Tier + listKind: TierList + plural: tiers + singular: tier + scope: Cluster + versions: + - name: v1 + schema: + openAPIV3Schema: + properties: + apiVersion: + description: 'APIVersion defines the versioned schema of this representation + of an object. Servers should convert recognized schemas to the latest + internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources' + type: string + kind: + description: 'Kind is a string value representing the REST resource this + object represents. Servers may infer this from the endpoint the client + submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds' + type: string + metadata: + type: object + spec: + description: TierSpec contains the specification for a security policy + tier resource. + properties: + defaultAction: + description: 'DefaultAction specifies the action applied to workloads + selected by a policy in the tier, but not rule matched the workload''s + traffic. [Default: Deny]' + enum: + - Pass + - Deny + type: string + order: + description: Order is an optional field that specifies the order in + which the tier is applied. Tiers with higher "order" are applied + after those with lower order. If the order is omitted, it may be + considered to be "infinite" - i.e. the tier will be applied last. Tiers + with identical order will be applied in alphanumerical order based + on the Tier "Name". + type: number + type: object + type: object + served: true + storage: true +status: + acceptedNames: + kind: "" + plural: "" + conditions: [] + storedVersions: [] +--- +# Source: calico/templates/kdd-crds.yaml +apiVersion: apiextensions.k8s.io/v1 +kind: CustomResourceDefinition +metadata: + annotations: + api-approved.kubernetes.io: https://github.com/kubernetes-sigs/network-policy-api/pull/30 + policy.networking.k8s.io/bundle-version: v0.1.1 + policy.networking.k8s.io/channel: experimental + creationTimestamp: null + name: adminnetworkpolicies.policy.networking.k8s.io +spec: + group: policy.networking.k8s.io + names: + kind: AdminNetworkPolicy + listKind: AdminNetworkPolicyList + plural: adminnetworkpolicies + shortNames: + - anp + singular: adminnetworkpolicy + scope: Cluster + versions: + - additionalPrinterColumns: + - jsonPath: .spec.priority + name: Priority + type: string + - jsonPath: .metadata.creationTimestamp + name: Age + type: date + name: v1alpha1 + schema: + openAPIV3Schema: + description: |- + AdminNetworkPolicy is a cluster level resource that is part of the + AdminNetworkPolicy API. + properties: + apiVersion: + description: |- + APIVersion defines the versioned schema of this representation of an object. + Servers should convert recognized schemas to the latest internal value, and + may reject unrecognized values. + More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources + type: string + kind: + description: |- + Kind is a string value representing the REST resource this object represents. + Servers may infer this from the endpoint the client submits requests to. + Cannot be updated. + In CamelCase. + More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds + type: string + metadata: + type: object + spec: + description: Specification of the desired behavior of AdminNetworkPolicy. + properties: + egress: + description: |- + Egress is the list of Egress rules to be applied to the selected pods. + A total of 100 rules will be allowed in each ANP instance. + The relative precedence of egress rules within a single ANP object (all of + which share the priority) will be determined by the order in which the rule + is written. Thus, a rule that appears at the top of the egress rules + would take the highest precedence. + ANPs with no egress rules do not affect egress traffic. + + + Support: Core + items: + description: |- + AdminNetworkPolicyEgressRule describes an action to take on a particular + set of traffic originating from pods selected by a AdminNetworkPolicy's + Subject field. + + properties: + action: + description: |- + Action specifies the effect this rule will have on matching traffic. + Currently the following actions are supported: + Allow: allows the selected traffic (even if it would otherwise have been denied by NetworkPolicy) + Deny: denies the selected traffic + Pass: instructs the selected traffic to skip any remaining ANP rules, and + then pass execution to any NetworkPolicies that select the pod. + If the pod is not selected by any NetworkPolicies then execution + is passed to any BaselineAdminNetworkPolicies that select the pod. + + + Support: Core + enum: + - Allow + - Deny + - Pass + type: string + name: + description: |- + Name is an identifier for this rule, that may be no more than 100 characters + in length. This field should be used by the implementation to help + improve observability, readability and error-reporting for any applied + AdminNetworkPolicies. + + + Support: Core + maxLength: 100 + type: string + ports: + description: |- + Ports allows for matching traffic based on port and protocols. + This field is a list of destination ports for the outgoing egress traffic. + If Ports is not set then the rule does not filter traffic via port. + + + Support: Core + items: + description: |- + AdminNetworkPolicyPort describes how to select network ports on pod(s). + Exactly one field must be set. + maxProperties: 1 + minProperties: 1 + properties: + namedPort: + description: |- + NamedPort selects a port on a pod(s) based on name. + + + Support: Extended + + + + type: string + portNumber: + description: |- + Port selects a port on a pod(s) based on number. + + + Support: Core + properties: + port: + description: |- + Number defines a network port value. + + + Support: Core + format: int32 + maximum: 65535 + minimum: 1 + type: integer + protocol: + default: TCP + description: |- + Protocol is the network protocol (TCP, UDP, or SCTP) which traffic must + match. If not specified, this field defaults to TCP. + + + Support: Core + type: string + required: + - port + - protocol + type: object + portRange: + description: |- + PortRange selects a port range on a pod(s) based on provided start and end + values. + + + Support: Core + properties: + end: + description: |- + End defines a network port that is the end of a port range, the End value + must be greater than Start. + + + Support: Core + format: int32 + maximum: 65535 + minimum: 1 + type: integer + protocol: + default: TCP + description: |- + Protocol is the network protocol (TCP, UDP, or SCTP) which traffic must + match. If not specified, this field defaults to TCP. + + + Support: Core + type: string + start: + description: |- + Start defines a network port that is the start of a port range, the Start + value must be less than End. + + + Support: Core + format: int32 + maximum: 65535 + minimum: 1 + type: integer + required: + - end + - start + type: object + type: object + maxItems: 100 + type: array + to: + description: |- + To is the List of destinations whose traffic this rule applies to. + If any AdminNetworkPolicyEgressPeer matches the destination of outgoing + traffic then the specified action is applied. + This field must be defined and contain at least one item. + + + Support: Core + items: + description: |- + AdminNetworkPolicyEgressPeer defines a peer to allow traffic to. + Exactly one of the selector pointers must be set for a given peer. If a + consumer observes none of its fields are set, they must assume an unknown + option has been specified and fail closed. + maxProperties: 1 + minProperties: 1 + properties: + namespaces: + description: |- + Namespaces defines a way to select all pods within a set of Namespaces. + Note that host-networked pods are not included in this type of peer. + + + Support: Core + properties: + matchExpressions: + description: matchExpressions is a list of label selector + requirements. The requirements are ANDed. + items: + description: |- + A label selector requirement is a selector that contains values, a key, and an operator that + relates the key and values. + properties: + key: + description: key is the label key that the selector + applies to. + type: string + operator: + description: |- + operator represents a key's relationship to a set of values. + Valid operators are In, NotIn, Exists and DoesNotExist. + type: string + values: + description: |- + values is an array of string values. If the operator is In or NotIn, + the values array must be non-empty. If the operator is Exists or DoesNotExist, + the values array must be empty. This array is replaced during a strategic + merge patch. + items: + type: string + type: array + required: + - key + - operator + type: object + type: array + matchLabels: + additionalProperties: + type: string + description: |- + matchLabels is a map of {key,value} pairs. A single {key,value} in the matchLabels + map is equivalent to an element of matchExpressions, whose key field is "key", the + operator is "In", and the values array contains only "value". The requirements are ANDed. + type: object + type: object + x-kubernetes-map-type: atomic + networks: + description: |- + Networks defines a way to select peers via CIDR blocks. + This is intended for representing entities that live outside the cluster, + which can't be selected by pods, namespaces and nodes peers, but note + that cluster-internal traffic will be checked against the rule as + well. So if you Allow or Deny traffic to `"0.0.0.0/0"`, that will allow + or deny all IPv4 pod-to-pod traffic as well. If you don't want that, + add a rule that Passes all pod traffic before the Networks rule. + + + Each item in Networks should be provided in the CIDR format and should be + IPv4 or IPv6, for example "10.0.0.0/8" or "fd00::/8". + + + Networks can have upto 25 CIDRs specified. + + + Support: Extended + + + + items: + description: |- + CIDR is an IP address range in CIDR notation (for example, "10.0.0.0/8" or "fd00::/8"). + This string must be validated by implementations using net.ParseCIDR + TODO: Introduce CEL CIDR validation regex isCIDR() in Kube 1.31 when it is available. + maxLength: 43 + type: string + x-kubernetes-validations: + - message: CIDR must be either an IPv4 or IPv6 address. + IPv4 address embedded in IPv6 addresses are not + supported + rule: self.contains(':') != self.contains('.') + maxItems: 25 + minItems: 1 + type: array + x-kubernetes-list-type: set + nodes: + description: |- + Nodes defines a way to select a set of nodes in + the cluster. This field follows standard label selector + semantics; if present but empty, it selects all Nodes. + + + Support: Extended + + + + properties: + matchExpressions: + description: matchExpressions is a list of label selector + requirements. The requirements are ANDed. + items: + description: |- + A label selector requirement is a selector that contains values, a key, and an operator that + relates the key and values. + properties: + key: + description: key is the label key that the selector + applies to. + type: string + operator: + description: |- + operator represents a key's relationship to a set of values. + Valid operators are In, NotIn, Exists and DoesNotExist. + type: string + values: + description: |- + values is an array of string values. If the operator is In or NotIn, + the values array must be non-empty. If the operator is Exists or DoesNotExist, + the values array must be empty. This array is replaced during a strategic + merge patch. + items: + type: string + type: array + required: + - key + - operator + type: object + type: array + matchLabels: + additionalProperties: + type: string + description: |- + matchLabels is a map of {key,value} pairs. A single {key,value} in the matchLabels + map is equivalent to an element of matchExpressions, whose key field is "key", the + operator is "In", and the values array contains only "value". The requirements are ANDed. + type: object + type: object + x-kubernetes-map-type: atomic + pods: + description: |- + Pods defines a way to select a set of pods in + a set of namespaces. Note that host-networked pods + are not included in this type of peer. + + + Support: Core + properties: + namespaceSelector: + description: |- + NamespaceSelector follows standard label selector semantics; if empty, + it selects all Namespaces. + properties: + matchExpressions: + description: matchExpressions is a list of label + selector requirements. The requirements are + ANDed. + items: + description: |- + A label selector requirement is a selector that contains values, a key, and an operator that + relates the key and values. + properties: + key: + description: key is the label key that the + selector applies to. + type: string + operator: + description: |- + operator represents a key's relationship to a set of values. + Valid operators are In, NotIn, Exists and DoesNotExist. + type: string + values: + description: |- + values is an array of string values. If the operator is In or NotIn, + the values array must be non-empty. If the operator is Exists or DoesNotExist, + the values array must be empty. This array is replaced during a strategic + merge patch. + items: + type: string + type: array + required: + - key + - operator + type: object + type: array + matchLabels: + additionalProperties: + type: string + description: |- + matchLabels is a map of {key,value} pairs. A single {key,value} in the matchLabels + map is equivalent to an element of matchExpressions, whose key field is "key", the + operator is "In", and the values array contains only "value". The requirements are ANDed. + type: object + type: object + x-kubernetes-map-type: atomic + podSelector: + description: |- + PodSelector is used to explicitly select pods within a namespace; if empty, + it selects all Pods. + properties: + matchExpressions: + description: matchExpressions is a list of label + selector requirements. The requirements are + ANDed. + items: + description: |- + A label selector requirement is a selector that contains values, a key, and an operator that + relates the key and values. + properties: + key: + description: key is the label key that the + selector applies to. + type: string + operator: + description: |- + operator represents a key's relationship to a set of values. + Valid operators are In, NotIn, Exists and DoesNotExist. + type: string + values: + description: |- + values is an array of string values. If the operator is In or NotIn, + the values array must be non-empty. If the operator is Exists or DoesNotExist, + the values array must be empty. This array is replaced during a strategic + merge patch. + items: + type: string + type: array + required: + - key + - operator + type: object + type: array + matchLabels: + additionalProperties: + type: string + description: |- + matchLabels is a map of {key,value} pairs. A single {key,value} in the matchLabels + map is equivalent to an element of matchExpressions, whose key field is "key", the + operator is "In", and the values array contains only "value". The requirements are ANDed. + type: object + type: object + x-kubernetes-map-type: atomic + required: + - namespaceSelector + - podSelector + type: object + type: object + maxItems: 100 + minItems: 1 + type: array + required: + - action + - to + type: object + x-kubernetes-validations: + - message: networks/nodes peer cannot be set with namedPorts since + there are no namedPorts for networks/nodes + rule: '!(self.to.exists(peer, has(peer.networks) || has(peer.nodes)) + && has(self.ports) && self.ports.exists(port, has(port.namedPort)))' + maxItems: 100 + type: array + ingress: + description: |- + Ingress is the list of Ingress rules to be applied to the selected pods. + A total of 100 rules will be allowed in each ANP instance. + The relative precedence of ingress rules within a single ANP object (all of + which share the priority) will be determined by the order in which the rule + is written. Thus, a rule that appears at the top of the ingress rules + would take the highest precedence. + ANPs with no ingress rules do not affect ingress traffic. + + + Support: Core + items: + description: |- + AdminNetworkPolicyIngressRule describes an action to take on a particular + set of traffic destined for pods selected by an AdminNetworkPolicy's + Subject field. + properties: + action: + description: |- + Action specifies the effect this rule will have on matching traffic. + Currently the following actions are supported: + Allow: allows the selected traffic (even if it would otherwise have been denied by NetworkPolicy) + Deny: denies the selected traffic + Pass: instructs the selected traffic to skip any remaining ANP rules, and + then pass execution to any NetworkPolicies that select the pod. + If the pod is not selected by any NetworkPolicies then execution + is passed to any BaselineAdminNetworkPolicies that select the pod. + + + Support: Core + enum: + - Allow + - Deny + - Pass + type: string + from: + description: |- + From is the list of sources whose traffic this rule applies to. + If any AdminNetworkPolicyIngressPeer matches the source of incoming + traffic then the specified action is applied. + This field must be defined and contain at least one item. + + + Support: Core + items: + description: |- + AdminNetworkPolicyIngressPeer defines an in-cluster peer to allow traffic from. + Exactly one of the selector pointers must be set for a given peer. If a + consumer observes none of its fields are set, they must assume an unknown + option has been specified and fail closed. + maxProperties: 1 + minProperties: 1 + properties: + namespaces: + description: |- + Namespaces defines a way to select all pods within a set of Namespaces. + Note that host-networked pods are not included in this type of peer. + + + Support: Core + properties: + matchExpressions: + description: matchExpressions is a list of label selector + requirements. The requirements are ANDed. + items: + description: |- + A label selector requirement is a selector that contains values, a key, and an operator that + relates the key and values. + properties: + key: + description: key is the label key that the selector + applies to. + type: string + operator: + description: |- + operator represents a key's relationship to a set of values. + Valid operators are In, NotIn, Exists and DoesNotExist. + type: string + values: + description: |- + values is an array of string values. If the operator is In or NotIn, + the values array must be non-empty. If the operator is Exists or DoesNotExist, + the values array must be empty. This array is replaced during a strategic + merge patch. + items: + type: string + type: array + required: + - key + - operator + type: object + type: array + matchLabels: + additionalProperties: + type: string + description: |- + matchLabels is a map of {key,value} pairs. A single {key,value} in the matchLabels + map is equivalent to an element of matchExpressions, whose key field is "key", the + operator is "In", and the values array contains only "value". The requirements are ANDed. + type: object + type: object + x-kubernetes-map-type: atomic + pods: + description: |- + Pods defines a way to select a set of pods in + a set of namespaces. Note that host-networked pods + are not included in this type of peer. + + + Support: Core + properties: + namespaceSelector: + description: |- + NamespaceSelector follows standard label selector semantics; if empty, + it selects all Namespaces. + properties: + matchExpressions: + description: matchExpressions is a list of label + selector requirements. The requirements are + ANDed. + items: + description: |- + A label selector requirement is a selector that contains values, a key, and an operator that + relates the key and values. + properties: + key: + description: key is the label key that the + selector applies to. + type: string + operator: + description: |- + operator represents a key's relationship to a set of values. + Valid operators are In, NotIn, Exists and DoesNotExist. + type: string + values: + description: |- + values is an array of string values. If the operator is In or NotIn, + the values array must be non-empty. If the operator is Exists or DoesNotExist, + the values array must be empty. This array is replaced during a strategic + merge patch. + items: + type: string + type: array + required: + - key + - operator + type: object + type: array + matchLabels: + additionalProperties: + type: string + description: |- + matchLabels is a map of {key,value} pairs. A single {key,value} in the matchLabels + map is equivalent to an element of matchExpressions, whose key field is "key", the + operator is "In", and the values array contains only "value". The requirements are ANDed. + type: object + type: object + x-kubernetes-map-type: atomic + podSelector: + description: |- + PodSelector is used to explicitly select pods within a namespace; if empty, + it selects all Pods. + properties: + matchExpressions: + description: matchExpressions is a list of label + selector requirements. The requirements are + ANDed. + items: + description: |- + A label selector requirement is a selector that contains values, a key, and an operator that + relates the key and values. + properties: + key: + description: key is the label key that the + selector applies to. + type: string + operator: + description: |- + operator represents a key's relationship to a set of values. + Valid operators are In, NotIn, Exists and DoesNotExist. + type: string + values: + description: |- + values is an array of string values. If the operator is In or NotIn, + the values array must be non-empty. If the operator is Exists or DoesNotExist, + the values array must be empty. This array is replaced during a strategic + merge patch. + items: + type: string + type: array + required: + - key + - operator + type: object + type: array + matchLabels: + additionalProperties: + type: string + description: |- + matchLabels is a map of {key,value} pairs. A single {key,value} in the matchLabels + map is equivalent to an element of matchExpressions, whose key field is "key", the + operator is "In", and the values array contains only "value". The requirements are ANDed. + type: object + type: object + x-kubernetes-map-type: atomic + required: + - namespaceSelector + - podSelector + type: object + type: object + maxItems: 100 + minItems: 1 + type: array + name: + description: |- + Name is an identifier for this rule, that may be no more than 100 characters + in length. This field should be used by the implementation to help + improve observability, readability and error-reporting for any applied + AdminNetworkPolicies. + + + Support: Core + maxLength: 100 + type: string + ports: + description: |- + Ports allows for matching traffic based on port and protocols. + This field is a list of ports which should be matched on + the pods selected for this policy i.e the subject of the policy. + So it matches on the destination port for the ingress traffic. + If Ports is not set then the rule does not filter traffic via port. + + + Support: Core + items: + description: |- + AdminNetworkPolicyPort describes how to select network ports on pod(s). + Exactly one field must be set. + maxProperties: 1 + minProperties: 1 + properties: + namedPort: + description: |- + NamedPort selects a port on a pod(s) based on name. + + + Support: Extended + + + + type: string + portNumber: + description: |- + Port selects a port on a pod(s) based on number. + + + Support: Core + properties: + port: + description: |- + Number defines a network port value. + + + Support: Core + format: int32 + maximum: 65535 + minimum: 1 + type: integer + protocol: + default: TCP + description: |- + Protocol is the network protocol (TCP, UDP, or SCTP) which traffic must + match. If not specified, this field defaults to TCP. + + + Support: Core + type: string + required: + - port + - protocol + type: object + portRange: + description: |- + PortRange selects a port range on a pod(s) based on provided start and end + values. + + + Support: Core + properties: + end: + description: |- + End defines a network port that is the end of a port range, the End value + must be greater than Start. + + + Support: Core + format: int32 + maximum: 65535 + minimum: 1 + type: integer + protocol: + default: TCP + description: |- + Protocol is the network protocol (TCP, UDP, or SCTP) which traffic must + match. If not specified, this field defaults to TCP. + + + Support: Core + type: string + start: + description: |- + Start defines a network port that is the start of a port range, the Start + value must be less than End. + + + Support: Core + format: int32 + maximum: 65535 + minimum: 1 + type: integer + required: + - end + - start + type: object + type: object + maxItems: 100 + type: array + required: + - action + - from + type: object + maxItems: 100 + type: array + priority: + description: |- + Priority is a value from 0 to 1000. Rules with lower priority values have + higher precedence, and are checked before rules with higher priority values. + All AdminNetworkPolicy rules have higher precedence than NetworkPolicy or + BaselineAdminNetworkPolicy rules + The behavior is undefined if two ANP objects have same priority. + + + Support: Core + format: int32 + maximum: 1000 + minimum: 0 + type: integer + subject: + description: |- + Subject defines the pods to which this AdminNetworkPolicy applies. + Note that host-networked pods are not included in subject selection. + + + Support: Core + maxProperties: 1 + minProperties: 1 + properties: + namespaces: + description: Namespaces is used to select pods via namespace selectors. + properties: + matchExpressions: + description: matchExpressions is a list of label selector + requirements. The requirements are ANDed. + items: + description: |- + A label selector requirement is a selector that contains values, a key, and an operator that + relates the key and values. + properties: + key: + description: key is the label key that the selector + applies to. + type: string + operator: + description: |- + operator represents a key's relationship to a set of values. + Valid operators are In, NotIn, Exists and DoesNotExist. + type: string + values: + description: |- + values is an array of string values. If the operator is In or NotIn, + the values array must be non-empty. If the operator is Exists or DoesNotExist, + the values array must be empty. This array is replaced during a strategic + merge patch. + items: + type: string + type: array + required: + - key + - operator + type: object + type: array + matchLabels: + additionalProperties: + type: string + description: |- + matchLabels is a map of {key,value} pairs. A single {key,value} in the matchLabels + map is equivalent to an element of matchExpressions, whose key field is "key", the + operator is "In", and the values array contains only "value". The requirements are ANDed. + type: object + type: object + x-kubernetes-map-type: atomic + pods: + description: Pods is used to select pods via namespace AND pod + selectors. + properties: + namespaceSelector: + description: |- + NamespaceSelector follows standard label selector semantics; if empty, + it selects all Namespaces. + properties: + matchExpressions: + description: matchExpressions is a list of label selector + requirements. The requirements are ANDed. + items: + description: |- + A label selector requirement is a selector that contains values, a key, and an operator that + relates the key and values. + properties: + key: + description: key is the label key that the selector + applies to. + type: string + operator: + description: |- + operator represents a key's relationship to a set of values. + Valid operators are In, NotIn, Exists and DoesNotExist. + type: string + values: + description: |- + values is an array of string values. If the operator is In or NotIn, + the values array must be non-empty. If the operator is Exists or DoesNotExist, + the values array must be empty. This array is replaced during a strategic + merge patch. + items: + type: string + type: array + required: + - key + - operator + type: object + type: array + matchLabels: + additionalProperties: + type: string + description: |- + matchLabels is a map of {key,value} pairs. A single {key,value} in the matchLabels + map is equivalent to an element of matchExpressions, whose key field is "key", the + operator is "In", and the values array contains only "value". The requirements are ANDed. + type: object + type: object + x-kubernetes-map-type: atomic + podSelector: + description: |- + PodSelector is used to explicitly select pods within a namespace; if empty, + it selects all Pods. + properties: + matchExpressions: + description: matchExpressions is a list of label selector + requirements. The requirements are ANDed. + items: + description: |- + A label selector requirement is a selector that contains values, a key, and an operator that + relates the key and values. + properties: + key: + description: key is the label key that the selector + applies to. + type: string + operator: + description: |- + operator represents a key's relationship to a set of values. + Valid operators are In, NotIn, Exists and DoesNotExist. + type: string + values: + description: |- + values is an array of string values. If the operator is In or NotIn, + the values array must be non-empty. If the operator is Exists or DoesNotExist, + the values array must be empty. This array is replaced during a strategic + merge patch. + items: + type: string + type: array + required: + - key + - operator + type: object + type: array + matchLabels: + additionalProperties: + type: string + description: |- + matchLabels is a map of {key,value} pairs. A single {key,value} in the matchLabels + map is equivalent to an element of matchExpressions, whose key field is "key", the + operator is "In", and the values array contains only "value". The requirements are ANDed. + type: object + type: object + x-kubernetes-map-type: atomic + required: + - namespaceSelector + - podSelector + type: object + type: object + required: + - priority + - subject + type: object + status: + description: Status is the status to be reported by the implementation. + properties: + conditions: + items: + description: "Condition contains details for one aspect of the current + state of this API Resource.\n---\nThis struct is intended for + direct use as an array at the field path .status.conditions. For + example,\n\n\n\ttype FooStatus struct{\n\t // Represents the + observations of a foo's current state.\n\t // Known .status.conditions.type + are: \"Available\", \"Progressing\", and \"Degraded\"\n\t // + +patchMergeKey=type\n\t // +patchStrategy=merge\n\t // +listType=map\n\t + \ // +listMapKey=type\n\t Conditions []metav1.Condition `json:\"conditions,omitempty\" + patchStrategy:\"merge\" patchMergeKey:\"type\" protobuf:\"bytes,1,rep,name=conditions\"`\n\n\n\t + \ // other fields\n\t}" + properties: + lastTransitionTime: + description: |- + lastTransitionTime is the last time the condition transitioned from one status to another. + This should be when the underlying condition changed. If that is not known, then using the time when the API field changed is acceptable. + format: date-time + type: string + message: + description: |- + message is a human readable message indicating details about the transition. + This may be an empty string. + maxLength: 32768 + type: string + observedGeneration: + description: |- + observedGeneration represents the .metadata.generation that the condition was set based upon. + For instance, if .metadata.generation is currently 12, but the .status.conditions[x].observedGeneration is 9, the condition is out of date + with respect to the current state of the instance. + format: int64 + minimum: 0 + type: integer + reason: + description: |- + reason contains a programmatic identifier indicating the reason for the condition's last transition. + Producers of specific condition types may define expected values and meanings for this field, + and whether the values are considered a guaranteed API. + The value should be a CamelCase string. + This field may not be empty. + maxLength: 1024 + minLength: 1 + pattern: ^[A-Za-z]([A-Za-z0-9_,:]*[A-Za-z0-9_])?$ + type: string + status: + description: status of the condition, one of True, False, Unknown. + enum: + - "True" + - "False" + - Unknown + type: string + type: + description: |- + type of condition in CamelCase or in foo.example.com/CamelCase. + --- + Many .condition.type values are consistent across resources like Available, but because arbitrary conditions can be + useful (see .node.status.conditions), the ability to deconflict is important. + The regex it matches is (dns1123SubdomainFmt/)?(qualifiedNameFmt) + maxLength: 316 + pattern: ^([a-z0-9]([-a-z0-9]*[a-z0-9])?(\.[a-z0-9]([-a-z0-9]*[a-z0-9])?)*/)?(([A-Za-z0-9][-A-Za-z0-9_.]*)?[A-Za-z0-9])$ + type: string + required: + - lastTransitionTime + - message + - reason + - status + - type + type: object + type: array + x-kubernetes-list-map-keys: + - type + x-kubernetes-list-type: map + required: + - conditions + type: object + required: + - metadata + - spec + type: object + served: true + storage: true + subresources: + status: {} +status: + acceptedNames: + kind: "" + plural: "" + conditions: null + storedVersions: null +--- +# Source: calico/templates/calico-kube-controllers-rbac.yaml +# Include a clusterrole for the kube-controllers component, +# and bind it to the calico-kube-controllers serviceaccount. +kind: ClusterRole +apiVersion: rbac.authorization.k8s.io/v1 +metadata: + name: calico-kube-controllers +rules: + # Nodes are watched to monitor for deletions. + - apiGroups: [""] + resources: + - nodes + verbs: + - watch + - list + - get + # Pods are watched to check for existence as part of IPAM controller. + - apiGroups: [""] + resources: + - pods + verbs: + - get + - list + - watch + # IPAM resources are manipulated in response to node and block updates, as well as periodic triggers. + - apiGroups: ["crd.projectcalico.org"] + resources: + - ipreservations + verbs: + - list + - apiGroups: ["crd.projectcalico.org"] + resources: + - blockaffinities + - ipamblocks + - ipamhandles + - tiers + verbs: + - get + - list + - create + - update + - delete + - watch + # Pools are watched to maintain a mapping of blocks to IP pools. + - apiGroups: ["crd.projectcalico.org"] + resources: + - ippools + verbs: + - list + - watch + # kube-controllers manages hostendpoints. + - apiGroups: ["crd.projectcalico.org"] + resources: + - hostendpoints + verbs: + - get + - list + - create + - update + - delete + # Needs access to update clusterinformations. + - apiGroups: ["crd.projectcalico.org"] + resources: + - clusterinformations + verbs: + - get + - list + - create + - update + - watch + # KubeControllersConfiguration is where it gets its config + - apiGroups: ["crd.projectcalico.org"] + resources: + - kubecontrollersconfigurations + verbs: + # read its own config + - get + # create a default if none exists + - create + # update status + - update + # watch for changes + - watch +--- +# Source: calico/templates/calico-node-rbac.yaml +# Include a clusterrole for the calico-node DaemonSet, +# and bind it to the calico-node serviceaccount. +kind: ClusterRole +apiVersion: rbac.authorization.k8s.io/v1 +metadata: + name: calico-node +rules: + # Used for creating service account tokens to be used by the CNI plugin + - apiGroups: [""] + resources: + - serviceaccounts/token + resourceNames: + - calico-cni-plugin + verbs: + - create + # The CNI plugin needs to get pods, nodes, and namespaces. + - apiGroups: [""] + resources: + - pods + - nodes + - namespaces + verbs: + - get + # EndpointSlices are used for Service-based network policy rule + # enforcement. + - apiGroups: ["discovery.k8s.io"] + resources: + - endpointslices + verbs: + - watch + - list + - apiGroups: [""] + resources: + - endpoints + - services + verbs: + # Used to discover service IPs for advertisement. + - watch + - list + # Used to discover Typhas. + - get + # Pod CIDR auto-detection on kubeadm needs access to config maps. + - apiGroups: [""] + resources: + - configmaps + verbs: + - get + - apiGroups: [""] + resources: + - nodes/status + verbs: + # Needed for clearing NodeNetworkUnavailable flag. + - patch + # Calico stores some configuration information in node annotations. + - update + # Watch for changes to Kubernetes NetworkPolicies. + - apiGroups: ["networking.k8s.io"] + resources: + - networkpolicies + verbs: + - watch + - list + # Watch for changes to Kubernetes AdminNetworkPolicies. + - apiGroups: ["policy.networking.k8s.io"] + resources: + - adminnetworkpolicies + verbs: + - watch + - list + # Used by Calico for policy information. + - apiGroups: [""] + resources: + - pods + - namespaces + - serviceaccounts + verbs: + - list + - watch + # The CNI plugin patches pods/status. + - apiGroups: [""] + resources: + - pods/status + verbs: + - patch + # Calico monitors various CRDs for config. + - apiGroups: ["crd.projectcalico.org"] + resources: + - globalfelixconfigs + - felixconfigurations + - bgppeers + - bgpfilters + - globalbgpconfigs + - bgpconfigurations + - ippools + - ipreservations + - ipamblocks + - globalnetworkpolicies + - globalnetworksets + - networkpolicies + - networksets + - clusterinformations + - hostendpoints + - blockaffinities + - caliconodestatuses + - tiers + verbs: + - get + - list + - watch + # Calico creates some tiers on startup. + - apiGroups: ["crd.projectcalico.org"] + resources: + - tiers + verbs: + - create + # Calico must create and update some CRDs on startup. + - apiGroups: ["crd.projectcalico.org"] + resources: + - ippools + - felixconfigurations + - clusterinformations + verbs: + - create + - update + # Calico must update some CRDs. + - apiGroups: ["crd.projectcalico.org"] + resources: + - caliconodestatuses + verbs: + - update + # Calico stores some configuration information on the node. + - apiGroups: [""] + resources: + - nodes + verbs: + - get + - list + - watch + # These permissions are only required for upgrade from v2.6, and can + # be removed after upgrade or on fresh installations. + - apiGroups: ["crd.projectcalico.org"] + resources: + - bgpconfigurations + - bgppeers + verbs: + - create + - update + # These permissions are required for Calico CNI to perform IPAM allocations. + - apiGroups: ["crd.projectcalico.org"] + resources: + - blockaffinities + - ipamblocks + - ipamhandles + verbs: + - get + - list + - create + - update + - delete + # The CNI plugin and calico/node need to be able to create a default + # IPAMConfiguration + - apiGroups: ["crd.projectcalico.org"] + resources: + - ipamconfigs + verbs: + - get + - create + # Block affinities must also be watchable by confd for route aggregation. + - apiGroups: ["crd.projectcalico.org"] + resources: + - blockaffinities + verbs: + - watch + # The Calico IPAM migration needs to get daemonsets. These permissions can be + # removed if not upgrading from an installation using host-local IPAM. + - apiGroups: ["apps"] + resources: + - daemonsets + verbs: + - get +--- +# Source: calico/templates/calico-node-rbac.yaml +# CNI cluster role +kind: ClusterRole +apiVersion: rbac.authorization.k8s.io/v1 +metadata: + name: calico-cni-plugin +rules: + - apiGroups: [""] + resources: + - pods + - nodes + - namespaces + verbs: + - get + - apiGroups: [""] + resources: + - pods/status + verbs: + - patch + - apiGroups: ["crd.projectcalico.org"] + resources: + - blockaffinities + - ipamblocks + - ipamhandles + - clusterinformations + - ippools + - ipreservations + - ipamconfigs + verbs: + - get + - list + - create + - update + - delete +--- +# Source: calico/templates/calico-kube-controllers-rbac.yaml +kind: ClusterRoleBinding +apiVersion: rbac.authorization.k8s.io/v1 +metadata: + name: calico-kube-controllers +roleRef: + apiGroup: rbac.authorization.k8s.io + kind: ClusterRole + name: calico-kube-controllers +subjects: +- kind: ServiceAccount + name: calico-kube-controllers + namespace: kube-system +--- +# Source: calico/templates/calico-node-rbac.yaml +apiVersion: rbac.authorization.k8s.io/v1 +kind: ClusterRoleBinding +metadata: + name: calico-node +roleRef: + apiGroup: rbac.authorization.k8s.io + kind: ClusterRole + name: calico-node +subjects: +- kind: ServiceAccount + name: calico-node + namespace: kube-system +--- +# Source: calico/templates/calico-node-rbac.yaml +apiVersion: rbac.authorization.k8s.io/v1 +kind: ClusterRoleBinding +metadata: + name: calico-cni-plugin +roleRef: + apiGroup: rbac.authorization.k8s.io + kind: ClusterRole + name: calico-cni-plugin +subjects: +- kind: ServiceAccount + name: calico-cni-plugin + namespace: kube-system +--- +# Source: calico/templates/calico-typha.yaml +# This manifest creates a Service, which will be backed by Calico's Typha daemon. +# Typha sits in between Felix and the API server, reducing Calico's load on the API server. + +apiVersion: v1 +kind: Service +metadata: + name: calico-typha + namespace: kube-system + labels: + k8s-app: calico-typha +spec: + ports: + - port: 5473 + protocol: TCP + targetPort: calico-typha + name: calico-typha + selector: + k8s-app: calico-typha +--- +# Source: calico/templates/calico-node.yaml +# This manifest installs the calico-node container, as well +# as the CNI plugins and network config on +# each master and worker node in a Kubernetes cluster. +kind: DaemonSet +apiVersion: apps/v1 +metadata: + name: calico-node + namespace: kube-system + labels: + k8s-app: calico-node +spec: + selector: + matchLabels: + k8s-app: calico-node + updateStrategy: + type: RollingUpdate + rollingUpdate: + maxUnavailable: 1 + template: + metadata: + labels: + k8s-app: calico-node + spec: + nodeSelector: + kubernetes.io/os: linux + hostNetwork: true + tolerations: + # Make sure calico-node gets scheduled on all nodes. + - effect: NoSchedule + operator: Exists + # Mark the pod as a critical add-on for rescheduling. + - key: CriticalAddonsOnly + operator: Exists + - effect: NoExecute + operator: Exists + serviceAccountName: calico-node + securityContext: + seccompProfile: + type: RuntimeDefault + # Minimize downtime during a rolling upgrade or deletion; tell Kubernetes to do a "force + # deletion": https://kubernetes.io/docs/concepts/workloads/pods/pod/#termination-of-pods. + terminationGracePeriodSeconds: 0 + priorityClassName: system-node-critical + initContainers: + # This container performs upgrade from host-local IPAM to calico-ipam. + # It can be deleted if this is a fresh installation, or if you have already + # upgraded to use calico-ipam. + - name: upgrade-ipam + image: docker.io/calico/cni:v3.29.1 + imagePullPolicy: IfNotPresent + command: ["/opt/cni/bin/calico-ipam", "-upgrade"] + envFrom: + - configMapRef: + # Allow KUBERNETES_SERVICE_HOST and KUBERNETES_SERVICE_PORT to be overridden for eBPF mode. + name: kubernetes-services-endpoint + optional: true + env: + - name: KUBERNETES_NODE_NAME + valueFrom: + fieldRef: + fieldPath: spec.nodeName + - name: CALICO_NETWORKING_BACKEND + valueFrom: + configMapKeyRef: + name: calico-config + key: calico_backend + volumeMounts: + - mountPath: /var/lib/cni/networks + name: host-local-net-dir + - mountPath: /host/opt/cni/bin + name: cni-bin-dir + securityContext: + privileged: true + # This container installs the CNI binaries + # and CNI network config file on each node. + - name: install-cni + image: docker.io/calico/cni:v3.29.1 + imagePullPolicy: IfNotPresent + command: ["/opt/cni/bin/install"] + envFrom: + - configMapRef: + # Allow KUBERNETES_SERVICE_HOST and KUBERNETES_SERVICE_PORT to be overridden for eBPF mode. + name: kubernetes-services-endpoint + optional: true + env: + # Name of the CNI config file to create. + - name: CNI_CONF_NAME + value: "10-calico.conflist" + # The CNI network config to install on each node. + - name: CNI_NETWORK_CONFIG + valueFrom: + configMapKeyRef: + name: calico-config + key: cni_network_config + # Set the hostname based on the k8s node name. + - name: KUBERNETES_NODE_NAME + valueFrom: + fieldRef: + fieldPath: spec.nodeName + # CNI MTU Config variable + - name: CNI_MTU + valueFrom: + configMapKeyRef: + name: calico-config + key: veth_mtu + # Prevents the container from sleeping forever. + - name: SLEEP + value: "false" + volumeMounts: + - mountPath: /host/opt/cni/bin + name: cni-bin-dir + - mountPath: /host/etc/cni/net.d + name: cni-net-dir + securityContext: + privileged: true + # This init container mounts the necessary filesystems needed by the BPF data plane + # i.e. bpf at /sys/fs/bpf and cgroup2 at /run/calico/cgroup. Calico-node initialisation is executed + # in best effort fashion, i.e. no failure for errors, to not disrupt pod creation in iptable mode. + - name: "mount-bpffs" + image: docker.io/calico/node:v3.29.1 + imagePullPolicy: IfNotPresent + command: ["calico-node", "-init", "-best-effort"] + volumeMounts: + - mountPath: /sys/fs + name: sys-fs + # Bidirectional is required to ensure that the new mount we make at /sys/fs/bpf propagates to the host + # so that it outlives the init container. + mountPropagation: Bidirectional + - mountPath: /var/run/calico + name: var-run-calico + # Bidirectional is required to ensure that the new mount we make at /run/calico/cgroup propagates to the host + # so that it outlives the init container. + mountPropagation: Bidirectional + # Mount /proc/ from host which usually is an init program at /nodeproc. It's needed by mountns binary, + # executed by calico-node, to mount root cgroup2 fs at /run/calico/cgroup to attach CTLB programs correctly. + - mountPath: /nodeproc + name: nodeproc + readOnly: true + securityContext: + privileged: true + containers: + # Runs calico-node container on each Kubernetes node. This + # container programs network policy and routes on each + # host. + - name: calico-node + image: docker.io/calico/node:v3.29.1 + imagePullPolicy: IfNotPresent + envFrom: + - configMapRef: + # Allow KUBERNETES_SERVICE_HOST and KUBERNETES_SERVICE_PORT to be overridden for eBPF mode. + name: kubernetes-services-endpoint + optional: true + env: + # Use Kubernetes API as the backing datastore. + - name: DATASTORE_TYPE + value: "kubernetes" + # Typha support: controlled by the ConfigMap. + - name: FELIX_TYPHAK8SSERVICENAME + valueFrom: + configMapKeyRef: + name: calico-config + key: typha_service_name + # Wait for the datastore. + - name: WAIT_FOR_DATASTORE + value: "true" + # Set based on the k8s node name. + - name: NODENAME + valueFrom: + fieldRef: + fieldPath: spec.nodeName + # Choose the backend to use. + - name: CALICO_NETWORKING_BACKEND + valueFrom: + configMapKeyRef: + name: calico-config + key: calico_backend + # Cluster type to identify the deployment type + - name: CLUSTER_TYPE + value: "k8s,bgp" + # Auto-detect the BGP IP address. + - name: IP + value: "autodetect" + # Enable IPIP + - name: CALICO_IPV4POOL_IPIP + value: "Always" + # Enable or Disable VXLAN on the default IP pool. + - name: CALICO_IPV4POOL_VXLAN + value: "Never" + # Enable or Disable VXLAN on the default IPv6 IP pool. + - name: CALICO_IPV6POOL_VXLAN + value: "Never" + # Set MTU for tunnel device used if ipip is enabled + - name: FELIX_IPINIPMTU + valueFrom: + configMapKeyRef: + name: calico-config + key: veth_mtu + # Set MTU for the VXLAN tunnel device. + - name: FELIX_VXLANMTU + valueFrom: + configMapKeyRef: + name: calico-config + key: veth_mtu + # Set MTU for the Wireguard tunnel device. + - name: FELIX_WIREGUARDMTU + valueFrom: + configMapKeyRef: + name: calico-config + key: veth_mtu + # The default IPv4 pool to create on startup if none exists. Pod IPs will be + # chosen from this range. Changing this value after installation will have + # no effect. This should fall within `--cluster-cidr`. + # - name: CALICO_IPV4POOL_CIDR + # value: "192.168.0.0/16" + # Disable file logging so `kubectl logs` works. + - name: CALICO_DISABLE_FILE_LOGGING + value: "true" + # Set Felix endpoint to host default action to ACCEPT. + - name: FELIX_DEFAULTENDPOINTTOHOSTACTION + value: "ACCEPT" + # Disable IPv6 on Kubernetes. + - name: FELIX_IPV6SUPPORT + value: "false" + - name: FELIX_HEALTHENABLED + value: "true" + securityContext: + privileged: true + resources: + requests: + cpu: 250m + lifecycle: + preStop: + exec: + command: + - /bin/calico-node + - -shutdown + livenessProbe: + exec: + command: + - /bin/calico-node + - -felix-live + - -bird-live + periodSeconds: 10 + initialDelaySeconds: 10 + failureThreshold: 6 + timeoutSeconds: 10 + readinessProbe: + exec: + command: + - /bin/calico-node + - -felix-ready + - -bird-ready + periodSeconds: 10 + timeoutSeconds: 10 + volumeMounts: + # For maintaining CNI plugin API credentials. + - mountPath: /host/etc/cni/net.d + name: cni-net-dir + readOnly: false + - mountPath: /lib/modules + name: lib-modules + readOnly: true + - mountPath: /run/xtables.lock + name: xtables-lock + readOnly: false + - mountPath: /var/run/calico + name: var-run-calico + readOnly: false + - mountPath: /var/lib/calico + name: var-lib-calico + readOnly: false + - name: policysync + mountPath: /var/run/nodeagent + # For eBPF mode, we need to be able to mount the BPF filesystem at /sys/fs/bpf so we mount in the + # parent directory. + - name: bpffs + mountPath: /sys/fs/bpf + - name: cni-log-dir + mountPath: /var/log/calico/cni + readOnly: true + volumes: + # Used by calico-node. + - name: lib-modules + hostPath: + path: /lib/modules + - name: var-run-calico + hostPath: + path: /var/run/calico + type: DirectoryOrCreate + - name: var-lib-calico + hostPath: + path: /var/lib/calico + type: DirectoryOrCreate + - name: xtables-lock + hostPath: + path: /run/xtables.lock + type: FileOrCreate + - name: sys-fs + hostPath: + path: /sys/fs/ + type: DirectoryOrCreate + - name: bpffs + hostPath: + path: /sys/fs/bpf + type: Directory + # mount /proc at /nodeproc to be used by mount-bpffs initContainer to mount root cgroup2 fs. + - name: nodeproc + hostPath: + path: /proc + # Used to install CNI. + - name: cni-bin-dir + hostPath: + path: /opt/cni/bin + type: DirectoryOrCreate + - name: cni-net-dir + hostPath: + path: /etc/cni/net.d + # Used to access CNI logs. + - name: cni-log-dir + hostPath: + path: /var/log/calico/cni + # Mount in the directory for host-local IPAM allocations. This is + # used when upgrading from host-local to calico-ipam, and can be removed + # if not using the upgrade-ipam init container. + - name: host-local-net-dir + hostPath: + path: /var/lib/cni/networks + # Used to create per-pod Unix Domain Sockets + - name: policysync + hostPath: + type: DirectoryOrCreate + path: /var/run/nodeagent +--- +# Source: calico/templates/calico-kube-controllers.yaml +# See https://github.com/projectcalico/kube-controllers +apiVersion: apps/v1 +kind: Deployment +metadata: + name: calico-kube-controllers + namespace: kube-system + labels: + k8s-app: calico-kube-controllers +spec: + # The controllers can only have a single active instance. + replicas: 1 + selector: + matchLabels: + k8s-app: calico-kube-controllers + strategy: + type: Recreate + template: + metadata: + name: calico-kube-controllers + namespace: kube-system + labels: + k8s-app: calico-kube-controllers + spec: + nodeSelector: + kubernetes.io/os: linux + tolerations: + # Mark the pod as a critical add-on for rescheduling. + - key: CriticalAddonsOnly + operator: Exists + - key: node-role.kubernetes.io/master + effect: NoSchedule + - key: node-role.kubernetes.io/control-plane + effect: NoSchedule + serviceAccountName: calico-kube-controllers + securityContext: + seccompProfile: + type: RuntimeDefault + priorityClassName: system-cluster-critical + containers: + - name: calico-kube-controllers + image: docker.io/calico/kube-controllers:v3.29.1 + imagePullPolicy: IfNotPresent + env: + # Choose which controllers to run. + - name: ENABLED_CONTROLLERS + value: node + - name: DATASTORE_TYPE + value: kubernetes + livenessProbe: + exec: + command: + - /usr/bin/check-status + - -l + periodSeconds: 10 + initialDelaySeconds: 10 + failureThreshold: 6 + timeoutSeconds: 10 + readinessProbe: + exec: + command: + - /usr/bin/check-status + - -r + periodSeconds: 10 + securityContext: + runAsNonRoot: true +--- +# Source: calico/templates/calico-typha.yaml +# This manifest creates a Deployment of Typha to back the above service. + +apiVersion: apps/v1 +kind: Deployment +metadata: + name: calico-typha + namespace: kube-system + labels: + k8s-app: calico-typha +spec: + # Number of Typha replicas. To enable Typha, set this to a non-zero value *and* set the + # typha_service_name variable in the calico-config ConfigMap above. + # + # We recommend using Typha if you have more than 50 nodes. Above 100 nodes it is essential + # (when using the Kubernetes datastore). Use one replica for every 100-200 nodes. In + # production, we recommend running at least 3 replicas to reduce the impact of rolling upgrade. + replicas: 1 + revisionHistoryLimit: 2 + selector: + matchLabels: + k8s-app: calico-typha + strategy: + rollingUpdate: + # 100% surge allows a complete up-level set of typha instances to start and become ready, + # which in turn allows all the back-level typha instances to start shutting down. This + # means that connections tend to bounce directly from a back-level instance to an up-level + # instance. + maxSurge: 100% + # In case the cluster is unable to schedule extra surge instances, allow at most one instance + # to shut down to make room. You can set this to 0 if you're sure there'll always be enough room to + # schedule extra typha instances during an upgrade (because setting it to 0 blocks shutdown until + # up-level typha instances are online and ready). + maxUnavailable: 1 + type: RollingUpdate + template: + metadata: + labels: + k8s-app: calico-typha + annotations: + cluster-autoscaler.kubernetes.io/safe-to-evict: 'true' + spec: + nodeSelector: + kubernetes.io/os: linux + hostNetwork: true + # Typha supports graceful shut down, disconnecting clients slowly during the grace period. + # The TYPHA_SHUTDOWNTIMEOUTSECS env var should be kept in sync with this value. + terminationGracePeriodSeconds: 300 + tolerations: + # Mark the pod as a critical add-on for rescheduling. + - key: CriticalAddonsOnly + operator: Exists + # Make sure Typha can get scheduled on any nodes. + - effect: NoSchedule + operator: Exists + - effect: NoExecute + operator: Exists + # Since Calico can't network a pod until Typha is up, we need to run Typha itself + # as a host-networked pod. + serviceAccountName: calico-node + priorityClassName: system-cluster-critical + # fsGroup allows using projected serviceaccount tokens as described here kubernetes/kubernetes#82573 + securityContext: + fsGroup: 65534 + seccompProfile: + type: RuntimeDefault + containers: + - image: docker.io/calico/typha:v3.29.1 + imagePullPolicy: IfNotPresent + name: calico-typha + ports: + - containerPort: 5473 + name: calico-typha + protocol: TCP + envFrom: + - configMapRef: + # Allow KUBERNETES_SERVICE_HOST and KUBERNETES_SERVICE_PORT to be overridden for eBPF mode. + name: kubernetes-services-endpoint + optional: true + env: + # Enable "info" logging by default. Can be set to "debug" to increase verbosity. + - name: TYPHA_LOGSEVERITYSCREEN + value: "info" + # Disable logging to file and syslog since those don't make sense in Kubernetes. + - name: TYPHA_LOGFILEPATH + value: "none" + - name: TYPHA_LOGSEVERITYSYS + value: "none" + # Monitor the Kubernetes API to find the number of running instances and rebalance + # connections. + - name: TYPHA_CONNECTIONREBALANCINGMODE + value: "kubernetes" + - name: TYPHA_DATASTORETYPE + value: "kubernetes" + - name: TYPHA_HEALTHENABLED + value: "true" + # Set this to the same value as terminationGracePeriodSeconds; it tells Typha how much time + # it has to shut down. + - name: TYPHA_SHUTDOWNTIMEOUTSECS + value: "300" + # Uncomment these lines to enable prometheus metrics. Since Typha is host-networked, + # this opens a port on the host, which may need to be secured. + #- name: TYPHA_PROMETHEUSMETRICSENABLED + # value: "true" + #- name: TYPHA_PROMETHEUSMETRICSPORT + # value: "9093" + livenessProbe: + httpGet: + path: /liveness + port: 9098 + host: localhost + periodSeconds: 30 + initialDelaySeconds: 30 + timeoutSeconds: 10 + securityContext: + runAsNonRoot: true + allowPrivilegeEscalation: false + readinessProbe: + httpGet: + path: /readiness + port: 9098 + host: localhost + periodSeconds: 10 + timeoutSeconds: 10 diff --git a/kubemarine/resources/configurations/compatibility/internal/kubernetes_images.yaml b/kubemarine/resources/configurations/compatibility/internal/kubernetes_images.yaml index d28333de5..6aa8f4c31 100644 --- a/kubemarine/resources/configurations/compatibility/internal/kubernetes_images.yaml +++ b/kubemarine/resources/configurations/compatibility/internal/kubernetes_images.yaml @@ -1,14 +1,6 @@ # !!!Do not change the content of the file!!! # This configuration is automatically generated and maintained by "scripts/thirdparties/sync.py" kube-apiserver: - v1.26.3: - version: v1.26.3 - v1.26.4: - version: v1.26.4 - v1.26.7: - version: v1.26.7 - v1.26.11: - version: v1.26.11 v1.27.1: version: v1.27.1 v1.27.4: @@ -44,14 +36,6 @@ kube-apiserver: v1.31.1: version: v1.31.1 kube-controller-manager: - v1.26.3: - version: v1.26.3 - v1.26.4: - version: v1.26.4 - v1.26.7: - version: v1.26.7 - v1.26.11: - version: v1.26.11 v1.27.1: version: v1.27.1 v1.27.4: @@ -87,14 +71,6 @@ kube-controller-manager: v1.31.1: version: v1.31.1 kube-scheduler: - v1.26.3: - version: v1.26.3 - v1.26.4: - version: v1.26.4 - v1.26.7: - version: v1.26.7 - v1.26.11: - version: v1.26.11 v1.27.1: version: v1.27.1 v1.27.4: @@ -130,14 +106,6 @@ kube-scheduler: v1.31.1: version: v1.31.1 kube-proxy: - v1.26.3: - version: v1.26.3 - v1.26.4: - version: v1.26.4 - v1.26.7: - version: v1.26.7 - v1.26.11: - version: v1.26.11 v1.27.1: version: v1.27.1 v1.27.4: @@ -173,14 +141,6 @@ kube-proxy: v1.31.1: version: v1.31.1 pause: - v1.26.3: - version: '3.9' - v1.26.4: - version: '3.9' - v1.26.7: - version: '3.9' - v1.26.11: - version: '3.9' v1.27.1: version: '3.9' v1.27.4: @@ -216,14 +176,6 @@ pause: v1.31.1: version: '3.10' etcd: - v1.26.3: - version: 3.5.6-0 - v1.26.4: - version: 3.5.6-0 - v1.26.7: - version: 3.5.6-0 - v1.26.11: - version: 3.5.9-0 v1.27.1: version: 3.5.7-0 v1.27.4: @@ -259,14 +211,6 @@ etcd: v1.31.1: version: 3.5.15-0 coredns/coredns: - v1.26.3: - version: v1.9.3 - v1.26.4: - version: v1.9.3 - v1.26.7: - version: v1.9.3 - v1.26.11: - version: v1.9.3 v1.27.1: version: v1.10.1 v1.27.4: diff --git a/kubemarine/resources/configurations/compatibility/internal/packages.yaml b/kubemarine/resources/configurations/compatibility/internal/packages.yaml index e34f949c2..699fc8050 100644 --- a/kubemarine/resources/configurations/compatibility/internal/packages.yaml +++ b/kubemarine/resources/configurations/compatibility/internal/packages.yaml @@ -5,14 +5,6 @@ # # After any change, kubemarine/patches/software_upgrade.yaml should also be manually changed. containerd: - v1.26.3: - version_debian: 1.6.* - v1.26.4: - version_debian: 1.6.* - v1.26.7: - version_debian: 1.6.* - v1.26.11: - version_debian: 1.6.* v1.27.1: version_debian: 1.6.* v1.27.4: @@ -48,22 +40,6 @@ containerd: v1.31.1: version_debian: 1.6.* containerdio: - v1.26.3: - version_rhel: 1.6* - version_rhel8: 1.6* - version_rhel9: 1.6* - v1.26.4: - version_rhel: 1.6* - version_rhel8: 1.6* - version_rhel9: 1.6* - v1.26.7: - version_rhel: 1.6* - version_rhel8: 1.6* - version_rhel9: 1.6* - v1.26.11: - version_rhel: 1.6* - version_rhel8: 1.6* - version_rhel9: 1.6* v1.27.1: version_rhel: 1.6* version_rhel8: 1.6* diff --git a/kubemarine/resources/configurations/compatibility/internal/plugins.yaml b/kubemarine/resources/configurations/compatibility/internal/plugins.yaml index f36c893df..6bc0ab4fa 100644 --- a/kubemarine/resources/configurations/compatibility/internal/plugins.yaml +++ b/kubemarine/resources/configurations/compatibility/internal/plugins.yaml @@ -5,14 +5,6 @@ # This is done for simplicity as the configuration is anyway generated, but this can be changed in the future. calico: - v1.26.3: - version: v3.24.2 - v1.26.4: - version: v3.25.1 - v1.26.7: - version: v3.26.1 - v1.26.11: - version: v3.26.4 v1.27.1: version: v3.26.4 v1.27.4: @@ -34,32 +26,20 @@ calico: v1.28.9: version: v3.27.3 v1.28.12: - version: v3.28.1 + version: v3.29.1 v1.29.1: version: v3.27.0 v1.29.4: version: v3.27.3 v1.29.7: - version: v3.28.1 + version: v3.29.1 v1.30.1: version: v3.27.3 v1.30.3: - version: v3.28.1 + version: v3.29.1 v1.31.1: - version: v3.28.1 + version: v3.29.1 nginx-ingress-controller: - v1.26.3: - version: v1.4.0 - webhook-version: v20220916-gd32f8c343 - v1.26.4: - version: v1.7.0 - webhook-version: v20230312-helm-chart-4.5.2-28-g66a760794 - v1.26.7: - version: v1.8.1 - webhook-version: v20230407 - v1.26.11: - version: v1.8.4 - webhook-version: v20231011-8b53cabe0 v1.27.1: version: v1.8.4 webhook-version: v20231011-8b53cabe0 @@ -112,18 +92,6 @@ nginx-ingress-controller: version: v1.11.1 webhook-version: v1.4.1 kubernetes-dashboard: - v1.26.3: - version: v2.7.0 - metrics-scraper-version: v1.0.8 - v1.26.4: - version: v2.7.0 - metrics-scraper-version: v1.0.8 - v1.26.7: - version: v2.7.0 - metrics-scraper-version: v1.0.8 - v1.26.11: - version: v2.7.0 - metrics-scraper-version: v1.0.8 v1.27.1: version: v2.7.0 metrics-scraper-version: v1.0.8 @@ -176,18 +144,6 @@ kubernetes-dashboard: version: v2.7.0 metrics-scraper-version: v1.0.8 local-path-provisioner: - v1.26.3: - version: v0.0.23 - busybox-version: 1.34.1 - v1.26.4: - version: v0.0.24 - busybox-version: 1.34.1 - v1.26.7: - version: v0.0.24 - busybox-version: 1.34.1 - v1.26.11: - version: v0.0.25 - busybox-version: 1.34.1 v1.27.1: version: v0.0.25 busybox-version: 1.34.1 diff --git a/kubemarine/resources/configurations/compatibility/internal/thirdparties.yaml b/kubemarine/resources/configurations/compatibility/internal/thirdparties.yaml index 8ca725d20..768926ac6 100644 --- a/kubemarine/resources/configurations/compatibility/internal/thirdparties.yaml +++ b/kubemarine/resources/configurations/compatibility/internal/thirdparties.yaml @@ -2,14 +2,6 @@ # This configuration is automatically generated and maintained by "scripts/thirdparties/sync.py" kubeadm: - v1.26.3: - sha1: 86e202f98d22c8fddcadda6656f6698d21eae6ca - v1.26.4: - sha1: 7b2f579d06d00d545b2dc279df837d3c9d3e7ae6 - v1.26.7: - sha1: 623952b584e5604da1e3e9e52c6da8dbecadf361 - v1.26.11: - sha1: 3d634ce0bf8db1ab5a9ea723443e7bce042c865b v1.27.1: sha1: d178257fa65dfa4694dea3262c618c6622804eb7 v1.27.4: @@ -45,14 +37,6 @@ kubeadm: v1.31.1: sha1: 0a682af6436ce4e7188f93ddeebff5c2f3be1592 kubelet: - v1.26.3: - sha1: 5fe320fedaabb91d3770da19135412b7454bb28b - v1.26.4: - sha1: 07fc5049dd34744d1c45258ed23291c3e58d4abf - v1.26.7: - sha1: 4e82efbb2f5880b5eec10ce4e73ba6f87ad9fbbe - v1.26.11: - sha1: 497d7785658f4b247404580a821c22958f7a4b93 v1.27.1: sha1: 43cb7231a889c01cfd88bd3f27441134e3d1cbff v1.27.4: @@ -88,14 +72,6 @@ kubelet: v1.31.1: sha1: fc7d0a9859c97ec2a2a4ac9ec1814b131e8d875f kubectl: - v1.26.3: - sha1: 56916d87c3caef05489db932fd9e48d32ebdf634 - v1.26.4: - sha1: b4880f3423aaf68992cdac384289739ff810a07a - v1.26.7: - sha1: 33c088e6d025b90e4311c4823d712d4b7a32e886 - v1.26.11: - sha1: df1fa17ce624c2a3e6a05b26157f93719b640bfa v1.27.1: sha1: 0c3f1e262a6c37719ba4d66885df6715f58b60e5 v1.27.4: @@ -133,18 +109,6 @@ kubectl: calicoctl: # calicoctl version is duplicated from kubemarine/resources/configurations/compatibility/kubernetes_versions.yaml # It also corresponds to the plugin version in kubemarine/resources/configurations/compatibility/internal/plugins.yaml - v1.26.3: - version: v3.24.2 - sha1: c4de7a203e5a3a942fdf130bc9ec180111fc2ab6 - v1.26.4: - version: v3.25.1 - sha1: bc3f20a899aca3f62451bb7d9b015b9805305216 - v1.26.7: - version: v3.26.1 - sha1: 4b9896ef195ec4fc4ff0d255c81960ebb5750922 - v1.26.11: - version: v3.26.4 - sha1: 46875b3d28318553fe382db0766a0916f2556217 v1.27.1: version: v3.26.4 sha1: 46875b3d28318553fe382db0766a0916f2556217 @@ -176,8 +140,8 @@ calicoctl: version: v3.27.3 sha1: 24468ab467fd59727d278dfdc3a5c6eec51cdff1 v1.28.12: - version: v3.28.1 - sha1: 80f164a8248b8fa501f35f12cbcac0b059e665e2 + version: v3.29.1 + sha1: 00be749d257eee5035d3ba408aca15fcaf8be7c2 v1.29.1: version: v3.27.0 sha1: 4d62cba82a4aee97ab20b96e7270da85d77ce20e @@ -185,32 +149,20 @@ calicoctl: version: v3.27.3 sha1: 24468ab467fd59727d278dfdc3a5c6eec51cdff1 v1.29.7: - version: v3.28.1 - sha1: 80f164a8248b8fa501f35f12cbcac0b059e665e2 + version: v3.29.1 + sha1: 00be749d257eee5035d3ba408aca15fcaf8be7c2 v1.30.1: version: v3.27.3 sha1: 24468ab467fd59727d278dfdc3a5c6eec51cdff1 v1.30.3: - version: v3.28.1 - sha1: 80f164a8248b8fa501f35f12cbcac0b059e665e2 + version: v3.29.1 + sha1: 00be749d257eee5035d3ba408aca15fcaf8be7c2 v1.31.1: - version: v3.28.1 - sha1: 80f164a8248b8fa501f35f12cbcac0b059e665e2 + version: v3.29.1 + sha1: 00be749d257eee5035d3ba408aca15fcaf8be7c2 crictl: # crictl version is duplicated from kubemarine/resources/configurations/compatibility/kubernetes_versions.yaml # for backward compatibility with clusters in a private environment. - v1.26.3: - version: v1.25.0 - sha1: b3a24e549ca3b4dfd105b7f4639014c0c508bea3 - v1.26.4: - version: v1.27.0 - sha1: 69b4973cd4542224c1790d1847e0e9a504f1a345 - v1.26.7: - version: v1.27.1 - sha1: c350624a1b8dff38dd58447d63518dc330cc5c0f - v1.26.11: - version: v1.28.0 - sha1: 1199411456ab5a1e0dd9524724f15e92aa3f9da7 v1.27.1: version: v1.28.0 sha1: 1199411456ab5a1e0dd9524724f15e92aa3f9da7 diff --git a/kubemarine/resources/configurations/compatibility/kubernetes_versions.yaml b/kubemarine/resources/configurations/compatibility/kubernetes_versions.yaml index 6b3749200..8a7a72020 100644 --- a/kubemarine/resources/configurations/compatibility/kubernetes_versions.yaml +++ b/kubemarine/resources/configurations/compatibility/kubernetes_versions.yaml @@ -1,7 +1,5 @@ # This section can be changed manually, but it is also synchronized automatically with the 'compatibility_map' section below. kubernetes_versions: - v1.26: - supported: false v1.27: supported: true v1.28: @@ -14,30 +12,6 @@ kubernetes_versions: supported: true compatibility_map: # This section should be changed manually. - v1.26.3: - calico: v3.24.2 - nginx-ingress-controller: v1.4.0 - kubernetes-dashboard: v2.7.0 - local-path-provisioner: v0.0.23 - crictl: v1.25.0 - v1.26.4: - calico: v3.25.1 - nginx-ingress-controller: v1.7.0 - kubernetes-dashboard: v2.7.0 - local-path-provisioner: v0.0.24 - crictl: v1.27.0 - v1.26.7: - calico: v3.26.1 - nginx-ingress-controller: v1.8.1 - kubernetes-dashboard: v2.7.0 - local-path-provisioner: v0.0.24 - crictl: v1.27.1 - v1.26.11: - calico: v3.26.4 - nginx-ingress-controller: v1.8.4 - kubernetes-dashboard: v2.7.0 - local-path-provisioner: v0.0.25 - crictl: v1.28.0 v1.27.1: calico: v3.26.4 nginx-ingress-controller: v1.8.4 @@ -99,7 +73,7 @@ compatibility_map: local-path-provisioner: v0.0.26 crictl: v1.30.0 v1.28.12: - calico: v3.28.1 + calico: v3.29.1 nginx-ingress-controller: v1.11.1 kubernetes-dashboard: v2.7.0 local-path-provisioner: v0.0.27 @@ -117,7 +91,7 @@ compatibility_map: local-path-provisioner: v0.0.26 crictl: v1.30.0 v1.29.7: - calico: v3.28.1 + calico: v3.29.1 nginx-ingress-controller: v1.11.1 kubernetes-dashboard: v2.7.0 local-path-provisioner: v0.0.27 @@ -129,13 +103,13 @@ compatibility_map: local-path-provisioner: v0.0.26 crictl: v1.30.0 v1.30.3: - calico: v3.28.1 + calico: v3.29.1 nginx-ingress-controller: v1.11.1 kubernetes-dashboard: v2.7.0 local-path-provisioner: v0.0.27 crictl: v1.30.0 v1.31.1: - calico: v3.28.1 + calico: v3.29.1 nginx-ingress-controller: v1.11.1 kubernetes-dashboard: v2.7.0 local-path-provisioner: v0.0.27