cmd/k8s-operator,k8s-operator: Allow the use of multiple tailnets (#18344)
This commit contains the implementation of multi-tailnet support within the Kubernetes Operator
Each of our custom resources now expose the `spec.tailnet` field. This field is a string that must match the name of an existing `Tailnet` resource. A `Tailnet` resource looks like this:
```yaml
apiVersion: tailscale.com/v1alpha1
kind: Tailnet
metadata:
name: example # This is the name that must be referenced by other resources
spec:
credentials:
secretName: example-oauth
```
Each `Tailnet` references a `Secret` resource that contains a set of oauth credentials. This secret must be created in the same namespace as the operator:
```yaml
apiVersion: v1
kind: Secret
metadata:
name: example-oauth # This is the name that's referenced by the Tailnet resource.
namespace: tailscale
stringData:
client_id: "client-id"
client_secret: "client-secret"
```
When created, the operator performs a basic check that the oauth client has access to all required scopes. This is done using read actions on devices, keys & services. While this doesn't capture a missing "write" permission, it catches completely missing permissions. Once this check passes, the `Tailnet` moves into a ready state and can be referenced. Attempting to use a `Tailnet` in a non-ready state will stall the deployment of `Connector`s, `ProxyGroup`s and `Recorder`s until the `Tailnet` becomes ready.
The `spec.tailnet` field informs the operator that a `Connector`, `ProxyGroup`, or `Recorder` must be given an auth key generated using the specified oauth client. For backwards compatibility, the set of credentials the operator is configured with are considered the default. That is, where `spec.tailnet` is not set, the resource will be deployed in the same tailnet as the operator.
Updates https://github.com/tailscale/corp/issues/34561
This commit is contained in:
@@ -25,6 +25,7 @@ import (
|
||||
"k8s.io/client-go/tools/record"
|
||||
"sigs.k8s.io/controller-runtime/pkg/client"
|
||||
"sigs.k8s.io/controller-runtime/pkg/reconcile"
|
||||
|
||||
tsoperator "tailscale.com/k8s-operator"
|
||||
tsapi "tailscale.com/k8s-operator/apis/v1alpha1"
|
||||
"tailscale.com/kube/kubetypes"
|
||||
@@ -207,6 +208,7 @@ func (a *ConnectorReconciler) maybeProvisionConnector(ctx context.Context, logge
|
||||
ProxyClassName: proxyClass,
|
||||
proxyType: proxyTypeConnector,
|
||||
LoginServer: a.ssr.loginServer,
|
||||
Tailnet: cn.Spec.Tailnet,
|
||||
}
|
||||
|
||||
if cn.Spec.SubnetRouter != nil && len(cn.Spec.SubnetRouter.AdvertiseRoutes) > 0 {
|
||||
@@ -276,7 +278,7 @@ func (a *ConnectorReconciler) maybeProvisionConnector(ctx context.Context, logge
|
||||
}
|
||||
|
||||
func (a *ConnectorReconciler) maybeCleanupConnector(ctx context.Context, logger *zap.SugaredLogger, cn *tsapi.Connector) (bool, error) {
|
||||
if done, err := a.ssr.Cleanup(ctx, logger, childResourceLabels(cn.Name, a.tsnamespace, "connector"), proxyTypeConnector); err != nil {
|
||||
if done, err := a.ssr.Cleanup(ctx, cn.Spec.Tailnet, logger, childResourceLabels(cn.Name, a.tsnamespace, "connector"), proxyTypeConnector); err != nil {
|
||||
return false, fmt.Errorf("failed to cleanup Connector resources: %w", err)
|
||||
} else if !done {
|
||||
logger.Debugf("Connector cleanup not done yet, waiting for next reconcile")
|
||||
|
||||
@@ -725,7 +725,7 @@ tailscale.com/cmd/k8s-operator dependencies: (generated by github.com/tailscale/
|
||||
k8s.io/utils/net from k8s.io/apimachinery/pkg/util/net+
|
||||
k8s.io/utils/ptr from k8s.io/client-go/tools/cache+
|
||||
k8s.io/utils/trace from k8s.io/client-go/tools/cache
|
||||
sigs.k8s.io/controller-runtime/pkg/builder from tailscale.com/cmd/k8s-operator
|
||||
sigs.k8s.io/controller-runtime/pkg/builder from tailscale.com/cmd/k8s-operator+
|
||||
sigs.k8s.io/controller-runtime/pkg/cache from sigs.k8s.io/controller-runtime/pkg/cluster+
|
||||
sigs.k8s.io/controller-runtime/pkg/cache/internal from sigs.k8s.io/controller-runtime/pkg/cache
|
||||
sigs.k8s.io/controller-runtime/pkg/certwatcher from sigs.k8s.io/controller-runtime/pkg/metrics/server+
|
||||
@@ -821,10 +821,12 @@ tailscale.com/cmd/k8s-operator dependencies: (generated by github.com/tailscale/
|
||||
tailscale.com/ipn/store from tailscale.com/ipn/ipnlocal+
|
||||
tailscale.com/ipn/store/kubestore from tailscale.com/cmd/k8s-operator
|
||||
tailscale.com/ipn/store/mem from tailscale.com/ipn/ipnlocal+
|
||||
tailscale.com/k8s-operator from tailscale.com/cmd/k8s-operator
|
||||
tailscale.com/k8s-operator from tailscale.com/cmd/k8s-operator+
|
||||
tailscale.com/k8s-operator/api-proxy from tailscale.com/cmd/k8s-operator
|
||||
tailscale.com/k8s-operator/apis from tailscale.com/k8s-operator/apis/v1alpha1
|
||||
tailscale.com/k8s-operator/apis/v1alpha1 from tailscale.com/cmd/k8s-operator+
|
||||
tailscale.com/k8s-operator/reconciler from tailscale.com/k8s-operator/reconciler/tailnet
|
||||
tailscale.com/k8s-operator/reconciler/tailnet from tailscale.com/cmd/k8s-operator
|
||||
tailscale.com/k8s-operator/sessionrecording from tailscale.com/k8s-operator/api-proxy
|
||||
tailscale.com/k8s-operator/sessionrecording/spdy from tailscale.com/k8s-operator/sessionrecording
|
||||
tailscale.com/k8s-operator/sessionrecording/tsrecorder from tailscale.com/k8s-operator/sessionrecording+
|
||||
|
||||
@@ -8,3 +8,4 @@
|
||||
/proxyclass.yaml
|
||||
/proxygroup.yaml
|
||||
/recorder.yaml
|
||||
/tailnet.yaml
|
||||
|
||||
@@ -37,6 +37,9 @@ rules:
|
||||
- apiGroups: ["tailscale.com"]
|
||||
resources: ["dnsconfigs", "dnsconfigs/status"]
|
||||
verbs: ["get", "list", "watch", "update"]
|
||||
- apiGroups: ["tailscale.com"]
|
||||
resources: ["tailnets", "tailnets/status"]
|
||||
verbs: ["get", "list", "watch", "update"]
|
||||
- apiGroups: ["tailscale.com"]
|
||||
resources: ["recorders", "recorders/status"]
|
||||
verbs: ["get", "list", "watch", "update"]
|
||||
|
||||
@@ -181,6 +181,14 @@ spec:
|
||||
items:
|
||||
type: string
|
||||
pattern: ^tag:[a-zA-Z][a-zA-Z0-9-]*$
|
||||
tailnet:
|
||||
description: |-
|
||||
Tailnet specifies the tailnet this Connector should join. If blank, the default tailnet is used. When set, this
|
||||
name must match that of a valid Tailnet resource. This field is immutable and cannot be changed once set.
|
||||
type: string
|
||||
x-kubernetes-validations:
|
||||
- rule: self == oldSelf
|
||||
message: Connector tailnet is immutable
|
||||
x-kubernetes-validations:
|
||||
- rule: has(self.subnetRouter) || (has(self.exitNode) && self.exitNode == true) || has(self.appConnector)
|
||||
message: A Connector needs to have at least one of exit node, subnet router or app connector configured.
|
||||
|
||||
@@ -139,6 +139,14 @@ spec:
|
||||
items:
|
||||
type: string
|
||||
pattern: ^tag:[a-zA-Z][a-zA-Z0-9-]*$
|
||||
tailnet:
|
||||
description: |-
|
||||
Tailnet specifies the tailnet this ProxyGroup should join. If blank, the default tailnet is used. When set, this
|
||||
name must match that of a valid Tailnet resource. This field is immutable and cannot be changed once set.
|
||||
type: string
|
||||
x-kubernetes-validations:
|
||||
- rule: self == oldSelf
|
||||
message: ProxyGroup tailnet is immutable
|
||||
type:
|
||||
description: |-
|
||||
Type of the ProxyGroup proxies. Supported types are egress, ingress, and kube-apiserver.
|
||||
|
||||
@@ -1680,6 +1680,14 @@ spec:
|
||||
items:
|
||||
type: string
|
||||
pattern: ^tag:[a-zA-Z][a-zA-Z0-9-]*$
|
||||
tailnet:
|
||||
description: |-
|
||||
Tailnet specifies the tailnet this Recorder should join. If blank, the default tailnet is used. When set, this
|
||||
name must match that of a valid Tailnet resource. This field is immutable and cannot be changed once set.
|
||||
type: string
|
||||
x-kubernetes-validations:
|
||||
- rule: self == oldSelf
|
||||
message: Recorder tailnet is immutable
|
||||
x-kubernetes-validations:
|
||||
- rule: '!(self.replicas > 1 && (!has(self.storage) || !has(self.storage.s3)))'
|
||||
message: S3 storage must be used when deploying multiple Recorder replicas
|
||||
|
||||
@@ -0,0 +1,141 @@
|
||||
apiVersion: apiextensions.k8s.io/v1
|
||||
kind: CustomResourceDefinition
|
||||
metadata:
|
||||
annotations:
|
||||
controller-gen.kubebuilder.io/version: v0.17.0
|
||||
name: tailnets.tailscale.com
|
||||
spec:
|
||||
group: tailscale.com
|
||||
names:
|
||||
kind: Tailnet
|
||||
listKind: TailnetList
|
||||
plural: tailnets
|
||||
shortNames:
|
||||
- tn
|
||||
singular: tailnet
|
||||
scope: Cluster
|
||||
versions:
|
||||
- additionalPrinterColumns:
|
||||
- jsonPath: .metadata.creationTimestamp
|
||||
name: Age
|
||||
type: date
|
||||
- description: Status of the deployed Tailnet resources.
|
||||
jsonPath: .status.conditions[?(@.type == "TailnetReady")].reason
|
||||
name: Status
|
||||
type: string
|
||||
name: v1alpha1
|
||||
schema:
|
||||
openAPIV3Schema:
|
||||
type: object
|
||||
required:
|
||||
- metadata
|
||||
- spec
|
||||
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: |-
|
||||
Spec describes the desired state of the Tailnet.
|
||||
More info:
|
||||
https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#spec-and-status
|
||||
type: object
|
||||
required:
|
||||
- credentials
|
||||
properties:
|
||||
credentials:
|
||||
description: Denotes the location of the OAuth credentials to use for authenticating with this Tailnet.
|
||||
type: object
|
||||
required:
|
||||
- secretName
|
||||
properties:
|
||||
secretName:
|
||||
description: |-
|
||||
The name of the secret containing the OAuth credentials. This secret must contain two fields "client_id" and
|
||||
"client_secret".
|
||||
type: string
|
||||
loginUrl:
|
||||
description: URL of the control plane to be used by all resources managed by the operator using this Tailnet.
|
||||
type: string
|
||||
status:
|
||||
description: |-
|
||||
Status describes the status of the Tailnet. This is set
|
||||
and managed by the Tailscale operator.
|
||||
type: object
|
||||
properties:
|
||||
conditions:
|
||||
type: array
|
||||
items:
|
||||
description: Condition contains details for one aspect of the current state of this API Resource.
|
||||
type: object
|
||||
required:
|
||||
- lastTransitionTime
|
||||
- message
|
||||
- reason
|
||||
- status
|
||||
- type
|
||||
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.
|
||||
type: string
|
||||
format: date-time
|
||||
message:
|
||||
description: |-
|
||||
message is a human readable message indicating details about the transition.
|
||||
This may be an empty string.
|
||||
type: string
|
||||
maxLength: 32768
|
||||
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.
|
||||
type: integer
|
||||
format: int64
|
||||
minimum: 0
|
||||
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.
|
||||
type: string
|
||||
maxLength: 1024
|
||||
minLength: 1
|
||||
pattern: ^[A-Za-z]([A-Za-z0-9_,:]*[A-Za-z0-9_])?$
|
||||
status:
|
||||
description: status of the condition, one of True, False, Unknown.
|
||||
type: string
|
||||
enum:
|
||||
- "True"
|
||||
- "False"
|
||||
- Unknown
|
||||
type:
|
||||
description: type of condition in CamelCase or in foo.example.com/CamelCase.
|
||||
type: string
|
||||
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])$
|
||||
x-kubernetes-list-map-keys:
|
||||
- type
|
||||
x-kubernetes-list-type: map
|
||||
served: true
|
||||
storage: true
|
||||
subresources:
|
||||
status: {}
|
||||
@@ -206,6 +206,14 @@ spec:
|
||||
pattern: ^tag:[a-zA-Z][a-zA-Z0-9-]*$
|
||||
type: string
|
||||
type: array
|
||||
tailnet:
|
||||
description: |-
|
||||
Tailnet specifies the tailnet this Connector should join. If blank, the default tailnet is used. When set, this
|
||||
name must match that of a valid Tailnet resource. This field is immutable and cannot be changed once set.
|
||||
type: string
|
||||
x-kubernetes-validations:
|
||||
- message: Connector tailnet is immutable
|
||||
rule: self == oldSelf
|
||||
type: object
|
||||
x-kubernetes-validations:
|
||||
- message: A Connector needs to have at least one of exit node, subnet router or app connector configured.
|
||||
@@ -3135,6 +3143,14 @@ spec:
|
||||
pattern: ^tag:[a-zA-Z][a-zA-Z0-9-]*$
|
||||
type: string
|
||||
type: array
|
||||
tailnet:
|
||||
description: |-
|
||||
Tailnet specifies the tailnet this ProxyGroup should join. If blank, the default tailnet is used. When set, this
|
||||
name must match that of a valid Tailnet resource. This field is immutable and cannot be changed once set.
|
||||
type: string
|
||||
x-kubernetes-validations:
|
||||
- message: ProxyGroup tailnet is immutable
|
||||
rule: self == oldSelf
|
||||
type:
|
||||
description: |-
|
||||
Type of the ProxyGroup proxies. Supported types are egress, ingress, and kube-apiserver.
|
||||
@@ -4950,6 +4966,14 @@ spec:
|
||||
pattern: ^tag:[a-zA-Z][a-zA-Z0-9-]*$
|
||||
type: string
|
||||
type: array
|
||||
tailnet:
|
||||
description: |-
|
||||
Tailnet specifies the tailnet this Recorder should join. If blank, the default tailnet is used. When set, this
|
||||
name must match that of a valid Tailnet resource. This field is immutable and cannot be changed once set.
|
||||
type: string
|
||||
x-kubernetes-validations:
|
||||
- message: Recorder tailnet is immutable
|
||||
rule: self == oldSelf
|
||||
type: object
|
||||
x-kubernetes-validations:
|
||||
- message: S3 storage must be used when deploying multiple Recorder replicas
|
||||
@@ -5059,6 +5083,148 @@ spec:
|
||||
subresources:
|
||||
status: {}
|
||||
---
|
||||
apiVersion: apiextensions.k8s.io/v1
|
||||
kind: CustomResourceDefinition
|
||||
metadata:
|
||||
annotations:
|
||||
controller-gen.kubebuilder.io/version: v0.17.0
|
||||
name: tailnets.tailscale.com
|
||||
spec:
|
||||
group: tailscale.com
|
||||
names:
|
||||
kind: Tailnet
|
||||
listKind: TailnetList
|
||||
plural: tailnets
|
||||
shortNames:
|
||||
- tn
|
||||
singular: tailnet
|
||||
scope: Cluster
|
||||
versions:
|
||||
- additionalPrinterColumns:
|
||||
- jsonPath: .metadata.creationTimestamp
|
||||
name: Age
|
||||
type: date
|
||||
- description: Status of the deployed Tailnet resources.
|
||||
jsonPath: .status.conditions[?(@.type == "TailnetReady")].reason
|
||||
name: Status
|
||||
type: string
|
||||
name: v1alpha1
|
||||
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: |-
|
||||
Spec describes the desired state of the Tailnet.
|
||||
More info:
|
||||
https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#spec-and-status
|
||||
properties:
|
||||
credentials:
|
||||
description: Denotes the location of the OAuth credentials to use for authenticating with this Tailnet.
|
||||
properties:
|
||||
secretName:
|
||||
description: |-
|
||||
The name of the secret containing the OAuth credentials. This secret must contain two fields "client_id" and
|
||||
"client_secret".
|
||||
type: string
|
||||
required:
|
||||
- secretName
|
||||
type: object
|
||||
loginUrl:
|
||||
description: URL of the control plane to be used by all resources managed by the operator using this Tailnet.
|
||||
type: string
|
||||
required:
|
||||
- credentials
|
||||
type: object
|
||||
status:
|
||||
description: |-
|
||||
Status describes the status of the Tailnet. This is set
|
||||
and managed by the Tailscale operator.
|
||||
properties:
|
||||
conditions:
|
||||
items:
|
||||
description: Condition contains details for one aspect of the current state of this API Resource.
|
||||
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.
|
||||
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
|
||||
type: object
|
||||
required:
|
||||
- metadata
|
||||
- spec
|
||||
type: object
|
||||
served: true
|
||||
storage: true
|
||||
subresources:
|
||||
status: {}
|
||||
---
|
||||
apiVersion: rbac.authorization.k8s.io/v1
|
||||
kind: ClusterRole
|
||||
metadata:
|
||||
@@ -5141,6 +5307,16 @@ rules:
|
||||
- list
|
||||
- watch
|
||||
- update
|
||||
- apiGroups:
|
||||
- tailscale.com
|
||||
resources:
|
||||
- tailnets
|
||||
- tailnets/status
|
||||
verbs:
|
||||
- get
|
||||
- list
|
||||
- watch
|
||||
- update
|
||||
- apiGroups:
|
||||
- tailscale.com
|
||||
resources:
|
||||
|
||||
@@ -26,12 +26,14 @@ const (
|
||||
dnsConfigCRDPath = operatorDeploymentFilesPath + "/crds/tailscale.com_dnsconfigs.yaml"
|
||||
recorderCRDPath = operatorDeploymentFilesPath + "/crds/tailscale.com_recorders.yaml"
|
||||
proxyGroupCRDPath = operatorDeploymentFilesPath + "/crds/tailscale.com_proxygroups.yaml"
|
||||
tailnetCRDPath = operatorDeploymentFilesPath + "/crds/tailscale.com_tailnets.yaml"
|
||||
helmTemplatesPath = operatorDeploymentFilesPath + "/chart/templates"
|
||||
connectorCRDHelmTemplatePath = helmTemplatesPath + "/connector.yaml"
|
||||
proxyClassCRDHelmTemplatePath = helmTemplatesPath + "/proxyclass.yaml"
|
||||
dnsConfigCRDHelmTemplatePath = helmTemplatesPath + "/dnsconfig.yaml"
|
||||
recorderCRDHelmTemplatePath = helmTemplatesPath + "/recorder.yaml"
|
||||
proxyGroupCRDHelmTemplatePath = helmTemplatesPath + "/proxygroup.yaml"
|
||||
tailnetCRDHelmTemplatePath = helmTemplatesPath + "/tailnet.yaml"
|
||||
|
||||
helmConditionalStart = "{{ if .Values.installCRDs -}}\n"
|
||||
helmConditionalEnd = "{{- end -}}"
|
||||
@@ -154,6 +156,7 @@ func generate(baseDir string) error {
|
||||
{dnsConfigCRDPath, dnsConfigCRDHelmTemplatePath},
|
||||
{recorderCRDPath, recorderCRDHelmTemplatePath},
|
||||
{proxyGroupCRDPath, proxyGroupCRDHelmTemplatePath},
|
||||
{tailnetCRDPath, tailnetCRDHelmTemplatePath},
|
||||
} {
|
||||
if err := addCRDToHelm(crd.crdPath, crd.templatePath); err != nil {
|
||||
return fmt.Errorf("error adding %s CRD to Helm templates: %w", crd.crdPath, err)
|
||||
|
||||
@@ -102,7 +102,7 @@ func (a *IngressReconciler) maybeCleanup(ctx context.Context, logger *zap.Sugare
|
||||
return nil
|
||||
}
|
||||
|
||||
if done, err := a.ssr.Cleanup(ctx, logger, childResourceLabels(ing.Name, ing.Namespace, "ingress"), proxyTypeIngressResource); err != nil {
|
||||
if done, err := a.ssr.Cleanup(ctx, operatorTailnet, logger, childResourceLabels(ing.Name, ing.Namespace, "ingress"), proxyTypeIngressResource); err != nil {
|
||||
return fmt.Errorf("failed to cleanup: %w", err)
|
||||
} else if !done {
|
||||
logger.Debugf("cleanup not done yet, waiting for next reconcile")
|
||||
|
||||
@@ -54,6 +54,7 @@ import (
|
||||
"tailscale.com/ipn/store/kubestore"
|
||||
apiproxy "tailscale.com/k8s-operator/api-proxy"
|
||||
tsapi "tailscale.com/k8s-operator/apis/v1alpha1"
|
||||
"tailscale.com/k8s-operator/reconciler/tailnet"
|
||||
"tailscale.com/kube/kubetypes"
|
||||
"tailscale.com/tsnet"
|
||||
"tailscale.com/tstime"
|
||||
@@ -325,6 +326,17 @@ func runReconcilers(opts reconcilerOpts) {
|
||||
startlog.Fatalf("could not create manager: %v", err)
|
||||
}
|
||||
|
||||
tailnetOptions := tailnet.ReconcilerOptions{
|
||||
Client: mgr.GetClient(),
|
||||
TailscaleNamespace: opts.tailscaleNamespace,
|
||||
Clock: tstime.DefaultClock{},
|
||||
Logger: opts.log,
|
||||
}
|
||||
|
||||
if err = tailnet.NewReconciler(tailnetOptions).Register(mgr); err != nil {
|
||||
startlog.Fatalf("could not register tailnet reconciler: %v", err)
|
||||
}
|
||||
|
||||
svcFilter := handler.EnqueueRequestsFromMapFunc(serviceHandler)
|
||||
svcChildFilter := handler.EnqueueRequestsFromMapFunc(managedResourceHandlerForType("svc"))
|
||||
// If a ProxyClass changes, enqueue all Services labeled with that
|
||||
|
||||
@@ -49,11 +49,12 @@ import (
|
||||
)
|
||||
|
||||
const (
|
||||
reasonProxyGroupCreationFailed = "ProxyGroupCreationFailed"
|
||||
reasonProxyGroupReady = "ProxyGroupReady"
|
||||
reasonProxyGroupAvailable = "ProxyGroupAvailable"
|
||||
reasonProxyGroupCreating = "ProxyGroupCreating"
|
||||
reasonProxyGroupInvalid = "ProxyGroupInvalid"
|
||||
reasonProxyGroupCreationFailed = "ProxyGroupCreationFailed"
|
||||
reasonProxyGroupReady = "ProxyGroupReady"
|
||||
reasonProxyGroupAvailable = "ProxyGroupAvailable"
|
||||
reasonProxyGroupCreating = "ProxyGroupCreating"
|
||||
reasonProxyGroupInvalid = "ProxyGroupInvalid"
|
||||
reasonProxyGroupTailnetUnavailable = "ProxyGroupTailnetUnavailable"
|
||||
|
||||
// Copied from k8s.io/apiserver/pkg/registry/generic/registry/store.go@cccad306d649184bf2a0e319ba830c53f65c445c
|
||||
optimisticLockErrorMsg = "the object has been modified; please apply your changes to the latest version and try again"
|
||||
@@ -117,6 +118,23 @@ func (r *ProxyGroupReconciler) Reconcile(ctx context.Context, req reconcile.Requ
|
||||
} else if err != nil {
|
||||
return reconcile.Result{}, fmt.Errorf("failed to get tailscale.com ProxyGroup: %w", err)
|
||||
}
|
||||
|
||||
tailscaleClient := r.tsClient
|
||||
if pg.Spec.Tailnet != "" {
|
||||
tc, err := clientForTailnet(ctx, r.Client, r.tsNamespace, pg.Spec.Tailnet)
|
||||
if err != nil {
|
||||
oldPGStatus := pg.Status.DeepCopy()
|
||||
nrr := ¬ReadyReason{
|
||||
reason: reasonProxyGroupTailnetUnavailable,
|
||||
message: err.Error(),
|
||||
}
|
||||
|
||||
return reconcile.Result{}, errors.Join(err, r.maybeUpdateStatus(ctx, logger, pg, oldPGStatus, nrr, make(map[string][]netip.AddrPort)))
|
||||
}
|
||||
|
||||
tailscaleClient = tc
|
||||
}
|
||||
|
||||
if markedForDeletion(pg) {
|
||||
logger.Debugf("ProxyGroup is being deleted, cleaning up resources")
|
||||
ix := xslices.Index(pg.Finalizers, FinalizerName)
|
||||
@@ -125,7 +143,7 @@ func (r *ProxyGroupReconciler) Reconcile(ctx context.Context, req reconcile.Requ
|
||||
return reconcile.Result{}, nil
|
||||
}
|
||||
|
||||
if done, err := r.maybeCleanup(ctx, pg); err != nil {
|
||||
if done, err := r.maybeCleanup(ctx, tailscaleClient, pg); err != nil {
|
||||
if strings.Contains(err.Error(), optimisticLockErrorMsg) {
|
||||
logger.Infof("optimistic lock error, retrying: %s", err)
|
||||
return reconcile.Result{}, nil
|
||||
@@ -144,7 +162,7 @@ func (r *ProxyGroupReconciler) Reconcile(ctx context.Context, req reconcile.Requ
|
||||
}
|
||||
|
||||
oldPGStatus := pg.Status.DeepCopy()
|
||||
staticEndpoints, nrr, err := r.reconcilePG(ctx, pg, logger)
|
||||
staticEndpoints, nrr, err := r.reconcilePG(ctx, tailscaleClient, pg, logger)
|
||||
return reconcile.Result{}, errors.Join(err, r.maybeUpdateStatus(ctx, logger, pg, oldPGStatus, nrr, staticEndpoints))
|
||||
}
|
||||
|
||||
@@ -152,7 +170,7 @@ func (r *ProxyGroupReconciler) Reconcile(ctx context.Context, req reconcile.Requ
|
||||
// for deletion. It is separated out from Reconcile to make a clear separation
|
||||
// between reconciling the ProxyGroup, and posting the status of its created
|
||||
// resources onto the ProxyGroup status field.
|
||||
func (r *ProxyGroupReconciler) reconcilePG(ctx context.Context, pg *tsapi.ProxyGroup, logger *zap.SugaredLogger) (map[string][]netip.AddrPort, *notReadyReason, error) {
|
||||
func (r *ProxyGroupReconciler) reconcilePG(ctx context.Context, tailscaleClient tsClient, pg *tsapi.ProxyGroup, logger *zap.SugaredLogger) (map[string][]netip.AddrPort, *notReadyReason, error) {
|
||||
if !slices.Contains(pg.Finalizers, FinalizerName) {
|
||||
// This log line is printed exactly once during initial provisioning,
|
||||
// because once the finalizer is in place this block gets skipped. So,
|
||||
@@ -193,7 +211,7 @@ func (r *ProxyGroupReconciler) reconcilePG(ctx context.Context, pg *tsapi.ProxyG
|
||||
return notReady(reasonProxyGroupInvalid, fmt.Sprintf("invalid ProxyGroup spec: %v", err))
|
||||
}
|
||||
|
||||
staticEndpoints, nrr, err := r.maybeProvision(ctx, pg, proxyClass)
|
||||
staticEndpoints, nrr, err := r.maybeProvision(ctx, tailscaleClient, pg, proxyClass)
|
||||
if err != nil {
|
||||
return nil, nrr, err
|
||||
}
|
||||
@@ -279,7 +297,7 @@ func (r *ProxyGroupReconciler) validate(ctx context.Context, pg *tsapi.ProxyGrou
|
||||
return errors.Join(errs...)
|
||||
}
|
||||
|
||||
func (r *ProxyGroupReconciler) maybeProvision(ctx context.Context, pg *tsapi.ProxyGroup, proxyClass *tsapi.ProxyClass) (map[string][]netip.AddrPort, *notReadyReason, error) {
|
||||
func (r *ProxyGroupReconciler) maybeProvision(ctx context.Context, tailscaleClient tsClient, pg *tsapi.ProxyGroup, proxyClass *tsapi.ProxyClass) (map[string][]netip.AddrPort, *notReadyReason, error) {
|
||||
logger := r.logger(pg.Name)
|
||||
r.mu.Lock()
|
||||
r.ensureAddedToGaugeForProxyGroup(pg)
|
||||
@@ -302,7 +320,7 @@ func (r *ProxyGroupReconciler) maybeProvision(ctx context.Context, pg *tsapi.Pro
|
||||
}
|
||||
}
|
||||
|
||||
staticEndpoints, err := r.ensureConfigSecretsCreated(ctx, pg, proxyClass, svcToNodePorts)
|
||||
staticEndpoints, err := r.ensureConfigSecretsCreated(ctx, tailscaleClient, pg, proxyClass, svcToNodePorts)
|
||||
if err != nil {
|
||||
var selectorErr *FindStaticEndpointErr
|
||||
if errors.As(err, &selectorErr) {
|
||||
@@ -414,7 +432,7 @@ func (r *ProxyGroupReconciler) maybeProvision(ctx context.Context, pg *tsapi.Pro
|
||||
return r.notReadyErrf(pg, logger, "error reconciling metrics resources: %w", err)
|
||||
}
|
||||
|
||||
if err := r.cleanupDanglingResources(ctx, pg, proxyClass); err != nil {
|
||||
if err := r.cleanupDanglingResources(ctx, tailscaleClient, pg, proxyClass); err != nil {
|
||||
return r.notReadyErrf(pg, logger, "error cleaning up dangling resources: %w", err)
|
||||
}
|
||||
|
||||
@@ -611,7 +629,7 @@ func (r *ProxyGroupReconciler) ensureNodePortServiceCreated(ctx context.Context,
|
||||
|
||||
// cleanupDanglingResources ensures we don't leak config secrets, state secrets, and
|
||||
// tailnet devices when the number of replicas specified is reduced.
|
||||
func (r *ProxyGroupReconciler) cleanupDanglingResources(ctx context.Context, pg *tsapi.ProxyGroup, pc *tsapi.ProxyClass) error {
|
||||
func (r *ProxyGroupReconciler) cleanupDanglingResources(ctx context.Context, tailscaleClient tsClient, pg *tsapi.ProxyGroup, pc *tsapi.ProxyClass) error {
|
||||
logger := r.logger(pg.Name)
|
||||
metadata, err := r.getNodeMetadata(ctx, pg)
|
||||
if err != nil {
|
||||
@@ -625,7 +643,7 @@ func (r *ProxyGroupReconciler) cleanupDanglingResources(ctx context.Context, pg
|
||||
|
||||
// Dangling resource, delete the config + state Secrets, as well as
|
||||
// deleting the device from the tailnet.
|
||||
if err := r.deleteTailnetDevice(ctx, m.tsID, logger); err != nil {
|
||||
if err := r.deleteTailnetDevice(ctx, tailscaleClient, m.tsID, logger); err != nil {
|
||||
return err
|
||||
}
|
||||
if err := r.Delete(ctx, m.stateSecret); err != nil && !apierrors.IsNotFound(err) {
|
||||
@@ -668,7 +686,7 @@ func (r *ProxyGroupReconciler) cleanupDanglingResources(ctx context.Context, pg
|
||||
// maybeCleanup just deletes the device from the tailnet. All the kubernetes
|
||||
// resources linked to a ProxyGroup will get cleaned up via owner references
|
||||
// (which we can use because they are all in the same namespace).
|
||||
func (r *ProxyGroupReconciler) maybeCleanup(ctx context.Context, pg *tsapi.ProxyGroup) (bool, error) {
|
||||
func (r *ProxyGroupReconciler) maybeCleanup(ctx context.Context, tailscaleClient tsClient, pg *tsapi.ProxyGroup) (bool, error) {
|
||||
logger := r.logger(pg.Name)
|
||||
|
||||
metadata, err := r.getNodeMetadata(ctx, pg)
|
||||
@@ -677,7 +695,7 @@ func (r *ProxyGroupReconciler) maybeCleanup(ctx context.Context, pg *tsapi.Proxy
|
||||
}
|
||||
|
||||
for _, m := range metadata {
|
||||
if err := r.deleteTailnetDevice(ctx, m.tsID, logger); err != nil {
|
||||
if err := r.deleteTailnetDevice(ctx, tailscaleClient, m.tsID, logger); err != nil {
|
||||
return false, err
|
||||
}
|
||||
}
|
||||
@@ -698,9 +716,9 @@ func (r *ProxyGroupReconciler) maybeCleanup(ctx context.Context, pg *tsapi.Proxy
|
||||
return true, nil
|
||||
}
|
||||
|
||||
func (r *ProxyGroupReconciler) deleteTailnetDevice(ctx context.Context, id tailcfg.StableNodeID, logger *zap.SugaredLogger) error {
|
||||
func (r *ProxyGroupReconciler) deleteTailnetDevice(ctx context.Context, tailscaleClient tsClient, id tailcfg.StableNodeID, logger *zap.SugaredLogger) error {
|
||||
logger.Debugf("deleting device %s from control", string(id))
|
||||
if err := r.tsClient.DeleteDevice(ctx, string(id)); err != nil {
|
||||
if err := tailscaleClient.DeleteDevice(ctx, string(id)); err != nil {
|
||||
errResp := &tailscale.ErrResponse{}
|
||||
if ok := errors.As(err, errResp); ok && errResp.Status == http.StatusNotFound {
|
||||
logger.Debugf("device %s not found, likely because it has already been deleted from control", string(id))
|
||||
@@ -714,7 +732,13 @@ func (r *ProxyGroupReconciler) deleteTailnetDevice(ctx context.Context, id tailc
|
||||
return nil
|
||||
}
|
||||
|
||||
func (r *ProxyGroupReconciler) ensureConfigSecretsCreated(ctx context.Context, pg *tsapi.ProxyGroup, proxyClass *tsapi.ProxyClass, svcToNodePorts map[string]uint16) (endpoints map[string][]netip.AddrPort, err error) {
|
||||
func (r *ProxyGroupReconciler) ensureConfigSecretsCreated(
|
||||
ctx context.Context,
|
||||
tailscaleClient tsClient,
|
||||
pg *tsapi.ProxyGroup,
|
||||
proxyClass *tsapi.ProxyClass,
|
||||
svcToNodePorts map[string]uint16,
|
||||
) (endpoints map[string][]netip.AddrPort, err error) {
|
||||
logger := r.logger(pg.Name)
|
||||
endpoints = make(map[string][]netip.AddrPort, pgReplicas(pg)) // keyed by Service name.
|
||||
for i := range pgReplicas(pg) {
|
||||
@@ -728,7 +752,7 @@ func (r *ProxyGroupReconciler) ensureConfigSecretsCreated(ctx context.Context, p
|
||||
}
|
||||
|
||||
var existingCfgSecret *corev1.Secret // unmodified copy of secret
|
||||
if err := r.Get(ctx, client.ObjectKeyFromObject(cfgSecret), cfgSecret); err == nil {
|
||||
if err = r.Get(ctx, client.ObjectKeyFromObject(cfgSecret), cfgSecret); err == nil {
|
||||
logger.Debugf("Secret %s/%s already exists", cfgSecret.GetNamespace(), cfgSecret.GetName())
|
||||
existingCfgSecret = cfgSecret.DeepCopy()
|
||||
} else if !apierrors.IsNotFound(err) {
|
||||
@@ -742,7 +766,7 @@ func (r *ProxyGroupReconciler) ensureConfigSecretsCreated(ctx context.Context, p
|
||||
if len(tags) == 0 {
|
||||
tags = r.defaultTags
|
||||
}
|
||||
key, err := newAuthKey(ctx, r.tsClient, tags)
|
||||
key, err := newAuthKey(ctx, tailscaleClient, tags)
|
||||
if err != nil {
|
||||
return nil, err
|
||||
}
|
||||
@@ -757,7 +781,7 @@ func (r *ProxyGroupReconciler) ensureConfigSecretsCreated(ctx context.Context, p
|
||||
Namespace: r.tsNamespace,
|
||||
},
|
||||
}
|
||||
if err := r.Get(ctx, client.ObjectKeyFromObject(stateSecret), stateSecret); err != nil && !apierrors.IsNotFound(err) {
|
||||
if err = r.Get(ctx, client.ObjectKeyFromObject(stateSecret), stateSecret); err != nil && !apierrors.IsNotFound(err) {
|
||||
return nil, err
|
||||
}
|
||||
|
||||
|
||||
+31
-6
@@ -107,6 +107,7 @@ const (
|
||||
letsEncryptStagingEndpoint = "https://acme-staging-v02.api.letsencrypt.org/directory"
|
||||
|
||||
mainContainerName = "tailscale"
|
||||
operatorTailnet = ""
|
||||
)
|
||||
|
||||
var (
|
||||
@@ -152,6 +153,9 @@ type tailscaleSTSConfig struct {
|
||||
// HostnamePrefix specifies the desired prefix for the device's hostname. The hostname will be suffixed with the
|
||||
// ordinal number generated by the StatefulSet.
|
||||
HostnamePrefix string
|
||||
|
||||
// Tailnet specifies the Tailnet resource to use for producing auth keys.
|
||||
Tailnet string
|
||||
}
|
||||
|
||||
type connector struct {
|
||||
@@ -194,6 +198,16 @@ func IsHTTPSEnabledOnTailnet(tsnetServer tsnetServer) bool {
|
||||
// Provision ensures that the StatefulSet for the given service is running and
|
||||
// up to date.
|
||||
func (a *tailscaleSTSReconciler) Provision(ctx context.Context, logger *zap.SugaredLogger, sts *tailscaleSTSConfig) (*corev1.Service, error) {
|
||||
tailscaleClient := a.tsClient
|
||||
if sts.Tailnet != "" {
|
||||
tc, err := clientForTailnet(ctx, a.Client, a.operatorNamespace, sts.Tailnet)
|
||||
if err != nil {
|
||||
return nil, err
|
||||
}
|
||||
|
||||
tailscaleClient = tc
|
||||
}
|
||||
|
||||
// Do full reconcile.
|
||||
// TODO (don't create Service for the Connector)
|
||||
hsvc, err := a.reconcileHeadlessService(ctx, logger, sts)
|
||||
@@ -213,7 +227,7 @@ func (a *tailscaleSTSReconciler) Provision(ctx context.Context, logger *zap.Suga
|
||||
}
|
||||
sts.ProxyClass = proxyClass
|
||||
|
||||
secretNames, err := a.provisionSecrets(ctx, logger, sts, hsvc)
|
||||
secretNames, err := a.provisionSecrets(ctx, tailscaleClient, logger, sts, hsvc)
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("failed to create or get API key secret: %w", err)
|
||||
}
|
||||
@@ -237,7 +251,18 @@ func (a *tailscaleSTSReconciler) Provision(ctx context.Context, logger *zap.Suga
|
||||
// Cleanup removes all resources associated that were created by Provision with
|
||||
// the given labels. It returns true when all resources have been removed,
|
||||
// otherwise it returns false and the caller should retry later.
|
||||
func (a *tailscaleSTSReconciler) Cleanup(ctx context.Context, logger *zap.SugaredLogger, labels map[string]string, typ string) (done bool, _ error) {
|
||||
func (a *tailscaleSTSReconciler) Cleanup(ctx context.Context, tailnet string, logger *zap.SugaredLogger, labels map[string]string, typ string) (done bool, _ error) {
|
||||
tailscaleClient := a.tsClient
|
||||
if tailnet != "" {
|
||||
tc, err := clientForTailnet(ctx, a.Client, a.operatorNamespace, tailnet)
|
||||
if err != nil {
|
||||
logger.Errorf("failed to get tailscale client: %v", err)
|
||||
return false, nil
|
||||
}
|
||||
|
||||
tailscaleClient = tc
|
||||
}
|
||||
|
||||
// Need to delete the StatefulSet first, and delete it with foreground
|
||||
// cascading deletion. That way, the pod that's writing to the Secret will
|
||||
// stop running before we start looking at the Secret's contents, and
|
||||
@@ -279,7 +304,7 @@ func (a *tailscaleSTSReconciler) Cleanup(ctx context.Context, logger *zap.Sugare
|
||||
for _, dev := range devices {
|
||||
if dev.id != "" {
|
||||
logger.Debugf("deleting device %s from control", string(dev.id))
|
||||
if err = a.tsClient.DeleteDevice(ctx, string(dev.id)); err != nil {
|
||||
if err = tailscaleClient.DeleteDevice(ctx, string(dev.id)); err != nil {
|
||||
errResp := &tailscale.ErrResponse{}
|
||||
if ok := errors.As(err, errResp); ok && errResp.Status == http.StatusNotFound {
|
||||
logger.Debugf("device %s not found, likely because it has already been deleted from control", string(dev.id))
|
||||
@@ -360,7 +385,7 @@ func (a *tailscaleSTSReconciler) reconcileHeadlessService(ctx context.Context, l
|
||||
return createOrUpdate(ctx, a.Client, a.operatorNamespace, hsvc, func(svc *corev1.Service) { svc.Spec = hsvc.Spec })
|
||||
}
|
||||
|
||||
func (a *tailscaleSTSReconciler) provisionSecrets(ctx context.Context, logger *zap.SugaredLogger, stsC *tailscaleSTSConfig, hsvc *corev1.Service) ([]string, error) {
|
||||
func (a *tailscaleSTSReconciler) provisionSecrets(ctx context.Context, tailscaleClient tsClient, logger *zap.SugaredLogger, stsC *tailscaleSTSConfig, hsvc *corev1.Service) ([]string, error) {
|
||||
secretNames := make([]string, stsC.Replicas)
|
||||
|
||||
// Start by ensuring we have Secrets for the desired number of replicas. This will handle both creating and scaling
|
||||
@@ -403,7 +428,7 @@ func (a *tailscaleSTSReconciler) provisionSecrets(ctx context.Context, logger *z
|
||||
if len(tags) == 0 {
|
||||
tags = a.defaultTags
|
||||
}
|
||||
authKey, err = newAuthKey(ctx, a.tsClient, tags)
|
||||
authKey, err = newAuthKey(ctx, tailscaleClient, tags)
|
||||
if err != nil {
|
||||
return nil, err
|
||||
}
|
||||
@@ -477,7 +502,7 @@ func (a *tailscaleSTSReconciler) provisionSecrets(ctx context.Context, logger *z
|
||||
if dev != nil && dev.id != "" {
|
||||
var errResp *tailscale.ErrResponse
|
||||
|
||||
err = a.tsClient.DeleteDevice(ctx, string(dev.id))
|
||||
err = tailscaleClient.DeleteDevice(ctx, string(dev.id))
|
||||
switch {
|
||||
case errors.As(err, &errResp) && errResp.Status == http.StatusNotFound:
|
||||
// This device has possibly already been deleted in the admin console. So we can ignore this
|
||||
|
||||
@@ -23,6 +23,7 @@ import (
|
||||
"k8s.io/client-go/tools/record"
|
||||
"sigs.k8s.io/controller-runtime/pkg/client"
|
||||
"sigs.k8s.io/controller-runtime/pkg/reconcile"
|
||||
|
||||
tsoperator "tailscale.com/k8s-operator"
|
||||
tsapi "tailscale.com/k8s-operator/apis/v1alpha1"
|
||||
"tailscale.com/kube/kubetypes"
|
||||
@@ -167,7 +168,7 @@ func (a *ServiceReconciler) maybeCleanup(ctx context.Context, logger *zap.Sugare
|
||||
proxyTyp = proxyTypeIngressService
|
||||
}
|
||||
|
||||
if done, err := a.ssr.Cleanup(ctx, logger, childResourceLabels(svc.Name, svc.Namespace, "svc"), proxyTyp); err != nil {
|
||||
if done, err := a.ssr.Cleanup(ctx, operatorTailnet, logger, childResourceLabels(svc.Name, svc.Namespace, "svc"), proxyTyp); err != nil {
|
||||
return fmt.Errorf("failed to cleanup: %w", err)
|
||||
} else if !done {
|
||||
logger.Debugf("cleanup not done yet, waiting for next reconcile")
|
||||
|
||||
@@ -0,0 +1,58 @@
|
||||
// Copyright (c) Tailscale Inc & AUTHORS
|
||||
// SPDX-License-Identifier: BSD-3-Clause
|
||||
|
||||
//go:build !plan9
|
||||
|
||||
package main
|
||||
|
||||
import (
|
||||
"context"
|
||||
"fmt"
|
||||
|
||||
"golang.org/x/oauth2"
|
||||
"golang.org/x/oauth2/clientcredentials"
|
||||
corev1 "k8s.io/api/core/v1"
|
||||
"sigs.k8s.io/controller-runtime/pkg/client"
|
||||
|
||||
"tailscale.com/internal/client/tailscale"
|
||||
"tailscale.com/ipn"
|
||||
operatorutils "tailscale.com/k8s-operator"
|
||||
tsapi "tailscale.com/k8s-operator/apis/v1alpha1"
|
||||
)
|
||||
|
||||
func clientForTailnet(ctx context.Context, cl client.Client, namespace, name string) (tsClient, error) {
|
||||
var tn tsapi.Tailnet
|
||||
if err := cl.Get(ctx, client.ObjectKey{Name: name}, &tn); err != nil {
|
||||
return nil, fmt.Errorf("failed to get tailnet %q: %w", name, err)
|
||||
}
|
||||
|
||||
if !operatorutils.TailnetIsReady(&tn) {
|
||||
return nil, fmt.Errorf("tailnet %q is not ready", name)
|
||||
}
|
||||
|
||||
var secret corev1.Secret
|
||||
if err := cl.Get(ctx, client.ObjectKey{Name: tn.Spec.Credentials.SecretName, Namespace: namespace}, &secret); err != nil {
|
||||
return nil, fmt.Errorf("failed to get Secret %q in namespace %q: %w", tn.Spec.Credentials.SecretName, namespace, err)
|
||||
}
|
||||
|
||||
baseURL := ipn.DefaultControlURL
|
||||
if tn.Spec.LoginURL != "" {
|
||||
baseURL = tn.Spec.LoginURL
|
||||
}
|
||||
|
||||
credentials := clientcredentials.Config{
|
||||
ClientID: string(secret.Data["client_id"]),
|
||||
ClientSecret: string(secret.Data["client_secret"]),
|
||||
TokenURL: baseURL + "/api/v2/oauth/token",
|
||||
}
|
||||
|
||||
source := credentials.TokenSource(ctx)
|
||||
httpClient := oauth2.NewClient(ctx, source)
|
||||
|
||||
ts := tailscale.NewClient(defaultTailnet, nil)
|
||||
ts.UserAgent = "tailscale-k8s-operator"
|
||||
ts.HTTPClient = httpClient
|
||||
ts.BaseURL = baseURL
|
||||
|
||||
return ts, nil
|
||||
}
|
||||
@@ -42,10 +42,11 @@ import (
|
||||
)
|
||||
|
||||
const (
|
||||
reasonRecorderCreationFailed = "RecorderCreationFailed"
|
||||
reasonRecorderCreating = "RecorderCreating"
|
||||
reasonRecorderCreated = "RecorderCreated"
|
||||
reasonRecorderInvalid = "RecorderInvalid"
|
||||
reasonRecorderCreationFailed = "RecorderCreationFailed"
|
||||
reasonRecorderCreating = "RecorderCreating"
|
||||
reasonRecorderCreated = "RecorderCreated"
|
||||
reasonRecorderInvalid = "RecorderInvalid"
|
||||
reasonRecorderTailnetUnavailable = "RecorderTailnetUnavailable"
|
||||
|
||||
currentProfileKey = "_current-profile"
|
||||
)
|
||||
@@ -84,27 +85,6 @@ func (r *RecorderReconciler) Reconcile(ctx context.Context, req reconcile.Reques
|
||||
} else if err != nil {
|
||||
return reconcile.Result{}, fmt.Errorf("failed to get tailscale.com Recorder: %w", err)
|
||||
}
|
||||
if markedForDeletion(tsr) {
|
||||
logger.Debugf("Recorder is being deleted, cleaning up resources")
|
||||
ix := xslices.Index(tsr.Finalizers, FinalizerName)
|
||||
if ix < 0 {
|
||||
logger.Debugf("no finalizer, nothing to do")
|
||||
return reconcile.Result{}, nil
|
||||
}
|
||||
|
||||
if done, err := r.maybeCleanup(ctx, tsr); err != nil {
|
||||
return reconcile.Result{}, err
|
||||
} else if !done {
|
||||
logger.Debugf("Recorder resource cleanup not yet finished, will retry...")
|
||||
return reconcile.Result{RequeueAfter: shortRequeue}, nil
|
||||
}
|
||||
|
||||
tsr.Finalizers = slices.Delete(tsr.Finalizers, ix, ix+1)
|
||||
if err = r.Update(ctx, tsr); err != nil {
|
||||
return reconcile.Result{}, err
|
||||
}
|
||||
return reconcile.Result{}, nil
|
||||
}
|
||||
|
||||
oldTSRStatus := tsr.Status.DeepCopy()
|
||||
setStatusReady := func(tsr *tsapi.Recorder, status metav1.ConditionStatus, reason, message string) (reconcile.Result, error) {
|
||||
@@ -119,6 +99,38 @@ func (r *RecorderReconciler) Reconcile(ctx context.Context, req reconcile.Reques
|
||||
return reconcile.Result{}, nil
|
||||
}
|
||||
|
||||
tailscaleClient := r.tsClient
|
||||
if tsr.Spec.Tailnet != "" {
|
||||
tc, err := clientForTailnet(ctx, r.Client, r.tsNamespace, tsr.Spec.Tailnet)
|
||||
if err != nil {
|
||||
return setStatusReady(tsr, metav1.ConditionFalse, reasonRecorderTailnetUnavailable, err.Error())
|
||||
}
|
||||
|
||||
tailscaleClient = tc
|
||||
}
|
||||
|
||||
if markedForDeletion(tsr) {
|
||||
logger.Debugf("Recorder is being deleted, cleaning up resources")
|
||||
ix := xslices.Index(tsr.Finalizers, FinalizerName)
|
||||
if ix < 0 {
|
||||
logger.Debugf("no finalizer, nothing to do")
|
||||
return reconcile.Result{}, nil
|
||||
}
|
||||
|
||||
if done, err := r.maybeCleanup(ctx, tsr, tailscaleClient); err != nil {
|
||||
return reconcile.Result{}, err
|
||||
} else if !done {
|
||||
logger.Debugf("Recorder resource cleanup not yet finished, will retry...")
|
||||
return reconcile.Result{RequeueAfter: shortRequeue}, nil
|
||||
}
|
||||
|
||||
tsr.Finalizers = slices.Delete(tsr.Finalizers, ix, ix+1)
|
||||
if err = r.Update(ctx, tsr); err != nil {
|
||||
return reconcile.Result{}, err
|
||||
}
|
||||
return reconcile.Result{}, nil
|
||||
}
|
||||
|
||||
if !slices.Contains(tsr.Finalizers, FinalizerName) {
|
||||
// This log line is printed exactly once during initial provisioning,
|
||||
// because once the finalizer is in place this block gets skipped. So,
|
||||
@@ -137,7 +149,7 @@ func (r *RecorderReconciler) Reconcile(ctx context.Context, req reconcile.Reques
|
||||
return setStatusReady(tsr, metav1.ConditionFalse, reasonRecorderInvalid, message)
|
||||
}
|
||||
|
||||
if err = r.maybeProvision(ctx, tsr); err != nil {
|
||||
if err = r.maybeProvision(ctx, tailscaleClient, tsr); err != nil {
|
||||
reason := reasonRecorderCreationFailed
|
||||
message := fmt.Sprintf("failed creating Recorder: %s", err)
|
||||
if strings.Contains(err.Error(), optimisticLockErrorMsg) {
|
||||
@@ -155,7 +167,7 @@ func (r *RecorderReconciler) Reconcile(ctx context.Context, req reconcile.Reques
|
||||
return setStatusReady(tsr, metav1.ConditionTrue, reasonRecorderCreated, reasonRecorderCreated)
|
||||
}
|
||||
|
||||
func (r *RecorderReconciler) maybeProvision(ctx context.Context, tsr *tsapi.Recorder) error {
|
||||
func (r *RecorderReconciler) maybeProvision(ctx context.Context, tailscaleClient tsClient, tsr *tsapi.Recorder) error {
|
||||
logger := r.logger(tsr.Name)
|
||||
|
||||
r.mu.Lock()
|
||||
@@ -163,7 +175,7 @@ func (r *RecorderReconciler) maybeProvision(ctx context.Context, tsr *tsapi.Reco
|
||||
gaugeRecorderResources.Set(int64(r.recorders.Len()))
|
||||
r.mu.Unlock()
|
||||
|
||||
if err := r.ensureAuthSecretsCreated(ctx, tsr); err != nil {
|
||||
if err := r.ensureAuthSecretsCreated(ctx, tailscaleClient, tsr); err != nil {
|
||||
return fmt.Errorf("error creating secrets: %w", err)
|
||||
}
|
||||
|
||||
@@ -241,13 +253,13 @@ func (r *RecorderReconciler) maybeProvision(ctx context.Context, tsr *tsapi.Reco
|
||||
|
||||
// If we have scaled the recorder down, we will have dangling state secrets
|
||||
// that we need to clean up.
|
||||
if err = r.maybeCleanupSecrets(ctx, tsr); err != nil {
|
||||
if err = r.maybeCleanupSecrets(ctx, tailscaleClient, tsr); err != nil {
|
||||
return fmt.Errorf("error cleaning up Secrets: %w", err)
|
||||
}
|
||||
|
||||
var devices []tsapi.RecorderTailnetDevice
|
||||
for replica := range replicas {
|
||||
dev, ok, err := r.getDeviceInfo(ctx, tsr.Name, replica)
|
||||
dev, ok, err := r.getDeviceInfo(ctx, tailscaleClient, tsr.Name, replica)
|
||||
switch {
|
||||
case err != nil:
|
||||
return fmt.Errorf("failed to get device info: %w", err)
|
||||
@@ -312,7 +324,7 @@ func (r *RecorderReconciler) maybeCleanupServiceAccounts(ctx context.Context, ts
|
||||
return nil
|
||||
}
|
||||
|
||||
func (r *RecorderReconciler) maybeCleanupSecrets(ctx context.Context, tsr *tsapi.Recorder) error {
|
||||
func (r *RecorderReconciler) maybeCleanupSecrets(ctx context.Context, tailscaleClient tsClient, tsr *tsapi.Recorder) error {
|
||||
options := []client.ListOption{
|
||||
client.InNamespace(r.tsNamespace),
|
||||
client.MatchingLabels(tsrLabels("recorder", tsr.Name, nil)),
|
||||
@@ -354,7 +366,7 @@ func (r *RecorderReconciler) maybeCleanupSecrets(ctx context.Context, tsr *tsapi
|
||||
var errResp *tailscale.ErrResponse
|
||||
|
||||
r.log.Debugf("deleting device %s", devicePrefs.Config.NodeID)
|
||||
err = r.tsClient.DeleteDevice(ctx, string(devicePrefs.Config.NodeID))
|
||||
err = tailscaleClient.DeleteDevice(ctx, string(devicePrefs.Config.NodeID))
|
||||
switch {
|
||||
case errors.As(err, &errResp) && errResp.Status == http.StatusNotFound:
|
||||
// This device has possibly already been deleted in the admin console. So we can ignore this
|
||||
@@ -375,7 +387,7 @@ func (r *RecorderReconciler) maybeCleanupSecrets(ctx context.Context, tsr *tsapi
|
||||
// maybeCleanup just deletes the device from the tailnet. All the kubernetes
|
||||
// resources linked to a Recorder will get cleaned up via owner references
|
||||
// (which we can use because they are all in the same namespace).
|
||||
func (r *RecorderReconciler) maybeCleanup(ctx context.Context, tsr *tsapi.Recorder) (bool, error) {
|
||||
func (r *RecorderReconciler) maybeCleanup(ctx context.Context, tsr *tsapi.Recorder, tailscaleClient tsClient) (bool, error) {
|
||||
logger := r.logger(tsr.Name)
|
||||
|
||||
var replicas int32 = 1
|
||||
@@ -399,7 +411,7 @@ func (r *RecorderReconciler) maybeCleanup(ctx context.Context, tsr *tsapi.Record
|
||||
|
||||
nodeID := string(devicePrefs.Config.NodeID)
|
||||
logger.Debugf("deleting device %s from control", nodeID)
|
||||
if err = r.tsClient.DeleteDevice(ctx, nodeID); err != nil {
|
||||
if err = tailscaleClient.DeleteDevice(ctx, nodeID); err != nil {
|
||||
errResp := &tailscale.ErrResponse{}
|
||||
if errors.As(err, errResp) && errResp.Status == http.StatusNotFound {
|
||||
logger.Debugf("device %s not found, likely because it has already been deleted from control", nodeID)
|
||||
@@ -425,7 +437,7 @@ func (r *RecorderReconciler) maybeCleanup(ctx context.Context, tsr *tsapi.Record
|
||||
return true, nil
|
||||
}
|
||||
|
||||
func (r *RecorderReconciler) ensureAuthSecretsCreated(ctx context.Context, tsr *tsapi.Recorder) error {
|
||||
func (r *RecorderReconciler) ensureAuthSecretsCreated(ctx context.Context, tailscaleClient tsClient, tsr *tsapi.Recorder) error {
|
||||
var replicas int32 = 1
|
||||
if tsr.Spec.Replicas != nil {
|
||||
replicas = *tsr.Spec.Replicas
|
||||
@@ -453,7 +465,7 @@ func (r *RecorderReconciler) ensureAuthSecretsCreated(ctx context.Context, tsr *
|
||||
return fmt.Errorf("failed to get Secret %q: %w", key.Name, err)
|
||||
}
|
||||
|
||||
authKey, err := newAuthKey(ctx, r.tsClient, tags.Stringify())
|
||||
authKey, err := newAuthKey(ctx, tailscaleClient, tags.Stringify())
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
@@ -555,7 +567,7 @@ func getDevicePrefs(secret *corev1.Secret) (prefs prefs, ok bool, err error) {
|
||||
return prefs, ok, nil
|
||||
}
|
||||
|
||||
func (r *RecorderReconciler) getDeviceInfo(ctx context.Context, tsrName string, replica int32) (d tsapi.RecorderTailnetDevice, ok bool, err error) {
|
||||
func (r *RecorderReconciler) getDeviceInfo(ctx context.Context, tailscaleClient tsClient, tsrName string, replica int32) (d tsapi.RecorderTailnetDevice, ok bool, err error) {
|
||||
secret, err := r.getStateSecret(ctx, tsrName, replica)
|
||||
if err != nil || secret == nil {
|
||||
return tsapi.RecorderTailnetDevice{}, false, err
|
||||
@@ -569,7 +581,7 @@ func (r *RecorderReconciler) getDeviceInfo(ctx context.Context, tsrName string,
|
||||
// TODO(tomhjp): The profile info doesn't include addresses, which is why we
|
||||
// need the API. Should maybe update tsrecorder to write IPs to the state
|
||||
// Secret like containerboot does.
|
||||
device, err := r.tsClient.Device(ctx, string(prefs.Config.NodeID), nil)
|
||||
device, err := tailscaleClient.Device(ctx, string(prefs.Config.NodeID), nil)
|
||||
if err != nil {
|
||||
return tsapi.RecorderTailnetDevice{}, false, fmt.Errorf("failed to get device info from API: %w", err)
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user