Inject Information into Pods Using a PodPreset

FEATURE STATE: Kubernetes v1.6 [alpha]

This page shows how to use PodPreset objects to inject information like Secrets, volume mounts, and environment variables into Pods at creation time.

Before you begin

You need to have a Kubernetes cluster, and the kubectl command-line tool must be configured to communicate with your cluster. If you do not already have a cluster, you can create one using Minikube. Make sure that you have enabled PodPreset in your cluster.

Use Pod presets to inject environment variables and volumes

In this step, you create a preset that has a volume mount and one environment variable. Here is the manifest for the PodPreset:

apiVersion: settings.k8s.io/v1alpha1
kind: PodPreset
metadata:
  name: allow-database
spec:
  selector:
    matchLabels:
      role: frontend
  env:
    - name: DB_PORT
      value: "6379"
  volumeMounts:
    - mountPath: /cache
      name: cache-volume
  volumes:
    - name: cache-volume
      emptyDir: {}

The name of a PodPreset object must be a valid DNS subdomain name.

In the manifest, you can see that the preset has an environment variable definition called DB_PORT and a volume mount definition called cache-volume which is mounted under /cache. The selector specifies that the preset will act upon any Pod that is labeled role:frontend.

Create the PodPreset:

kubectl apply -f https://k8s.io/examples/podpreset/preset.yaml

Verify that the PodPreset has been created:

kubectl get podpreset
NAME             CREATED AT
allow-database   2020-01-24T08:54:29Z

This manifest defines a Pod labelled role: frontend (matching the PodPreset's selector):

apiVersion: v1
kind: Pod
metadata:
  name: website
  labels:
    app: website
    role: frontend
spec:
  containers:
    - name: website
      image: nginx
      ports:
        - containerPort: 80

Create the Pod:

kubectl create -f https://k8s.io/examples/podpreset/pod.yaml

Verify that the Pod is running:

kubectl get pods

The output shows that the Pod is running:

NAME      READY     STATUS    RESTARTS   AGE
website   1/1       Running   0          4m

View the Pod spec altered by the admission controller in order to see the effects of the preset having been applied:

kubectl get pod website -o yaml
apiVersion: v1
kind: Pod
metadata:
  name: website
  labels:
    app: website
    role: frontend
  annotations:
    podpreset.admission.kubernetes.io/podpreset-allow-database: "resource version"
spec:
  containers:
    - name: website
      image: nginx
      volumeMounts:
        - mountPath: /cache
          name: cache-volume
      ports:
        - containerPort: 80
      env:
        - name: DB_PORT
          value: "6379"
  volumes:
    - name: cache-volume
      emptyDir: {}

The DB_PORT environment variable, the volumeMount and the podpreset.admission.kubernetes.io annotation of the Pod verify that the preset has been applied.

Pod spec with ConfigMap example

This is an example to show how a Pod spec is modified by a Pod preset that references a ConfigMap containing environment variables.

Here is the manifest containing the definition of the ConfigMap:

apiVersion: v1
kind: ConfigMap
metadata:
  name: etcd-env-config
data:
  number_of_members: "1"
  initial_cluster_state: new
  initial_cluster_token: DUMMY_ETCD_INITIAL_CLUSTER_TOKEN
  discovery_token: DUMMY_ETCD_DISCOVERY_TOKEN
  discovery_url: http://etcd_discovery:2379
  etcdctl_peers: http://etcd:2379
  duplicate_key: FROM_CONFIG_MAP
  REPLACE_ME: "a value"

Create the ConfigMap:

kubectl create -f https://k8s.io/examples/podpreset/configmap.yaml

Here is a PodPreset manifest referencing that ConfigMap:

apiVersion: settings.k8s.io/v1alpha1
kind: PodPreset
metadata:
  name: allow-database
spec:
  selector:
    matchLabels:
      role: frontend
  env:
    - name: DB_PORT
      value: "6379"
    - name: duplicate_key
      value: FROM_ENV
    - name: expansion
      value: $(REPLACE_ME)
  envFrom:
    - configMapRef:
        name: etcd-env-config
  volumeMounts:
    - mountPath: /cache
      name: cache-volume
  volumes:
    - name: cache-volume
      emptyDir: {}

Create the preset that references the ConfigMap:

kubectl create -f https://k8s.io/examples/podpreset/allow-db.yaml

The following manifest defines a Pod matching the PodPreset for this example:

apiVersion: v1
kind: Pod
metadata:
  name: website
  labels:
    app: website
    role: frontend
spec:
  containers:
    - name: website
      image: nginx
      ports:
        - containerPort: 80

Create the Pod:

kubectl create -f https://k8s.io/examples/podpreset/pod.yaml

View the Pod spec altered by the admission controller in order to see the effects of the preset having been applied:

kubectl get pod website -o yaml
apiVersion: v1
kind: Pod
metadata:
  name: website
  labels:
    app: website
    role: frontend
  annotations:
    podpreset.admission.kubernetes.io/podpreset-allow-database: "resource version"
spec:
  containers:
    - name: website
      image: nginx
      volumeMounts:
        - mountPath: /cache
          name: cache-volume
      ports:
        - containerPort: 80
      env:
        - name: DB_PORT
          value: "6379"
        - name: duplicate_key
          value: FROM_ENV
        - name: expansion
          value: $(REPLACE_ME)
      envFrom:
        - configMapRef:
            name: etcd-env-config
  volumes:
    - name: cache-volume
      emptyDir: {}

The DB_PORT environment variable and the podpreset.admission.kubernetes.io annotation of the Pod verify that the preset has been applied.

ReplicaSet with Pod spec example

This is an example to show that only Pod specs are modified by Pod presets. Other workload types like ReplicaSets or Deployments are unaffected.

Here is the manifest for the PodPreset for this example:

apiVersion: settings.k8s.io/v1alpha1
kind: PodPreset
metadata:
  name: allow-database
spec:
  selector:
    matchLabels:
      role: frontend
  env:
    - name: DB_PORT
      value: "6379"
  volumeMounts:
    - mountPath: /cache
      name: cache-volume
  volumes:
    - name: cache-volume
      emptyDir: {}

Create the preset:

kubectl apply -f https://k8s.io/examples/podpreset/preset.yaml

This manifest defines a ReplicaSet that manages three application Pods:

apiVersion: apps/v1
kind: ReplicaSet
metadata:
  name: frontend
spec:
  replicas: 3
  selector:
    matchLabels:
      role: frontend
    matchExpressions:
      - {key: role, operator: In, values: [frontend]}
  template:
    metadata:
      labels:
        app: guestbook
        role: frontend
    spec:
      containers:
      - name: php-redis
        image: gcr.io/google_samples/gb-frontend:v3
        resources:
          requests:
            cpu: 100m
            memory: 100Mi
        env:
          - name: GET_HOSTS_FROM
            value: dns
        ports:
          - containerPort: 80

Create the ReplicaSet:

kubectl create -f https://k8s.io/examples/podpreset/replicaset.yaml

Verify that the Pods created by the ReplicaSet are running:

kubectl get pods

The output shows that the Pods are running:

NAME             READY   STATUS    RESTARTS   AGE
frontend-2l94q   1/1     Running   0          2m18s
frontend-6vdgn   1/1     Running   0          2m18s
frontend-jzt4p   1/1     Running   0          2m18s

View the spec of the ReplicaSet:

kubectl get replicasets frontend -o yaml
Note:

The ReplicaSet object's spec was not changed, nor does the ReplicaSet contain a podpreset.admission.kubernetes.io annotation. This is because a PodPreset only applies to Pod objects.

To see the effects of the preset having been applied, you need to look at individual Pods.

The command to view the specs of the affected Pods is:

kubectl get pod --selector=role=frontend -o yaml
apiVersion: v1
kind: Pod
metadata:
  name: frontend
  labels:
    app: guestbook
    role: frontend
  annotations:
    podpreset.admission.kubernetes.io/podpreset-allow-database: "resource version"
spec:
  containers:
  - name: php-redis
    image: gcr.io/google_samples/gb-frontend:v3
    resources:
      requests:
        cpu: 100m
        memory: 100Mi
    volumeMounts:
    - mountPath: /cache
      name: cache-volume
    env:
    - name: GET_HOSTS_FROM
      value: dns
    - name: DB_PORT
      value: "6379"
    ports:
    - containerPort: 80
  volumes:
  - name: cache-volume
    emptyDir: {}

Again the podpreset.admission.kubernetes.io annotation of the Pods verifies that the preset has been applied.

Multiple Pod presets example

This is an example to show how a Pod spec is modified by multiple Pod presets.

Here is the manifest for the first PodPreset:

apiVersion: settings.k8s.io/v1alpha1
kind: PodPreset
metadata:
  name: allow-database
spec:
  selector:
    matchLabels:
      role: frontend
  env:
    - name: DB_PORT
      value: "6379"
  volumeMounts:
    - mountPath: /cache
      name: cache-volume
  volumes:
    - name: cache-volume
      emptyDir: {}

Create the first PodPreset for this example:

kubectl apply -f https://k8s.io/examples/podpreset/preset.yaml

Here is the manifest for the second PodPreset:

apiVersion: settings.k8s.io/v1alpha1
kind: PodPreset
metadata:
  name: proxy
spec:
  selector:
    matchLabels:
      role: frontend
  volumeMounts:
    - mountPath: /etc/proxy/configs
      name: proxy-volume
  volumes:
    - name: proxy-volume
      emptyDir: {}

Create the second preset:

kubectl apply -f https://k8s.io/examples/podpreset/proxy.yaml

Here's a manifest containing the definition of an applicable Pod (matched by two PodPresets):

apiVersion: v1
kind: Pod
metadata:
  name: website
  labels:
    app: website
    role: frontend
spec:
  containers:
    - name: website
      image: nginx
      ports:
        - containerPort: 80

Create the Pod:

kubectl create -f https://k8s.io/examples/podpreset/pod.yaml

View the Pod spec altered by the admission controller in order to see the effects of both presets having been applied:

kubectl get pod website -o yaml
apiVersion: v1
kind: Pod
metadata:
  name: website
  labels:
    app: website
    role: frontend
  annotations:
    podpreset.admission.kubernetes.io/podpreset-allow-database: "resource version"
    podpreset.admission.kubernetes.io/podpreset-proxy: "resource version"
spec:
  containers:
    - name: website
      image: nginx
      volumeMounts:
        - mountPath: /cache
          name: cache-volume
        - mountPath: /etc/proxy/configs
          name: proxy-volume
      ports:
        - containerPort: 80
      env:
        - name: DB_PORT
          value: "6379"
  volumes:
    - name: cache-volume
      emptyDir: {}
    - name: proxy-volume
      emptyDir: {}

The DB_PORT environment variable, the proxy-volume VolumeMount and the two podpreset.admission.kubernetes.io annotations of the Pod verify that both presets have been applied.

Conflict example

This is an example to show how a Pod spec is not modified by a Pod preset when there is a conflict. The conflict in this example consists of a VolumeMount in the PodPreset conflicting with a Pod that defines the same mountPath.

Here is the manifest for the PodPreset:

apiVersion: settings.k8s.io/v1alpha1
kind: PodPreset
metadata:
  name: allow-database
spec:
  selector:
    matchLabels:
      role: frontend
  env:
    - name: DB_PORT
      value: "6379"
  volumeMounts:
    - mountPath: /cache
      name: other-volume
  volumes:
    - name: other-volume
      emptyDir: {}

Note the mountPath value of /cache.

Create the preset:

kubectl apply -f https://k8s.io/examples/podpreset/conflict-preset.yaml

Here is the manifest for the Pod:

apiVersion: v1
kind: Pod
metadata:
  name: website
  labels:
    app: website
    role: frontend
spec:
  containers:
    - name: website
      image: nginx
      volumeMounts:
        - mountPath: /cache
          name: cache-volume
      ports:
        - containerPort: 80
  volumes:
    - name: cache-volume
      emptyDir: {}

Note the volumeMount element with the same path as in the PodPreset.

Create the Pod:

kubectl create -f https://k8s.io/examples/podpreset/conflict-pod.yaml

View the Pod spec:

kubectl get pod website -o yaml
apiVersion: v1
kind: Pod
metadata:
  name: website
  labels:
    app: website
    role: frontend
spec:
  containers:
    - name: website
      image: nginx
      volumeMounts:
        - mountPath: /cache
          name: cache-volume
      ports:
        - containerPort: 80
  volumes:
    - name: cache-volume
      emptyDir: {}

You can see there is no preset annotation (podpreset.admission.kubernetes.io). Seeing no annotation tells you that no preset has not been applied to the Pod.

However, the PodPreset admission controller logs a warning containing details of the conflict. You can view the warning using kubectl:

kubectl -n kube-system logs -l=component=kube-apiserver

The output should look similar to:

W1214 13:00:12.987884       1 admission.go:147] conflict occurred while applying podpresets: allow-database on pod:  err: merging volume mounts for allow-database has a conflict on mount path /cache:
v1.VolumeMount{Name:"other-volume", ReadOnly:false, MountPath:"/cache", SubPath:"", MountPropagation:(*v1.MountPropagationMode)(nil), SubPathExpr:""}
does not match
core.VolumeMount{Name:"cache-volume", ReadOnly:false, MountPath:"/cache", SubPath:"", MountPropagation:(*core.MountPropagationMode)(nil), SubPathExpr:""}
 in container

Note the conflict message on the path for the VolumeMount.

Deleting a PodPreset

Once you don't need a PodPreset anymore, you can delete it with kubectl:

kubectl delete podpreset allow-database

The output shows that the PodPreset was deleted:

podpreset "allow-database" deleted
Last modified October 22, 2020 at 3:01 PM PST: Fix links in the tasks section (740eb340d)