1 - Configuration Best Practices

This document highlights and consolidates configuration best practices that are introduced throughout the user guide, Getting Started documentation, and examples.

This is a living document. If you think of something that is not on this list but might be useful to others, please don't hesitate to file an issue or submit a PR.

General Configuration Tips

  • When defining configurations, specify the latest stable API version.

  • Configuration files should be stored in version control before being pushed to the cluster. This allows you to quickly roll back a configuration change if necessary. It also aids cluster re-creation and restoration.

  • Write your configuration files using YAML rather than JSON. Though these formats can be used interchangeably in almost all scenarios, YAML tends to be more user-friendly.

  • Group related objects into a single file whenever it makes sense. One file is often easier to manage than several. See the guestbook-all-in-one.yaml file as an example of this syntax.

  • Note also that many kubectl commands can be called on a directory. For example, you can call kubectl apply on a directory of config files.

  • Don't specify default values unnecessarily: simple, minimal configuration will make errors less likely.

  • Put object descriptions in annotations, to allow better introspection.

"Naked" Pods versus ReplicaSets, Deployments, and Jobs

  • Don't use naked Pods (that is, Pods not bound to a ReplicaSet or Deployment) if you can avoid it. Naked Pods will not be rescheduled in the event of a node failure.

    A Deployment, which both creates a ReplicaSet to ensure that the desired number of Pods is always available, and specifies a strategy to replace Pods (such as RollingUpdate), is almost always preferable to creating Pods directly, except for some explicit restartPolicy: Never scenarios. A Job may also be appropriate.

Services

  • Create a Service before its corresponding backend workloads (Deployments or ReplicaSets), and before any workloads that need to access it. When Kubernetes starts a container, it provides environment variables pointing to all the Services which were running when the container was started. For example, if a Service named foo exists, all containers will get the following variables in their initial environment:

    FOO_SERVICE_HOST=<the host the Service is running on>
    FOO_SERVICE_PORT=<the port the Service is running on>
    

    This does imply an ordering requirement - any Service that a Pod wants to access must be created before the Pod itself, or else the environment variables will not be populated. DNS does not have this restriction.

  • An optional (though strongly recommended) cluster add-on is a DNS server. The DNS server watches the Kubernetes API for new Services and creates a set of DNS records for each. If DNS has been enabled throughout the cluster then all Pods should be able to do name resolution of Services automatically.

  • Don't specify a hostPort for a Pod unless it is absolutely necessary. When you bind a Pod to a hostPort, it limits the number of places the Pod can be scheduled, because each <hostIP, hostPort, protocol> combination must be unique. If you don't specify the hostIP and protocol explicitly, Kubernetes will use 0.0.0.0 as the default hostIP and TCP as the default protocol.

    If you only need access to the port for debugging purposes, you can use the apiserver proxy or kubectl port-forward.

    If you explicitly need to expose a Pod's port on the node, consider using a NodePort Service before resorting to hostPort.

  • Avoid using hostNetwork, for the same reasons as hostPort.

  • Use headless Services (which have a ClusterIP of None) for service discovery when you don't need kube-proxy load balancing.

Using Labels

  • Define and use labels that identify semantic attributes of your application or Deployment, such as { app: myapp, tier: frontend, phase: test, deployment: v3 }. You can use these labels to select the appropriate Pods for other resources; for example, a Service that selects all tier: frontend Pods, or all phase: test components of app: myapp. See the guestbook app for examples of this approach.

A Service can be made to span multiple Deployments by omitting release-specific labels from its selector. When you need to update a running service without downtime, use a Deployment.

A desired state of an object is described by a Deployment, and if changes to that spec are applied, the deployment controller changes the actual state to the desired state at a controlled rate.

  • Use the Kubernetes common labels for common use cases. These standardized labels enrich the metadata in a way that allows tools, including kubectl and dashboard, to work in an interoperable way.

  • You can manipulate labels for debugging. Because Kubernetes controllers (such as ReplicaSet) and Services match to Pods using selector labels, removing the relevant labels from a Pod will stop it from being considered by a controller or from being served traffic by a Service. If you remove the labels of an existing Pod, its controller will create a new Pod to take its place. This is a useful way to debug a previously "live" Pod in a "quarantine" environment. To interactively remove or add labels, use kubectl label.

Using kubectl

  • Use kubectl apply -f <directory>. This looks for Kubernetes configuration in all .yaml, .yml, and .json files in <directory> and passes it to apply.

  • Use label selectors for get and delete operations instead of specific object names. See the sections on label selectors and using labels effectively.

  • Use kubectl create deployment and kubectl expose to quickly create single-container Deployments and Services. See Use a Service to Access an Application in a Cluster for an example.

2 - ConfigMaps

A ConfigMap is an API object used to store non-confidential data in key-value pairs. Pods can consume ConfigMaps as environment variables, command-line arguments, or as configuration files in a volume.

A ConfigMap allows you to decouple environment-specific configuration from your container images, so that your applications are easily portable.

Motivation

Use a ConfigMap for setting configuration data separately from application code.

For example, imagine that you are developing an application that you can run on your own computer (for development) and in the cloud (to handle real traffic). You write the code to look in an environment variable named DATABASE_HOST. Locally, you set that variable to localhost. In the cloud, you set it to refer to a Kubernetes Service that exposes the database component to your cluster. This lets you fetch a container image running in the cloud and debug the exact same code locally if needed.

A ConfigMap is not designed to hold large chunks of data. The data stored in a ConfigMap cannot exceed 1 MiB. If you need to store settings that are larger than this limit, you may want to consider mounting a volume or use a separate database or file service.

ConfigMap object

A ConfigMap is an API object that lets you store configuration for other objects to use. Unlike most Kubernetes objects that have a spec, a ConfigMap has data and binaryData fields. These fields accept key-value pairs as their values. Both the data field and the binaryData are optional. The data field is designed to contain UTF-8 strings while the binaryData field is designed to contain binary data as base64-encoded strings.

The name of a ConfigMap must be a valid DNS subdomain name.

Each key under the data or the binaryData field must consist of alphanumeric characters, -, _ or .. The keys stored in data must not overlap with the keys in the binaryData field.

Starting from v1.19, you can add an immutable field to a ConfigMap definition to create an immutable ConfigMap.

ConfigMaps and Pods

You can write a Pod spec that refers to a ConfigMap and configures the container(s) in that Pod based on the data in the ConfigMap. The Pod and the ConfigMap must be in the same namespace.

Here's an example ConfigMap that has some keys with single values, and other keys where the value looks like a fragment of a configuration format.

apiVersion: v1
kind: ConfigMap
metadata:
  name: game-demo
data:
  # property-like keys; each key maps to a simple value
  player_initial_lives: "3"
  ui_properties_file_name: "user-interface.properties"

  # file-like keys
  game.properties: |
    enemy.types=aliens,monsters
    player.maximum-lives=5    
  user-interface.properties: |
    color.good=purple
    color.bad=yellow
    allow.textmode=true    

There are four different ways that you can use a ConfigMap to configure a container inside a Pod:

  1. Inside a container command and args
  2. Environment variables for a container
  3. Add a file in read-only volume, for the application to read
  4. Write code to run inside the Pod that uses the Kubernetes API to read a ConfigMap

These different methods lend themselves to different ways of modeling the data being consumed. For the first three methods, the kubelet uses the data from the ConfigMap when it launches container(s) for a Pod.

The fourth method means you have to write code to read the ConfigMap and its data. However, because you're using the Kubernetes API directly, your application can subscribe to get updates whenever the ConfigMap changes, and react when that happens. By accessing the Kubernetes API directly, this technique also lets you access a ConfigMap in a different namespace.

Here's an example Pod that uses values from game-demo to configure a Pod:

apiVersion: v1
kind: Pod
metadata:
  name: configmap-demo-pod
spec:
  containers:
    - name: demo
      image: alpine
      command: ["sleep", "3600"]
      env:
        # Define the environment variable
        - name: PLAYER_INITIAL_LIVES # Notice that the case is different here
                                     # from the key name in the ConfigMap.
          valueFrom:
            configMapKeyRef:
              name: game-demo           # The ConfigMap this value comes from.
              key: player_initial_lives # The key to fetch.
        - name: UI_PROPERTIES_FILE_NAME
          valueFrom:
            configMapKeyRef:
              name: game-demo
              key: ui_properties_file_name
      volumeMounts:
      - name: config
        mountPath: "/config"
        readOnly: true
  volumes:
    # You set volumes at the Pod level, then mount them into containers inside that Pod
    - name: config
      configMap:
        # Provide the name of the ConfigMap you want to mount.
        name: game-demo
        # An array of keys from the ConfigMap to create as files
        items:
        - key: "game.properties"
          path: "game.properties"
        - key: "user-interface.properties"
          path: "user-interface.properties"

A ConfigMap doesn't differentiate between single line property values and multi-line file-like values. What matters is how Pods and other objects consume those values.

For this example, defining a volume and mounting it inside the demo container as /config creates two files, /config/game.properties and /config/user-interface.properties, even though there are four keys in the ConfigMap. This is because the Pod definition specifies an items array in the volumes section. If you omit the items array entirely, every key in the ConfigMap becomes a file with the same name as the key, and you get 4 files.

Using ConfigMaps

ConfigMaps can be mounted as data volumes. ConfigMaps can also be used by other parts of the system, without being directly exposed to the Pod. For example, ConfigMaps can hold data that other parts of the system should use for configuration.

The most common way to use ConfigMaps is to configure settings for containers running in a Pod in the same namespace. You can also use a ConfigMap separately.

For example, you might encounter addons or operators that adjust their behavior based on a ConfigMap.

Using ConfigMaps as files from a Pod

To consume a ConfigMap in a volume in a Pod:

  1. Create a ConfigMap or use an existing one. Multiple Pods can reference the same ConfigMap.
  2. Modify your Pod definition to add a volume under .spec.volumes[]. Name the volume anything, and have a .spec.volumes[].configMap.name field set to reference your ConfigMap object.
  3. Add a .spec.containers[].volumeMounts[] to each container that needs the ConfigMap. Specify .spec.containers[].volumeMounts[].readOnly = true and .spec.containers[].volumeMounts[].mountPath to an unused directory name where you would like the ConfigMap to appear.
  4. Modify your image or command line so that the program looks for files in that directory. Each key in the ConfigMap data map becomes the filename under mountPath.

This is an example of a Pod that mounts a ConfigMap in a volume:

apiVersion: v1
kind: Pod
metadata:
  name: mypod
spec:
  containers:
  - name: mypod
    image: redis
    volumeMounts:
    - name: foo
      mountPath: "/etc/foo"
      readOnly: true
  volumes:
  - name: foo
    configMap:
      name: myconfigmap

Each ConfigMap you want to use needs to be referred to in .spec.volumes.

If there are multiple containers in the Pod, then each container needs its own volumeMounts block, but only one .spec.volumes is needed per ConfigMap.

Mounted ConfigMaps are updated automatically

When a ConfigMap currently consumed in a volume is updated, projected keys are eventually updated as well. The kubelet checks whether the mounted ConfigMap is fresh on every periodic sync. However, the kubelet uses its local cache for getting the current value of the ConfigMap. The type of the cache is configurable using the ConfigMapAndSecretChangeDetectionStrategy field in the KubeletConfiguration struct. A ConfigMap can be either propagated by watch (default), ttl-based, or by redirecting all requests directly to the API server. As a result, the total delay from the moment when the ConfigMap is updated to the moment when new keys are projected to the Pod can be as long as the kubelet sync period + cache propagation delay, where the cache propagation delay depends on the chosen cache type (it equals to watch propagation delay, ttl of cache, or zero correspondingly).

ConfigMaps consumed as environment variables are not updated automatically and require a pod restart.

Immutable ConfigMaps

FEATURE STATE: Kubernetes v1.21 [stable]

The Kubernetes feature Immutable Secrets and ConfigMaps provides an option to set individual Secrets and ConfigMaps as immutable. For clusters that extensively use ConfigMaps (at least tens of thousands of unique ConfigMap to Pod mounts), preventing changes to their data has the following advantages:

  • protects you from accidental (or unwanted) updates that could cause applications outages
  • improves performance of your cluster by significantly reducing load on kube-apiserver, by closing watches for ConfigMaps marked as immutable.

This feature is controlled by the ImmutableEphemeralVolumes feature gate. You can create an immutable ConfigMap by setting the immutable field to true. For example:

apiVersion: v1
kind: ConfigMap
metadata:
  ...
data:
  ...
immutable: true

Once a ConfigMap is marked as immutable, it is not possible to revert this change nor to mutate the contents of the data or the binaryData field. You can only delete and recreate the ConfigMap. Because existing Pods maintain a mount point to the deleted ConfigMap, it is recommended to recreate these pods.

What's next

3 - Secrets

A Secret is an object that contains a small amount of sensitive data such as a password, a token, or a key. Such information might otherwise be put in a Pod specification or in a container image. Using a Secret means that you don't need to include confidential data in your application code.

Because Secrets can be created independently of the Pods that use them, there is less risk of the Secret (and its data) being exposed during the workflow of creating, viewing, and editing Pods. Kubernetes, and applications that run in your cluster, can also take additional precautions with Secrets, such as avoiding writing secret data to nonvolatile storage.

Secrets are similar to ConfigMaps but are specifically intended to hold confidential data.

See Information security for Secrets for more details.

Uses for Secrets

There are three main ways for a Pod to use a Secret:

The Kubernetes control plane also uses Secrets; for example, bootstrap token Secrets are a mechanism to help automate node registration.

Alternatives to Secrets

Rather than using a Secret to protect confidential data, you can pick from alternatives.

Here are some of your options:

  • if your cloud-native component needs to authenticate to another application that you know is running within the same Kubernetes cluster, you can use a ServiceAccount and its tokens to identify your client.
  • there are third-party tools that you can run, either within or outside your cluster, that provide secrets management. For example, a service that Pods access over HTTPS, that reveals a secret if the client correctly authenticates (for example, with a ServiceAccount token).
  • for authentication, you can implement a custom signer for X.509 certificates, and use CertificateSigningRequests to let that custom signer issue certificates to Pods that need them.
  • you can use a device plugin to expose node-local encryption hardware to a specific Pod. For example, you can schedule trusted Pods onto nodes that provide a Trusted Platform Module, configured out-of-band.

You can also combine two or more of those options, including the option to use Secret objects themselves.

For example: implement (or deploy) an operator that fetches short-lived session tokens from an external service, and then creates Secrets based on those short-lived session tokens. Pods running in your cluster can make use of the session tokens, and operator ensures they are valid. This separation means that you can run Pods that are unaware of the exact mechanisms for issuing and refreshing those session tokens.

Working with Secrets

Creating a Secret

There are several options to create a Secret:

Constraints on Secret names and data

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

You can specify the data and/or the stringData field when creating a configuration file for a Secret. The data and the stringData fields are optional. The values for all keys in the data field have to be base64-encoded strings. If the conversion to base64 string is not desirable, you can choose to specify the stringData field instead, which accepts arbitrary strings as values.

The keys of data and stringData must consist of alphanumeric characters, -, _ or .. All key-value pairs in the stringData field are internally merged into the data field. If a key appears in both the data and the stringData field, the value specified in the stringData field takes precedence.

Size limit

Individual secrets are limited to 1MiB in size. This is to discourage creation of very large secrets that could exhaust the API server and kubelet memory. However, creation of many smaller secrets could also exhaust memory. You can use a resource quota to limit the number of Secrets (or other resources) in a namespace.

Editing a Secret

You can edit an existing Secret using kubectl:

kubectl edit secrets mysecret

This opens your default editor and allows you to update the base64 encoded Secret values in the data field; for example:

# Please edit the object below. Lines beginning with a '#' will be ignored,
# and an empty file will abort the edit. If an error occurs while saving this file, it will be
# reopened with the relevant failures.
#
apiVersion: v1
data:
  username: YWRtaW4=
  password: MWYyZDFlMmU2N2Rm
kind: Secret
metadata:
  annotations:
    kubectl.kubernetes.io/last-applied-configuration: { ... }
  creationTimestamp: 2020-01-22T18:41:56Z
  name: mysecret
  namespace: default
  resourceVersion: "164619"
  uid: cfee02d6-c137-11e5-8d73-42010af00002
type: Opaque

That example manifest defines a Secret with two keys in the data field: username and password. The values are Base64 strings in the manifest; however, when you use the Secret with a Pod then the kubelet provides the decoded data to the Pod and its containers.

You can package many keys and values into one Secret, or use many Secrets, whichever is convenient.

Using a Secret

Secrets can be mounted as data volumes or exposed as environment variables to be used by a container in a Pod. Secrets can also be used by other parts of the system, without being directly exposed to the Pod. For example, Secrets can hold credentials that other parts of the system should use to interact with external systems on your behalf.

Secret volume sources are validated to ensure that the specified object reference actually points to an object of type Secret. Therefore, a Secret needs to be created before any Pods that depend on it.

If the Secret cannot be fetched (perhaps because it does not exist, or due to a temporary lack of connection to the API server) the kubelet periodically retries running that Pod. The kubelet also reports an Event for that Pod, including details of the problem fetching the Secret.

Optional Secrets

When you define a container environment variable based on a Secret, you can mark it as optional. The default is for the Secret to be required.

None of a Pod's containers will start until all non-optional Secrets are available.

If a Pod references a specific key in a Secret and that Secret does exist, but is missing the named key, the Pod fails during startup.

Using Secrets as files from a Pod

If you want to access data from a Secret in a Pod, one way to do that is to have Kubernetes make the value of that Secret be available as a file inside the filesystem of one or more of the Pod's containers.

To configure that, you:

  1. Create a secret or use an existing one. Multiple Pods can reference the same secret.
  2. Modify your Pod definition to add a volume under .spec.volumes[]. Name the volume anything, and have a .spec.volumes[].secret.secretName field equal to the name of the Secret object.
  3. Add a .spec.containers[].volumeMounts[] to each container that needs the secret. Specify .spec.containers[].volumeMounts[].readOnly = true and .spec.containers[].volumeMounts[].mountPath to an unused directory name where you would like the secrets to appear.
  4. Modify your image or command line so that the program looks for files in that directory. Each key in the secret data map becomes the filename under mountPath.

This is an example of a Pod that mounts a Secret named mysecret in a volume:

apiVersion: v1
kind: Pod
metadata:
  name: mypod
spec:
  containers:
  - name: mypod
    image: redis
    volumeMounts:
    - name: foo
      mountPath: "/etc/foo"
      readOnly: true
  volumes:
  - name: foo
    secret:
      secretName: mysecret
      optional: false # default setting; "mysecret" must exist

Each Secret you want to use needs to be referred to in .spec.volumes.

If there are multiple containers in the Pod, then each container needs its own volumeMounts block, but only one .spec.volumes is needed per Secret.

Projection of Secret keys to specific paths

You can also control the paths within the volume where Secret keys are projected. You can use the .spec.volumes[].secret.items field to change the target path of each key:

apiVersion: v1
kind: Pod
metadata:
  name: mypod
spec:
  containers:
  - name: mypod
    image: redis
    volumeMounts:
    - name: foo
      mountPath: "/etc/foo"
      readOnly: true
  volumes:
  - name: foo
    secret:
      secretName: mysecret
      items:
      - key: username
        path: my-group/my-username

What will happen:

  • the username key from mysecret is available to the container at the path /etc/foo/my-group/my-username instead of at /etc/foo/username.
  • the password key from that Secret object is not projected.

If .spec.volumes[].secret.items is used, only keys specified in items are projected. To consume all keys from the Secret, all of them must be listed in the items field.

If you list keys explicitly, then all listed keys must exist in the corresponding Secret. Otherwise, the volume is not created.

Secret files permissions

You can set the POSIX file access permission bits for a single Secret key. If you don't specify any permissions, 0644 is used by default. You can also set a default mode for the entire Secret volume and override per key if needed.

For example, you can specify a default mode like this:

apiVersion: v1
kind: Pod
metadata:
  name: mypod
spec:
  containers:
  - name: mypod
    image: redis
    volumeMounts:
    - name: foo
      mountPath: "/etc/foo"
  volumes:
  - name: foo
    secret:
      secretName: mysecret
      defaultMode: 0400

The secret is mounted on /etc/foo; all the files created by the secret volume mount have permission 0400.

Consuming Secret values from volumes

Inside the container that mounts a secret volume, the secret keys appear as files. The secret values are base64 decoded and stored inside these files.

This is the result of commands executed inside the container from the example above:

ls /etc/foo/

The output is similar to:

username
password
cat /etc/foo/username

The output is similar to:

admin
cat /etc/foo/password

The output is similar to:

1f2d1e2e67df

The program in a container is responsible for reading the secret data from these files, as needed.

Mounted Secrets are updated automatically

When a volume contains data from a Secret, and that Secret is updated, Kubernetes tracks this and updates the data in the volume, using an eventually-consistent approach.

The kubelet keeps a cache of the current keys and values for the Secrets that are used in volumes for pods on that node. You can configure the way that the kubelet detects changes from the cached values. The configMapAndSecretChangeDetectionStrategy field in the kubelet configuration controls which strategy the kubelet uses. The default strategy is Watch.

Updates to Secrets can be either propagated by an API watch mechanism (the default), based on a cache with a defined time-to-live, or polled from the cluster API server on each kubelet synchronisation loop.

As a result, the total delay from the moment when the Secret is updated to the moment when new keys are projected to the Pod can be as long as the kubelet sync period + cache propagation delay, where the cache propagation delay depends on the chosen cache type (following the same order listed in the previous paragraph, these are: watch propagation delay, the configured cache TTL, or zero for direct polling).

Using Secrets as environment variables

To use a Secret in an environment variable in a Pod:

  1. Create a Secret (or use an existing one). Multiple Pods can reference the same Secret.
  2. Modify your Pod definition in each container that you wish to consume the value of a secret key to add an environment variable for each secret key you wish to consume. The environment variable that consumes the secret key should populate the secret's name and key in env[].valueFrom.secretKeyRef.
  3. Modify your image and/or command line so that the program looks for values in the specified environment variables.

This is an example of a Pod that uses a Secret via environment variables:

apiVersion: v1
kind: Pod
metadata:
  name: secret-env-pod
spec:
  containers:
  - name: mycontainer
    image: redis
    env:
      - name: SECRET_USERNAME
        valueFrom:
          secretKeyRef:
            name: mysecret
            key: username
            optional: false # same as default; "mysecret" must exist
                            # and include a key named "username"
      - name: SECRET_PASSWORD
        valueFrom:
          secretKeyRef:
            name: mysecret
            key: password
            optional: false # same as default; "mysecret" must exist
                            # and include a key named "password"
  restartPolicy: Never

Invalid environment variables

Secrets used to populate environment variables by the envFrom field that have keys that are considered invalid environment variable names will have those keys skipped. The Pod is allowed to start.

If you define a Pod with an invalid variable name, the failed Pod startup includes an event with the reason set to InvalidVariableNames and a message that lists the skipped invalid keys. The following example shows a Pod that refers to a Secret named mysecret, where mysecret contains 2 invalid keys: 1badkey and 2alsobad.

kubectl get events

The output is similar to:

LASTSEEN   FIRSTSEEN   COUNT     NAME            KIND      SUBOBJECT                         TYPE      REASON
0s         0s          1         dapi-test-pod   Pod                                         Warning   InvalidEnvironmentVariableNames   kubelet, 127.0.0.1      Keys [1badkey, 2alsobad] from the EnvFrom secret default/mysecret were skipped since they are considered invalid environment variable names.

Consuming Secret values from environment variables

Inside a container that consumes a Secret using environment variables, the secret keys appear as normal environment variables. The values of those variables are the base64 decoded values of the secret data.

This is the result of commands executed inside the container from the example above:

echo "$SECRET_USERNAME"

The output is similar to:

admin
echo "$SECRET_PASSWORD"

The output is similar to:

1f2d1e2e67df

Container image pull secrets

If you want to fetch container images from a private repository, you need a way for the kubelet on each node to authenticate to that repository. You can configure image pull secrets to make this possible. These secrets are configured at the Pod level.

The imagePullSecrets field for a Pod is a list of references to Secrets in the same namespace as the Pod. You can use an imagePullSecrets to pass image registry access credentials to the kubelet. The kubelet uses this information to pull a private image on behalf of your Pod. See PodSpec in the Pod API reference for more information about the imagePullSecrets field.

Using imagePullSecrets

The imagePullSecrets field is a list of references to secrets in the same namespace. You can use an imagePullSecrets to pass a secret that contains a Docker (or other) image registry password to the kubelet. The kubelet uses this information to pull a private image on behalf of your Pod. See the PodSpec API for more information about the imagePullSecrets field.

Manually specifying an imagePullSecret

You can learn how to specify imagePullSecrets from the container images documentation.

Arranging for imagePullSecrets to be automatically attached

You can manually create imagePullSecrets, and reference these from a ServiceAccount. Any Pods created with that ServiceAccount or created with that ServiceAccount by default, will get their imagePullSecrets field set to that of the service account. See Add ImagePullSecrets to a service account for a detailed explanation of that process.

Using Secrets with static Pods

You cannot use ConfigMaps or Secrets with static Pods.

Use cases

Use case: As container environment variables

Create a secret

apiVersion: v1
kind: Secret
metadata:
  name: mysecret
type: Opaque
data:
  USER_NAME: YWRtaW4=
  PASSWORD: MWYyZDFlMmU2N2Rm

Create the Secret:

kubectl apply -f mysecret.yaml

Use envFrom to define all of the Secret's data as container environment variables. The key from the Secret becomes the environment variable name in the Pod.

apiVersion: v1
kind: Pod
metadata:
  name: secret-test-pod
spec:
  containers:
    - name: test-container
      image: k8s.gcr.io/busybox
      command: [ "/bin/sh", "-c", "env" ]
      envFrom:
      - secretRef:
          name: mysecret
  restartPolicy: Never

Use case: Pod with SSH keys

Create a Secret containing some SSH keys:

kubectl create secret generic ssh-key-secret --from-file=ssh-privatekey=/path/to/.ssh/id_rsa --from-file=ssh-publickey=/path/to/.ssh/id_rsa.pub

The output is similar to:

secret "ssh-key-secret" created

You can also create a kustomization.yaml with a secretGenerator field containing ssh keys.

Now you can create a Pod which references the secret with the SSH key and consumes it in a volume:

apiVersion: v1
kind: Pod
metadata:
  name: secret-test-pod
  labels:
    name: secret-test
spec:
  volumes:
  - name: secret-volume
    secret:
      secretName: ssh-key-secret
  containers:
  - name: ssh-test-container
    image: mySshImage
    volumeMounts:
    - name: secret-volume
      readOnly: true
      mountPath: "/etc/secret-volume"

When the container's command runs, the pieces of the key will be available in:

/etc/secret-volume/ssh-publickey
/etc/secret-volume/ssh-privatekey

The container is then free to use the secret data to establish an SSH connection.

Use case: Pods with prod / test credentials

This example illustrates a Pod which consumes a secret containing production credentials and another Pod which consumes a secret with test environment credentials.

You can create a kustomization.yaml with a secretGenerator field or run kubectl create secret.

kubectl create secret generic prod-db-secret --from-literal=username=produser --from-literal=password=Y4nys7f11

The output is similar to:

secret "prod-db-secret" created

You can also create a secret for test environment credentials.

kubectl create secret generic test-db-secret --from-literal=username=testuser --from-literal=password=iluvtests

The output is similar to:

secret "test-db-secret" created

Now make the Pods:

cat <<EOF > pod.yaml
apiVersion: v1
kind: List
items:
- kind: Pod
  apiVersion: v1
  metadata:
    name: prod-db-client-pod
    labels:
      name: prod-db-client
  spec:
    volumes:
    - name: secret-volume
      secret:
        secretName: prod-db-secret
    containers:
    - name: db-client-container
      image: myClientImage
      volumeMounts:
      - name: secret-volume
        readOnly: true
        mountPath: "/etc/secret-volume"
- kind: Pod
  apiVersion: v1
  metadata:
    name: test-db-client-pod
    labels:
      name: test-db-client
  spec:
    volumes:
    - name: secret-volume
      secret:
        secretName: test-db-secret
    containers:
    - name: db-client-container
      image: myClientImage
      volumeMounts:
      - name: secret-volume
        readOnly: true
        mountPath: "/etc/secret-volume"
EOF

Add the pods to the same kustomization.yaml:

cat <<EOF >> kustomization.yaml
resources:
- pod.yaml
EOF

Apply all those objects on the API server by running:

kubectl apply -k .

Both containers will have the following files present on their filesystems with the values for each container's environment:

/etc/secret-volume/username
/etc/secret-volume/password

Note how the specs for the two Pods differ only in one field; this facilitates creating Pods with different capabilities from a common Pod template.

You could further simplify the base Pod specification by using two service accounts:

  1. prod-user with the prod-db-secret
  2. test-user with the test-db-secret

The Pod specification is shortened to:

apiVersion: v1
kind: Pod
metadata:
  name: prod-db-client-pod
  labels:
    name: prod-db-client
spec:
  serviceAccount: prod-db-client
  containers:
  - name: db-client-container
    image: myClientImage

Use case: dotfiles in a secret volume

You can make your data "hidden" by defining a key that begins with a dot. This key represents a dotfile or "hidden" file. For example, when the following secret is mounted into a volume, secret-volume:

apiVersion: v1
kind: Secret
metadata:
  name: dotfile-secret
data:
  .secret-file: dmFsdWUtMg0KDQo=
---
apiVersion: v1
kind: Pod
metadata:
  name: secret-dotfiles-pod
spec:
  volumes:
  - name: secret-volume
    secret:
      secretName: dotfile-secret
  containers:
  - name: dotfile-test-container
    image: k8s.gcr.io/busybox
    command:
    - ls
    - "-l"
    - "/etc/secret-volume"
    volumeMounts:
    - name: secret-volume
      readOnly: true
      mountPath: "/etc/secret-volume"

The volume will contain a single file, called .secret-file, and the dotfile-test-container will have this file present at the path /etc/secret-volume/.secret-file.

Use case: Secret visible to one container in a Pod

Consider a program that needs to handle HTTP requests, do some complex business logic, and then sign some messages with an HMAC. Because it has complex application logic, there might be an unnoticed remote file reading exploit in the server, which could expose the private key to an attacker.

This could be divided into two processes in two containers: a frontend container which handles user interaction and business logic, but which cannot see the private key; and a signer container that can see the private key, and responds to simple signing requests from the frontend (for example, over localhost networking).

With this partitioned approach, an attacker now has to trick the application server into doing something rather arbitrary, which may be harder than getting it to read a file.

Types of Secret

When creating a Secret, you can specify its type using the type field of the Secret resource, or certain equivalent kubectl command line flags (if available). The Secret type is used to facilitate programmatic handling of the Secret data.

Kubernetes provides several built-in types for some common usage scenarios. These types vary in terms of the validations performed and the constraints Kubernetes imposes on them.

Built-in TypeUsage
Opaquearbitrary user-defined data
kubernetes.io/service-account-tokenServiceAccount token
kubernetes.io/dockercfgserialized ~/.dockercfg file
kubernetes.io/dockerconfigjsonserialized ~/.docker/config.json file
kubernetes.io/basic-authcredentials for basic authentication
kubernetes.io/ssh-authcredentials for SSH authentication
kubernetes.io/tlsdata for a TLS client or server
bootstrap.kubernetes.io/tokenbootstrap token data

You can define and use your own Secret type by assigning a non-empty string as the type value for a Secret object (an empty string is treated as an Opaque type).

Kubernetes doesn't impose any constraints on the type name. However, if you are using one of the built-in types, you must meet all the requirements defined for that type.

If you are defining a type of secret that's for public use, follow the convention and structure the secret type to have your domain name before the name, separated by a /. For example: cloud-hosting.example.net/cloud-api-credentials.

Opaque secrets

Opaque is the default Secret type if omitted from a Secret configuration file. When you create a Secret using kubectl, you will use the generic subcommand to indicate an Opaque Secret type. For example, the following command creates an empty Secret of type Opaque.

kubectl create secret generic empty-secret
kubectl get secret empty-secret

The output looks like:

NAME           TYPE     DATA   AGE
empty-secret   Opaque   0      2m6s

The DATA column shows the number of data items stored in the Secret. In this case, 0 means you have created an empty Secret.

Service account token Secrets

A kubernetes.io/service-account-token type of Secret is used to store a token credential that identifies a service account.

Since 1.22, this type of Secret is no longer used to mount credentials into Pods, and obtaining tokens via the TokenRequest API is recommended instead of using service account token Secret objects. Tokens obtained from the TokenRequest API are more secure than ones stored in Secret objects, because they have a bounded lifetime and are not readable by other API clients. You can use the kubectl create token command to obtain a token from the TokenRequest API.

You should only create a service account token Secret object if you can't use the TokenRequest API to obtain a token, and the security exposure of persisting a non-expiring token credential in a readable API object is acceptable to you.

When using this Secret type, you need to ensure that the kubernetes.io/service-account.name annotation is set to an existing service account name. If you are creating both the ServiceAccount and the Secret objects, you should create the ServiceAccount object first.

After the Secret is created, a Kubernetes controller fills in some other fields such as the kubernetes.io/service-account.uid annotation, and the token key in the data field, which is populated with an authentication token.

The following example configuration declares a service account token Secret:

apiVersion: v1
kind: Secret
metadata:
  name: secret-sa-sample
  annotations:
    kubernetes.io/service-account.name: "sa-name"
type: kubernetes.io/service-account-token
data:
  # You can include additional key value pairs as you do with Opaque Secrets
  extra: YmFyCg==

After creating the Secret, wait for Kubernetes to populate the token key in the data field.

See the ServiceAccount documentation for more information on how service accounts work.
You can also check the automountServiceAccountToken field and the serviceAccountName field of the Pod for information on referencing service account credentials from within Pods.

Docker config Secrets

You can use one of the following type values to create a Secret to store the credentials for accessing a container image registry:

  • kubernetes.io/dockercfg
  • kubernetes.io/dockerconfigjson

The kubernetes.io/dockercfg type is reserved to store a serialized ~/.dockercfg which is the legacy format for configuring Docker command line. When using this Secret type, you have to ensure the Secret data field contains a .dockercfg key whose value is content of a ~/.dockercfg file encoded in the base64 format.

The kubernetes.io/dockerconfigjson type is designed for storing a serialized JSON that follows the same format rules as the ~/.docker/config.json file which is a new format for ~/.dockercfg. When using this Secret type, the data field of the Secret object must contain a .dockerconfigjson key, in which the content for the ~/.docker/config.json file is provided as a base64 encoded string.

Below is an example for a kubernetes.io/dockercfg type of Secret:

apiVersion: v1
kind: Secret
metadata:
  name: secret-dockercfg
type: kubernetes.io/dockercfg
data:
  .dockercfg: |
        "<base64 encoded ~/.dockercfg file>"

When you create these types of Secrets using a manifest, the API server checks whether the expected key exists in the data field, and it verifies if the value provided can be parsed as a valid JSON. The API server doesn't validate if the JSON actually is a Docker config file.

When you do not have a Docker config file, or you want to use kubectl to create a Secret for accessing a container registry, you can do:

kubectl create secret docker-registry secret-tiger-docker \
  --docker-email=tiger@acme.example \
  --docker-username=tiger \
  --docker-password=pass1234 \
  --docker-server=my-registry.example:5000

That command creates a Secret of type kubernetes.io/dockerconfigjson. If you dump the .data.dockerconfigjson field from that new Secret and then decode it from base64:

kubectl get secret secret-tiger-docker -o jsonpath='{.data.*}' | base64 -d

then the output is equivalent to this JSON document (which is also a valid Docker configuration file):

{
  "auths": {
    "my-registry.example:5000": {
      "username": "tiger",
      "password": "pass1234",
      "email": "tiger@acme.example",
      "auth": "dGlnZXI6cGFzczEyMzQ="
    }
  }
}

Basic authentication Secret

The kubernetes.io/basic-auth type is provided for storing credentials needed for basic authentication. When using this Secret type, the data field of the Secret must contain one of the following two keys:

  • username: the user name for authentication
  • password: the password or token for authentication

Both values for the above two keys are base64 encoded strings. You can, of course, provide the clear text content using the stringData for Secret creation.

The following manifest is an example of a basic authentication Secret:

apiVersion: v1
kind: Secret
metadata:
  name: secret-basic-auth
type: kubernetes.io/basic-auth
stringData:
  username: admin      # required field for kubernetes.io/basic-auth
  password: t0p-Secret # required field for kubernetes.io/basic-auth

The basic authentication Secret type is provided only for convenience. You can create an Opaque type for credentials used for basic authentication. However, using the defined and public Secret type (kubernetes.io/basic-auth) helps other people to understand the purpose of your Secret, and sets a convention for what key names to expect. The Kubernetes API verifies that the required keys are set for a Secret of this type.

SSH authentication secrets

The builtin type kubernetes.io/ssh-auth is provided for storing data used in SSH authentication. When using this Secret type, you will have to specify a ssh-privatekey key-value pair in the data (or stringData) field as the SSH credential to use.

The following manifest is an example of a Secret used for SSH public/private key authentication:

apiVersion: v1
kind: Secret
metadata:
  name: secret-ssh-auth
type: kubernetes.io/ssh-auth
data:
  # the data is abbreviated in this example
  ssh-privatekey: |
          MIIEpQIBAAKCAQEAulqb/Y ...

The SSH authentication Secret type is provided only for user's convenience. You could instead create an Opaque type Secret for credentials used for SSH authentication. However, using the defined and public Secret type (kubernetes.io/ssh-auth) helps other people to understand the purpose of your Secret, and sets a convention for what key names to expect. and the API server does verify if the required keys are provided in a Secret configuration.

TLS secrets

Kubernetes provides a builtin Secret type kubernetes.io/tls for storing a certificate and its associated key that are typically used for TLS.

One common use for TLS secrets is to configure encryption in transit for an Ingress, but you can also use it with other resources or directly in your workload. When using this type of Secret, the tls.key and the tls.crt key must be provided in the data (or stringData) field of the Secret configuration, although the API server doesn't actually validate the values for each key.

The following YAML contains an example config for a TLS Secret:

apiVersion: v1
kind: Secret
metadata:
  name: secret-tls
type: kubernetes.io/tls
data:
  # the data is abbreviated in this example
  tls.crt: |
        MIIC2DCCAcCgAwIBAgIBATANBgkqh ...
  tls.key: |
        MIIEpgIBAAKCAQEA7yn3bRHQ5FHMQ ...

The TLS Secret type is provided for user's convenience. You can create an Opaque for credentials used for TLS server and/or client. However, using the builtin Secret type helps ensure the consistency of Secret format in your project; the API server does verify if the required keys are provided in a Secret configuration.

When creating a TLS Secret using kubectl, you can use the tls subcommand as shown in the following example:

kubectl create secret tls my-tls-secret \
  --cert=path/to/cert/file \
  --key=path/to/key/file

The public/private key pair must exist before hand. The public key certificate for --cert must be DER format as per Section 5.1 of RFC 7468, and must match the given private key for --key (PKCS #8 in DER format; Section 11 of RFC 7468).

Bootstrap token Secrets

A bootstrap token Secret can be created by explicitly specifying the Secret type to bootstrap.kubernetes.io/token. This type of Secret is designed for tokens used during the node bootstrap process. It stores tokens used to sign well-known ConfigMaps.

A bootstrap token Secret is usually created in the kube-system namespace and named in the form bootstrap-token-<token-id> where <token-id> is a 6 character string of the token ID.

As a Kubernetes manifest, a bootstrap token Secret might look like the following:

apiVersion: v1
kind: Secret
metadata:
  name: bootstrap-token-5emitj
  namespace: kube-system
type: bootstrap.kubernetes.io/token
data:
  auth-extra-groups: c3lzdGVtOmJvb3RzdHJhcHBlcnM6a3ViZWFkbTpkZWZhdWx0LW5vZGUtdG9rZW4=
  expiration: MjAyMC0wOS0xM1QwNDozOToxMFo=
  token-id: NWVtaXRq
  token-secret: a3E0Z2lodnN6emduMXAwcg==
  usage-bootstrap-authentication: dHJ1ZQ==
  usage-bootstrap-signing: dHJ1ZQ==

A bootstrap type Secret has the following keys specified under data:

  • token-id: A random 6 character string as the token identifier. Required.
  • token-secret: A random 16 character string as the actual token secret. Required.
  • description: A human-readable string that describes what the token is used for. Optional.
  • expiration: An absolute UTC time using RFC3339 specifying when the token should be expired. Optional.
  • usage-bootstrap-<usage>: A boolean flag indicating additional usage for the bootstrap token.
  • auth-extra-groups: A comma-separated list of group names that will be authenticated as in addition to the system:bootstrappers group.

The above YAML may look confusing because the values are all in base64 encoded strings. In fact, you can create an identical Secret using the following YAML:

apiVersion: v1
kind: Secret
metadata:
  # Note how the Secret is named
  name: bootstrap-token-5emitj
  # A bootstrap token Secret usually resides in the kube-system namespace
  namespace: kube-system
type: bootstrap.kubernetes.io/token
stringData:
  auth-extra-groups: "system:bootstrappers:kubeadm:default-node-token"
  expiration: "2020-09-13T04:39:10Z"
  # This token ID is used in the name
  token-id: "5emitj"
  token-secret: "kq4gihvszzgn1p0r"
  # This token can be used for authentication
  usage-bootstrap-authentication: "true"
  # and it can be used for signing
  usage-bootstrap-signing: "true"

Immutable Secrets

FEATURE STATE: Kubernetes v1.21 [stable]

Kubernetes lets you mark specific Secrets (and ConfigMaps) as immutable. Preventing changes to the data of an existing Secret has the following benefits:

  • protects you from accidental (or unwanted) updates that could cause applications outages
  • (for clusters that extensively use Secrets - at least tens of thousands of unique Secret to Pod mounts), switching to immutable Secrets improves the performance of your cluster by significantly reducing load on kube-apiserver. The kubelet does not need to maintain a [watch] on any Secrets that are marked as immutable.

Marking a Secret as immutable

You can create an immutable Secret by setting the immutable field to true. For example,

apiVersion: v1
kind: Secret
metadata:
  ...
data:
  ...
immutable: true

You can also update any existing mutable Secret to make it immutable.

Information security for Secrets

Although ConfigMap and Secret work similarly, Kubernetes applies some additional protection for Secret objects.

Secrets often hold values that span a spectrum of importance, many of which can cause escalations within Kubernetes (e.g. service account tokens) and to external systems. Even if an individual app can reason about the power of the Secrets it expects to interact with, other apps within the same namespace can render those assumptions invalid.

A Secret is only sent to a node if a Pod on that node requires it. For mounting secrets into Pods, the kubelet stores a copy of the data into a tmpfs so that the confidential data is not written to durable storage. Once the Pod that depends on the Secret is deleted, the kubelet deletes its local copy of the confidential data from the Secret.

There may be several containers in a Pod. By default, containers you define only have access to the default ServiceAccount and its related Secret. You must explicitly define environment variables or map a volume into a container in order to provide access to any other Secret.

There may be Secrets for several Pods on the same node. However, only the Secrets that a Pod requests are potentially visible within its containers. Therefore, one Pod does not have access to the Secrets of another Pod.

Security recommendations for developers

  • Applications still need to protect the value of confidential information after reading it from an environment variable or volume. For example, your application must avoid logging the secret data in the clear or transmitting it to an untrusted party.
  • If you are defining multiple containers in a Pod, and only one of those containers needs access to a Secret, define the volume mount or environment variable configuration so that the other containers do not have access to that Secret.
  • If you configure a Secret through a manifest, with the secret data encoded as base64, sharing this file or checking it in to a source repository means the secret is available to everyone who can read the manifest. Base64 encoding is not an encryption method, it provides no additional confidentiality over plain text.
  • When deploying applications that interact with the Secret API, you should limit access using authorization policies such as RBAC.
  • In the Kubernetes API, watch and list requests for Secrets within a namespace are extremely powerful capabilities. Avoid granting this access where feasible, since listing Secrets allows the clients to inspect the values of every Secret in that namespace.

Security recommendations for cluster administrators

  • Reserve the ability to watch or list all secrets in a cluster (using the Kubernetes API), so that only the most privileged, system-level components can perform this action.
  • When deploying applications that interact with the Secret API, you should limit access using authorization policies such as RBAC.
  • In the API server, objects (including Secrets) are persisted into etcd; therefore:
    • only allow cluster admistrators to access etcd (this includes read-only access);
    • enable encryption at rest for Secret objects, so that the data of these Secrets are not stored in the clear into etcd;
    • consider wiping / shredding the durable storage used by etcd once it is no longer in use;
    • if there are multiple etcd instances, make sure that etcd is using SSL/TLS for communication between etcd peers.

What's next

4 - Resource Management for Pods and Containers

When you specify a Pod, you can optionally specify how much of each resource a container needs. The most common resources to specify are CPU and memory (RAM); there are others.

When you specify the resource request for containers in a Pod, the kube-scheduler uses this information to decide which node to place the Pod on. When you specify a resource limit for a container, the kubelet enforces those limits so that the running container is not allowed to use more of that resource than the limit you set. The kubelet also reserves at least the request amount of that system resource specifically for that container to use.

Requests and limits

If the node where a Pod is running has enough of a resource available, it's possible (and allowed) for a container to use more resource than its request for that resource specifies. However, a container is not allowed to use more than its resource limit.

For example, if you set a memory request of 256 MiB for a container, and that container is in a Pod scheduled to a Node with 8GiB of memory and no other Pods, then the container can try to use more RAM.

If you set a memory limit of 4GiB for that container, the kubelet (and container runtime) enforce the limit. The runtime prevents the container from using more than the configured resource limit. For example: when a process in the container tries to consume more than the allowed amount of memory, the system kernel terminates the process that attempted the allocation, with an out of memory (OOM) error.

Limits can be implemented either reactively (the system intervenes once it sees a violation) or by enforcement (the system prevents the container from ever exceeding the limit). Different runtimes can have different ways to implement the same restrictions.

Resource types

CPU and memory are each a resource type. A resource type has a base unit. CPU represents compute processing and is specified in units of Kubernetes CPUs. Memory is specified in units of bytes. For Linux workloads, you can specify huge page resources. Huge pages are a Linux-specific feature where the node kernel allocates blocks of memory that are much larger than the default page size.

For example, on a system where the default page size is 4KiB, you could specify a limit, hugepages-2Mi: 80Mi. If the container tries allocating over 40 2MiB huge pages (a total of 80 MiB), that allocation fails.

CPU and memory are collectively referred to as compute resources, or resources. Compute resources are measurable quantities that can be requested, allocated, and consumed. They are distinct from API resources. API resources, such as Pods and Services are objects that can be read and modified through the Kubernetes API server.

Resource requests and limits of Pod and container

For each container, you can specify resource limits and requests, including the following:

  • spec.containers[].resources.limits.cpu
  • spec.containers[].resources.limits.memory
  • spec.containers[].resources.limits.hugepages-<size>
  • spec.containers[].resources.requests.cpu
  • spec.containers[].resources.requests.memory
  • spec.containers[].resources.requests.hugepages-<size>

Although you can only specify requests and limits for individual containers, it is also useful to think about the overall resource requests and limits for a Pod. For a particular resource, a Pod resource request/limit is the sum of the resource requests/limits of that type for each container in the Pod.

Resource units in Kubernetes

CPU resource units

Limits and requests for CPU resources are measured in cpu units. In Kubernetes, 1 CPU unit is equivalent to 1 physical CPU core, or 1 virtual core, depending on whether the node is a physical host or a virtual machine running inside a physical machine.

Fractional requests are allowed. When you define a container with spec.containers[].resources.requests.cpu set to 0.5, you are requesting half as much CPU time compared to if you asked for 1.0 CPU. For CPU resource units, the quantity expression 0.1 is equivalent to the expression 100m, which can be read as "one hundred millicpu". Some people say "one hundred millicores", and this is understood to mean the same thing.

CPU resource is always specified as an absolute amount of resource, never as a relative amount. For example, 500m CPU represents the roughly same amount of computing power whether that container runs on a single-core, dual-core, or 48-core machine.

Memory resource units

Limits and requests for memory are measured in bytes. You can express memory as a plain integer or as a fixed-point number using one of these quantity suffixes: E, P, T, G, M, k. You can also use the power-of-two equivalents: Ei, Pi, Ti, Gi, Mi, Ki. For example, the following represent roughly the same value:

128974848, 129e6, 129M,  128974848000m, 123Mi

Pay attention to the case of the suffixes. If you request 400m of memory, this is a request for 0.4 bytes. Someone who types that probably meant to ask for 400 mebibytes (400Mi) or 400 megabytes (400M).

Container resources example

The following Pod has two containers. Both containers are defined with a request for 0.25 CPU and 64MiB (226 bytes) of memory. Each container has a limit of 0.5 CPU and 128MiB of memory. You can say the Pod has a request of 0.5 CPU and 128 MiB of memory, and a limit of 1 CPU and 256MiB of memory.

---
apiVersion: v1
kind: Pod
metadata:
  name: frontend
spec:
  containers:
  - name: app
    image: images.my-company.example/app:v4
    resources:
      requests:
        memory: "64Mi"
        cpu: "250m"
      limits:
        memory: "128Mi"
        cpu: "500m"
  - name: log-aggregator
    image: images.my-company.example/log-aggregator:v6
    resources:
      requests:
        memory: "64Mi"
        cpu: "250m"
      limits:
        memory: "128Mi"
        cpu: "500m"

How Pods with resource requests are scheduled

When you create a Pod, the Kubernetes scheduler selects a node for the Pod to run on. Each node has a maximum capacity for each of the resource types: the amount of CPU and memory it can provide for Pods. The scheduler ensures that, for each resource type, the sum of the resource requests of the scheduled containers is less than the capacity of the node. Note that although actual memory or CPU resource usage on nodes is very low, the scheduler still refuses to place a Pod on a node if the capacity check fails. This protects against a resource shortage on a node when resource usage later increases, for example, during a daily peak in request rate.

How Kubernetes applies resource requests and limits

When the kubelet starts a container as part of a Pod, the kubelet passes that container's requests and limits for memory and CPU to the container runtime.

On Linux, the container runtime typically configures kernel cgroups that apply and enforce the limits you defined.

  • The CPU limit defines a hard ceiling on how much CPU time that the container can use. During each scheduling interval (time slice), the Linux kernel checks to see if this limit is exceeded; if so, the kernel waits before allowing that cgroup to resume execution.
  • The CPU request typically defines a weighting. If several different containers (cgroups) want to run on a contended system, workloads with larger CPU requests are allocated more CPU time than workloads with small requests.
  • The memory request is mainly used during (Kubernetes) Pod scheduling. On a node that uses cgroups v2, the container runtime might use the memory request as a hint to set memory.min and memory.low.
  • The memory limit defines a memory limit for that cgroup. If the container tries to allocate more memory than this limit, the Linux kernel out-of-memory subsystem activates and, typically, intervenes by stopping one of the processes in the container that tried to allocate memory. If that process is the container's PID 1, and the container is marked as restartable, Kubernetes restarts the container.
  • The memory limit for the Pod or container can also apply to pages in memory backed volumes, such as an emptyDir. The kubelet tracks tmpfs emptyDir volumes as container memory use, rather than as local ephemeral storage.

If a container exceeds its memory request and the node that it runs on becomes short of memory overall, it is likely that the Pod the container belongs to will be evicted.

A container might or might not be allowed to exceed its CPU limit for extended periods of time. However, container runtimes don't terminate Pods or containers for excessive CPU usage.

To determine whether a container cannot be scheduled or is being killed due to resource limits, see the Troubleshooting section.

Monitoring compute & memory resource usage

The kubelet reports the resource usage of a Pod as part of the Pod status.

If optional tools for monitoring are available in your cluster, then Pod resource usage can be retrieved either from the Metrics API directly or from your monitoring tools.

Local ephemeral storage

FEATURE STATE: Kubernetes v1.10 [beta]

Nodes have local ephemeral storage, backed by locally-attached writeable devices or, sometimes, by RAM. "Ephemeral" means that there is no long-term guarantee about durability.

Pods use ephemeral local storage for scratch space, caching, and for logs. The kubelet can provide scratch space to Pods using local ephemeral storage to mount emptyDir volumes into containers.

The kubelet also uses this kind of storage to hold node-level container logs, container images, and the writable layers of running containers.

As a beta feature, Kubernetes lets you track, reserve and limit the amount of ephemeral local storage a Pod can consume.

Configurations for local ephemeral storage

Kubernetes supports two ways to configure local ephemeral storage on a node:

In this configuration, you place all different kinds of ephemeral local data (emptyDir volumes, writeable layers, container images, logs) into one filesystem. The most effective way to configure the kubelet means dedicating this filesystem to Kubernetes (kubelet) data.

The kubelet also writes node-level container logs and treats these similarly to ephemeral local storage.

The kubelet writes logs to files inside its configured log directory (/var/log by default); and has a base directory for other locally stored data (/var/lib/kubelet by default).

Typically, both /var/lib/kubelet and /var/log are on the system root filesystem, and the kubelet is designed with that layout in mind.

Your node can have as many other filesystems, not used for Kubernetes, as you like.

You have a filesystem on the node that you're using for ephemeral data that comes from running Pods: logs, and emptyDir volumes. You can use this filesystem for other data (for example: system logs not related to Kubernetes); it can even be the root filesystem.

The kubelet also writes node-level container logs into the first filesystem, and treats these similarly to ephemeral local storage.

You also use a separate filesystem, backed by a different logical storage device. In this configuration, the directory where you tell the kubelet to place container image layers and writeable layers is on this second filesystem.

The first filesystem does not hold any image layers or writeable layers.

Your node can have as many other filesystems, not used for Kubernetes, as you like.

The kubelet can measure how much local storage it is using. It does this provided that:

  • the LocalStorageCapacityIsolation feature gate is enabled (the feature is on by default), and
  • you have set up the node using one of the supported configurations for local ephemeral storage.

If you have a different configuration, then the kubelet does not apply resource limits for ephemeral local storage.

Setting requests and limits for local ephemeral storage

You can specify ephemeral-storage for managing local ephemeral storage. Each container of a Pod can specify either or both of the following:

  • spec.containers[].resources.limits.ephemeral-storage
  • spec.containers[].resources.requests.ephemeral-storage

Limits and requests for ephemeral-storage are measured in byte quantities. You can express storage as a plain integer or as a fixed-point number using one of these suffixes: E, P, T, G, M, K. You can also use the power-of-two equivalents: Ei, Pi, Ti, Gi, Mi, Ki. For example, the following quantities all represent roughly the same value:

  • 128974848
  • 129e6
  • 129M
  • 123Mi

In the following example, the Pod has two containers. Each container has a request of 2GiB of local ephemeral storage. Each container has a limit of 4GiB of local ephemeral storage. Therefore, the Pod has a request of 4GiB of local ephemeral storage, and a limit of 8GiB of local ephemeral storage.

apiVersion: v1
kind: Pod
metadata:
  name: frontend
spec:
  containers:
  - name: app
    image: images.my-company.example/app:v4
    resources:
      requests:
        ephemeral-storage: "2Gi"
      limits:
        ephemeral-storage: "4Gi"
    volumeMounts:
    - name: ephemeral
      mountPath: "/tmp"
  - name: log-aggregator
    image: images.my-company.example/log-aggregator:v6
    resources:
      requests:
        ephemeral-storage: "2Gi"
      limits:
        ephemeral-storage: "4Gi"
    volumeMounts:
    - name: ephemeral
      mountPath: "/tmp"
  volumes:
    - name: ephemeral
      emptyDir: {}

How Pods with ephemeral-storage requests are scheduled

When you create a Pod, the Kubernetes scheduler selects a node for the Pod to run on. Each node has a maximum amount of local ephemeral storage it can provide for Pods. For more information, see Node Allocatable.

The scheduler ensures that the sum of the resource requests of the scheduled containers is less than the capacity of the node.

Ephemeral storage consumption management

If the kubelet is managing local ephemeral storage as a resource, then the kubelet measures storage use in:

  • emptyDir volumes, except tmpfs emptyDir volumes
  • directories holding node-level logs
  • writeable container layers

If a Pod is using more ephemeral storage than you allow it to, the kubelet sets an eviction signal that triggers Pod eviction.

For container-level isolation, if a container's writable layer and log usage exceeds its storage limit, the kubelet marks the Pod for eviction.

For pod-level isolation the kubelet works out an overall Pod storage limit by summing the limits for the containers in that Pod. In this case, if the sum of the local ephemeral storage usage from all containers and also the Pod's emptyDir volumes exceeds the overall Pod storage limit, then the kubelet also marks the Pod for eviction.

The kubelet supports different ways to measure Pod storage use:

The kubelet performs regular, scheduled checks that scan each emptyDir volume, container log directory, and writeable container layer.

The scan measures how much space is used.

FEATURE STATE: Kubernetes v1.15 [alpha]

Project quotas are an operating-system level feature for managing storage use on filesystems. With Kubernetes, you can enable project quotas for monitoring storage use. Make sure that the filesystem backing the emptyDir volumes, on the node, provides project quota support. For example, XFS and ext4fs offer project quotas.

Kubernetes uses project IDs starting from 1048576. The IDs in use are registered in /etc/projects and /etc/projid. If project IDs in this range are used for other purposes on the system, those project IDs must be registered in /etc/projects and /etc/projid so that Kubernetes does not use them.

Quotas are faster and more accurate than directory scanning. When a directory is assigned to a project, all files created under a directory are created in that project, and the kernel merely has to keep track of how many blocks are in use by files in that project. If a file is created and deleted, but has an open file descriptor, it continues to consume space. Quota tracking records that space accurately whereas directory scans overlook the storage used by deleted files.

If you want to use project quotas, you should:

  • Enable the LocalStorageCapacityIsolationFSQuotaMonitoring=true feature gate using the featureGates field in the kubelet configuration or the --feature-gates command line flag.

  • Ensure that the root filesystem (or optional runtime filesystem) has project quotas enabled. All XFS filesystems support project quotas. For ext4 filesystems, you need to enable the project quota tracking feature while the filesystem is not mounted.

    # For ext4, with /dev/block-device not mounted
    sudo tune2fs -O project -Q prjquota /dev/block-device
    
  • Ensure that the root filesystem (or optional runtime filesystem) is mounted with project quotas enabled. For both XFS and ext4fs, the mount option is named prjquota.

Extended resources

Extended resources are fully-qualified resource names outside the kubernetes.io domain. They allow cluster operators to advertise and users to consume the non-Kubernetes-built-in resources.

There are two steps required to use Extended Resources. First, the cluster operator must advertise an Extended Resource. Second, users must request the Extended Resource in Pods.

Managing extended resources

Node-level extended resources

Node-level extended resources are tied to nodes.

Device plugin managed resources

See Device Plugin for how to advertise device plugin managed resources on each node.

Other resources

To advertise a new node-level extended resource, the cluster operator can submit a PATCH HTTP request to the API server to specify the available quantity in the status.capacity for a node in the cluster. After this operation, the node's status.capacity will include a new resource. The status.allocatable field is updated automatically with the new resource asynchronously by the kubelet.

Because the scheduler uses the node's status.allocatable value when evaluating Pod fitness, the scheduler only takes account of the new value after that asynchronous update. There may be a short delay between patching the node capacity with a new resource and the time when the first Pod that requests the resource can be scheduled on that node.

Example:

Here is an example showing how to use curl to form an HTTP request that advertises five "example.com/foo" resources on node k8s-node-1 whose master is k8s-master.

curl --header "Content-Type: application/json-patch+json" \
--request PATCH \
--data '[{"op": "add", "path": "/status/capacity/example.com~1foo", "value": "5"}]' \
http://k8s-master:8080/api/v1/nodes/k8s-node-1/status

Cluster-level extended resources

Cluster-level extended resources are not tied to nodes. They are usually managed by scheduler extenders, which handle the resource consumption and resource quota.

You can specify the extended resources that are handled by scheduler extenders in scheduler configuration

Example:

The following configuration for a scheduler policy indicates that the cluster-level extended resource "example.com/foo" is handled by the scheduler extender.

  • The scheduler sends a Pod to the scheduler extender only if the Pod requests "example.com/foo".
  • The ignoredByScheduler field specifies that the scheduler does not check the "example.com/foo" resource in its PodFitsResources predicate.
{
  "kind": "Policy",
  "apiVersion": "v1",
  "extenders": [
    {
      "urlPrefix":"<extender-endpoint>",
      "bindVerb": "bind",
      "managedResources": [
        {
          "name": "example.com/foo",
          "ignoredByScheduler": true
        }
      ]
    }
  ]
}

Consuming extended resources

Users can consume extended resources in Pod specs like CPU and memory. The scheduler takes care of the resource accounting so that no more than the available amount is simultaneously allocated to Pods.

The API server restricts quantities of extended resources to whole numbers. Examples of valid quantities are 3, 3000m and 3Ki. Examples of invalid quantities are 0.5 and 1500m.

To consume an extended resource in a Pod, include the resource name as a key in the spec.containers[].resources.limits map in the container spec.

A Pod is scheduled only if all of the resource requests are satisfied, including CPU, memory and any extended resources. The Pod remains in the PENDING state as long as the resource request cannot be satisfied.

Example:

The Pod below requests 2 CPUs and 1 "example.com/foo" (an extended resource).

apiVersion: v1
kind: Pod
metadata:
  name: my-pod
spec:
  containers:
  - name: my-container
    image: myimage
    resources:
      requests:
        cpu: 2
        example.com/foo: 1
      limits:
        example.com/foo: 1

PID limiting

Process ID (PID) limits allow for the configuration of a kubelet to limit the number of PIDs that a given Pod can consume. See PID Limiting for information.

Troubleshooting

My Pods are pending with event message FailedScheduling

If the scheduler cannot find any node where a Pod can fit, the Pod remains unscheduled until a place can be found. An Event is produced each time the scheduler fails to find a place for the Pod. You can use kubectl to view the events for a Pod; for example:

kubectl describe pod frontend | grep -A 9999999999 Events
Events:
  Type     Reason            Age   From               Message
  ----     ------            ----  ----               -------
  Warning  FailedScheduling  23s   default-scheduler  0/42 nodes available: insufficient cpu

In the preceding example, the Pod named "frontend" fails to be scheduled due to insufficient CPU resource on any node. Similar error messages can also suggest failure due to insufficient memory (PodExceedsFreeMemory). In general, if a Pod is pending with a message of this type, there are several things to try:

  • Add more nodes to the cluster.
  • Terminate unneeded Pods to make room for pending Pods.
  • Check that the Pod is not larger than all the nodes. For example, if all the nodes have a capacity of cpu: 1, then a Pod with a request of cpu: 1.1 will never be scheduled.
  • Check for node taints. If most of your nodes are tainted, and the new Pod does not tolerate that taint, the scheduler only considers placements onto the remaining nodes that don't have that taint.

You can check node capacities and amounts allocated with the kubectl describe nodes command. For example:

kubectl describe nodes e2e-test-node-pool-4lw4
Name:            e2e-test-node-pool-4lw4
[ ... lines removed for clarity ...]
Capacity:
 cpu:                               2
 memory:                            7679792Ki
 pods:                              110
Allocatable:
 cpu:                               1800m
 memory:                            7474992Ki
 pods:                              110
[ ... lines removed for clarity ...]
Non-terminated Pods:        (5 in total)
  Namespace    Name                                  CPU Requests  CPU Limits  Memory Requests  Memory Limits
  ---------    ----                                  ------------  ----------  ---------------  -------------
  kube-system  fluentd-gcp-v1.38-28bv1               100m (5%)     0 (0%)      200Mi (2%)       200Mi (2%)
  kube-system  kube-dns-3297075139-61lj3             260m (13%)    0 (0%)      100Mi (1%)       170Mi (2%)
  kube-system  kube-proxy-e2e-test-...               100m (5%)     0 (0%)      0 (0%)           0 (0%)
  kube-system  monitoring-influxdb-grafana-v4-z1m12  200m (10%)    200m (10%)  600Mi (8%)       600Mi (8%)
  kube-system  node-problem-detector-v0.1-fj7m3      20m (1%)      200m (10%)  20Mi (0%)        100Mi (1%)
Allocated resources:
  (Total limits may be over 100 percent, i.e., overcommitted.)
  CPU Requests    CPU Limits    Memory Requests    Memory Limits
  ------------    ----------    ---------------    -------------
  680m (34%)      400m (20%)    920Mi (11%)        1070Mi (13%)

In the preceding output, you can see that if a Pod requests more than 1.120 CPUs or more than 6.23Gi of memory, that Pod will not fit on the node.

By looking at the “Pods” section, you can see which Pods are taking up space on the node.

The amount of resources available to Pods is less than the node capacity because system daemons use a portion of the available resources. Within the Kubernetes API, each Node has a .status.allocatable field (see NodeStatus for details).

The .status.allocatable field describes the amount of resources that are available to Pods on that node (for example: 15 virtual CPUs and 7538 MiB of memory). For more information on node allocatable resources in Kubernetes, see Reserve Compute Resources for System Daemons.

You can configure resource quotas to limit the total amount of resources that a namespace can consume. Kubernetes enforces quotas for objects in particular namespace when there is a ResourceQuota in that namespace. For example, if you assign specific namespaces to different teams, you can add ResourceQuotas into those namespaces. Setting resource quotas helps to prevent one team from using so much of any resource that this over-use affects other teams.

You should also consider what access you grant to that namespace: full write access to a namespace allows someone with that access to remove any resource, including a configured ResourceQuota.

My container is terminated

Your container might get terminated because it is resource-starved. To check whether a container is being killed because it is hitting a resource limit, call kubectl describe pod on the Pod of interest:

kubectl describe pod simmemleak-hra99

The output is similar to:

Name:                           simmemleak-hra99
Namespace:                      default
Image(s):                       saadali/simmemleak
Node:                           kubernetes-node-tf0f/10.240.216.66
Labels:                         name=simmemleak
Status:                         Running
Reason:
Message:
IP:                             10.244.2.75
Containers:
  simmemleak:
    Image:  saadali/simmemleak:latest
    Limits:
      cpu:          100m
      memory:       50Mi
    State:          Running
      Started:      Tue, 07 Jul 2019 12:54:41 -0700
    Last State:     Terminated
      Reason:       OOMKilled
      Exit Code:    137
      Started:      Fri, 07 Jul 2019 12:54:30 -0700
      Finished:     Fri, 07 Jul 2019 12:54:33 -0700
    Ready:          False
    Restart Count:  5
Conditions:
  Type      Status
  Ready     False
Events:
  Type    Reason     Age   From               Message
  ----    ------     ----  ----               -------
  Normal  Scheduled  42s   default-scheduler  Successfully assigned simmemleak-hra99 to kubernetes-node-tf0f
  Normal  Pulled     41s   kubelet            Container image "saadali/simmemleak:latest" already present on machine
  Normal  Created    41s   kubelet            Created container simmemleak
  Normal  Started    40s   kubelet            Started container simmemleak
  Normal  Killing    32s   kubelet            Killing container with id ead3fb35-5cf5-44ed-9ae1-488115be66c6: Need to kill Pod

In the preceding example, the Restart Count: 5 indicates that the simmemleak container in the Pod was terminated and restarted five times (so far). The OOMKilled reason shows that the container tried to use more memory than its limit.

Your next step might be to check the application code for a memory leak. If you find that the application is behaving how you expect, consider setting a higher memory limit (and possibly request) for that container.

What's next

5 - Organizing Cluster Access Using kubeconfig Files

Use kubeconfig files to organize information about clusters, users, namespaces, and authentication mechanisms. The kubectl command-line tool uses kubeconfig files to find the information it needs to choose a cluster and communicate with the API server of a cluster.

By default, kubectl looks for a file named config in the $HOME/.kube directory. You can specify other kubeconfig files by setting the KUBECONFIG environment variable or by setting the --kubeconfig flag.

For step-by-step instructions on creating and specifying kubeconfig files, see Configure Access to Multiple Clusters.

Supporting multiple clusters, users, and authentication mechanisms

Suppose you have several clusters, and your users and components authenticate in a variety of ways. For example:

  • A running kubelet might authenticate using certificates.
  • A user might authenticate using tokens.
  • Administrators might have sets of certificates that they provide to individual users.

With kubeconfig files, you can organize your clusters, users, and namespaces. You can also define contexts to quickly and easily switch between clusters and namespaces.

Context

A context element in a kubeconfig file is used to group access parameters under a convenient name. Each context has three parameters: cluster, namespace, and user. By default, the kubectl command-line tool uses parameters from the current context to communicate with the cluster.

To choose the current context:

kubectl config use-context

The KUBECONFIG environment variable

The KUBECONFIG environment variable holds a list of kubeconfig files. For Linux and Mac, the list is colon-delimited. For Windows, the list is semicolon-delimited. The KUBECONFIG environment variable is not required. If the KUBECONFIG environment variable doesn't exist, kubectl uses the default kubeconfig file, $HOME/.kube/config.

If the KUBECONFIG environment variable does exist, kubectl uses an effective configuration that is the result of merging the files listed in the KUBECONFIG environment variable.

Merging kubeconfig files

To see your configuration, enter this command:

kubectl config view

As described previously, the output might be from a single kubeconfig file, or it might be the result of merging several kubeconfig files.

Here are the rules that kubectl uses when it merges kubeconfig files:

  1. If the --kubeconfig flag is set, use only the specified file. Do not merge. Only one instance of this flag is allowed.

    Otherwise, if the KUBECONFIG environment variable is set, use it as a list of files that should be merged. Merge the files listed in the KUBECONFIG environment variable according to these rules:

    • Ignore empty filenames.
    • Produce errors for files with content that cannot be deserialized.
    • The first file to set a particular value or map key wins.
    • Never change the value or map key. Example: Preserve the context of the first file to set current-context. Example: If two files specify a red-user, use only values from the first file's red-user. Even if the second file has non-conflicting entries under red-user, discard them.

    For an example of setting the KUBECONFIG environment variable, see Setting the KUBECONFIG environment variable.

    Otherwise, use the default kubeconfig file, $HOME/.kube/config, with no merging.

  2. Determine the context to use based on the first hit in this chain:

    1. Use the --context command-line flag if it exists.
    2. Use the current-context from the merged kubeconfig files.

    An empty context is allowed at this point.

  3. Determine the cluster and user. At this point, there might or might not be a context. Determine the cluster and user based on the first hit in this chain, which is run twice: once for user and once for cluster:

    1. Use a command-line flag if it exists: --user or --cluster.
    2. If the context is non-empty, take the user or cluster from the context.

    The user and cluster can be empty at this point.

  4. Determine the actual cluster information to use. At this point, there might or might not be cluster information. Build each piece of the cluster information based on this chain; the first hit wins:

    1. Use command line flags if they exist: --server, --certificate-authority, --insecure-skip-tls-verify.
    2. If any cluster information attributes exist from the merged kubeconfig files, use them.
    3. If there is no server location, fail.
  5. Determine the actual user information to use. Build user information using the same rules as cluster information, except allow only one authentication technique per user:

    1. Use command line flags if they exist: --client-certificate, --client-key, --username, --password, --token.
    2. Use the user fields from the merged kubeconfig files.
    3. If there are two conflicting techniques, fail.
  6. For any information still missing, use default values and potentially prompt for authentication information.

File references

File and path references in a kubeconfig file are relative to the location of the kubeconfig file. File references on the command line are relative to the current working directory. In $HOME/.kube/config, relative paths are stored relatively, and absolute paths are stored absolutely.

Proxy

You can configure kubectl to use a proxy per cluster using proxy-url in your kubeconfig file, like this:

apiVersion: v1
kind: Config

clusters:
- cluster:
    proxy-url: http://proxy.example.org:3128
    server: https://k8s.example.org/k8s/clusters/c-xxyyzz
  name: development

users:
- name: developer

contexts:
- context:
  name: development

What's next

6 - Resource Management for Windows nodes

This page outlines the differences in how resources are managed between Linux and Windows.

On Linux nodes, cgroups are used as a pod boundary for resource control. Containers are created within that boundary for network, process and file system isolation. The Linux cgroup APIs can be used to gather CPU, I/O, and memory use statistics.

In contrast, Windows uses a job object per container with a system namespace filter to contain all processes in a container and provide logical isolation from the host. (Job objects are a Windows process isolation mechanism and are different from what Kubernetes refers to as a Job).

There is no way to run a Windows container without the namespace filtering in place. This means that system privileges cannot be asserted in the context of the host, and thus privileged containers are not available on Windows. Containers cannot assume an identity from the host because the Security Account Manager (SAM) is separate.

Memory management

Windows does not have an out-of-memory process killer as Linux does. Windows always treats all user-mode memory allocations as virtual, and pagefiles are mandatory.

Windows nodes do not overcommit memory for processes. The net effect is that Windows won't reach out of memory conditions the same way Linux does, and processes page to disk instead of being subject to out of memory (OOM) termination. If memory is over-provisioned and all physical memory is exhausted, then paging can slow down performance.

CPU management

Windows can limit the amount of CPU time allocated for different processes but cannot guarantee a minimum amount of CPU time.

On Windows, the kubelet supports a command-line flag to set the scheduling priority of the kubelet process: --windows-priorityclass. This flag allows the kubelet process to get more CPU time slices when compared to other processes running on the Windows host. More information on the allowable values and their meaning is available at Windows Priority Classes. To ensure that running Pods do not starve the kubelet of CPU cycles, set this flag to ABOVE_NORMAL_PRIORITY_CLASS or above.

Resource reservation

To account for memory and CPU used by the operating system, the container runtime, and by Kubernetes host processes such as the kubelet, you can (and should) reserve memory and CPU resources with the --kube-reserved and/or --system-reserved kubelet flags. On Windows these values are only used to calculate the node's allocatable resources.

On Windows, a good practice is to reserve at least 2GiB of memory.

To determine how much CPU to reserve, identify the maximum pod density for each node and monitor the CPU usage of the system services running there, then choose a value that meets your workload needs.