فهرست فصل‌ها دانلود آرشیوناشر

Kubernetes Docs · نسخهٔ f284a9e7c87f · CC BY 4.0 برای مستندات و مجوزهای همراه

<!-- overview -->

_Pods_ are the smallest deployable units of computing that you can create and manage in Kubernetes.

A _Pod_ (as in a pod of whales or pea pod) is a group of one or more {{< glossary_tooltip text="containers" term_id="container" >}}, with shared storage and network resources, and a specification for how to run the containers. A Pod's contents are always co-located and co-scheduled, and run in a shared context. A Pod models an application-specific "logical host": it contains one or more application containers which are relatively tightly coupled. In non-cloud contexts, applications executed on the same physical or virtual machine are analogous to cloud applications executed on the same logical host.

As well as application containers, a Pod can contain {{< glossary_tooltip text="init containers" term_id="init-container" >}} that run during Pod startup. You can also inject {{< glossary_tooltip text="ephemeral containers" term_id="ephemeral-container" >}} for debugging a running Pod.

<!-- body -->

What is a Pod?

{{< note >}} You need to install a container runtime into each node in the cluster so that Pods can run there. {{< /note >}}

The shared context of a Pod is a set of Linux namespaces, cgroups, and potentially other facets of isolation - the same things that isolate a {{< glossary_tooltip text="container" term_id="container" >}}. Within a Pod's context, the individual applications may have further sub-isolations applied.

A Pod is similar to a set of containers with shared namespaces and shared filesystem volumes.

Pods in a Kubernetes cluster are used in two main ways:

  • Pods that run a single container. The "one-container-per-Pod" model is the

most common Kubernetes use case; in this case, you can think of a Pod as a wrapper around a single container; Kubernetes manages Pods rather than managing the containers directly.

  • Pods that run multiple containers that need to work together. A Pod can

encapsulate an application composed of multiple co-located containers that are tightly coupled and need to share resources. These co-located containers form a single cohesive unit.

Grouping multiple co-located and co-managed containers in a single Pod is a relatively advanced use case. You should use this pattern only in specific instances in which your containers are tightly coupled.

You don't need to run multiple containers to provide replication (for resilience or capacity); if you need multiple replicas, see Workload management.

Using Pods

The following is an example of a Pod which consists of a container running the image nginx:1.14.2.

{{% code_sample file="pods/simple-pod.yaml" %}}

To create the Pod shown above, run the following command:

shell
kubectl apply -f https://k8s.io/examples/pods/simple-pod.yaml

Pods are generally not created directly and are created using workload resources. See Working with Pods for more information on how Pods are used with workload resources.

Workload resources for managing pods

Usually you don't need to create Pods directly, even singleton Pods. Instead, create them using workload resources such as {{< glossary_tooltip text="Deployment" term_id="deployment" >}} or {{< glossary_tooltip text="Job" term_id="job" >}}. If your Pods need to track state, consider the {{< glossary_tooltip text="StatefulSet" term_id="statefulset" >}} resource.

Each Pod is meant to run a single instance of a given application. If you want to scale your application horizontally (to provide more overall resources by running more instances), you should use multiple Pods, one for each instance. In Kubernetes, this is typically referred to as _replication_. Replicated Pods are usually created and managed as a group by a workload resource and its {{< glossary_tooltip text="controller" term_id="controller" >}}.

See Pods and controllers for more information on how Kubernetes uses workload resources, and their controllers, to implement application scaling and auto-healing.

Pods natively provide two kinds of shared resources for their constituent containers: networking and storage.

Working with Pods

You'll rarely create individual Pods directly in Kubernetes—even singleton Pods. This is because Pods are designed as relatively ephemeral, disposable entities. When a Pod gets created (directly by you, or indirectly by a {{< glossary_tooltip text="controller" term_id="controller" >}}), the new Pod is scheduled to run on a {{< glossary_tooltip term_id="node" >}} in your cluster. The Pod remains on that node until the Pod finishes execution, the Pod object is deleted, the Pod is *evicted* for lack of resources, or the node fails.

{{< note >}} Restarting a container in a Pod should not be confused with restarting a Pod. A Pod is not a process, but an environment for running container(s). A Pod persists until it is deleted. {{< /note >}}

The name of a Pod must be a valid DNS subdomain value, but this can produce unexpected results for the Pod hostname. For best compatibility, the name should follow the more restrictive rules for a DNS label.

Pod OS

{{< feature-state state="stable" for_k8s_version="v1.25" >}}

You should set the .spec.os.name field to either windows or linux to indicate the operating system that the containers in that Pod require. These two are the only operating systems supported for now by Kubernetes. In the future, this list may be expanded.

The kubelet refuses to run a Pod if the value of .spec.os.name does not match the operating system of the node. However, in Kubernetes v{{< skew currentVersion >}}, the value of .spec.os.name does not affect how the {{< glossary_tooltip text="kube-scheduler" term_id="kube-scheduler" >}} picks a node for the Pod to run on. In any cluster where there is more than one operating system for running nodes, you should set the kubernetes.io/os label correctly on each node, and define pods with a nodeSelector based on the operating system label. If you do not specify an OS-based nodeSelector, the kube-scheduler may assign the Pod to a node whose OS is unsuitable for its containers. The Pod security standards also use the .spec.os.name field to avoid enforcing policies that aren't relevant to the operating system.

Pods and controllers

You can use workload resources to create and manage multiple Pods for you. A controller for the resource handles replication and rollout and automatic healing in case of Pod failure. For example, if a Node fails, a controller notices that Pods on that Node have stopped working and creates a replacement Pod. The scheduler places the replacement Pod onto a healthy Node.

Here are some examples of workload resources that manage one or more Pods:

  • {{< glossary_tooltip text="Deployment" term_id="deployment" >}}
  • {{< glossary_tooltip text="StatefulSet" term_id="statefulset" >}}
  • {{< glossary_tooltip text="DaemonSet" term_id="daemonset" >}}

Specifying a scheduling group

{{< feature-state feature_gate_name="GenericWorkload" >}}

By default, Kubernetes schedules every Pod individually. However, some tightly-coupled applications need a group of Pods to be scheduled simultaneously to function correctly.

You can link a Pod to a PodGroup using the scheduling group field (spec.schedulingGroup). This tells the kube-scheduler that the Pod belongs to a specific group, enabling it to apply group-level coordinated placement decisions for the entire group at once.

Pod templates

Controllers for {{< glossary_tooltip text="workload" term_id="workload" >}} resources create Pods from a _pod template_ and manage those Pods on your behalf.

PodTemplates are specifications for creating Pods, and are included in workload resources such as Deployments, Jobs, and DaemonSets.

Each controller for a workload resource uses the PodTemplate inside the workload object to make actual Pods. The PodTemplate is part of the desired state of whatever workload resource you used to run your app.

When you create a Pod, you can include environment variables in the Pod template for the containers that run in the Pod.

The sample below is a manifest for a simple Job with a template that starts one container. The container in that Pod prints a message then pauses.

yaml
apiVersion: batch/v1
kind: Job
metadata:
  name: hello
spec:
  template:
    # This is the pod template
    spec:
      containers:
      - name: hello
        image: busybox:1.28
        command: ['sh', '-c', 'echo "Hello, Kubernetes!" && sleep 3600']
      restartPolicy: OnFailure
    # The pod template ends here

Modifying the pod template or switching to a new pod template has no direct effect on the Pods that already exist. If you change the pod template for a workload resource, that resource needs to create replacement Pods that use the updated template.

For example, the StatefulSet controller ensures that the running Pods match the current pod template for each StatefulSet object. If you edit the StatefulSet to change its pod template, the StatefulSet starts to create new Pods based on the updated template. Eventually, all of the old Pods are replaced with new Pods, and the update is complete.

Each workload resource implements its own rules for handling changes to the Pod template. If you want to read more about StatefulSet specifically, read Update strategy in the StatefulSet Basics tutorial.

On Nodes, the {{< glossary_tooltip term_id="kubelet" text="kubelet" >}} does not directly observe or manage any of the details around pod templates and updates; those details are abstracted away. That abstraction and separation of concerns simplifies system semantics, and makes it feasible to extend the cluster's behavior without changing existing code.

Pod update and replacement

As mentioned in the previous section, when the Pod template for a workload resource is changed, the controller creates new Pods based on the updated template instead of updating or patching the existing Pods.

Kubernetes doesn't prevent you from managing Pods directly. It is possible to update some fields of a running Pod, in place. However, Pod update operations like [patch](/docs/reference/generated/kubernetes-api/{{< param "version" >}}/#patch-pod-v1-core), and [replace](/docs/reference/generated/kubernetes-api/{{< param "version" >}}/#replace-pod-v1-core) have some limitations:

  • Most of the metadata about a Pod is immutable. For example, you cannot

change the namespace, name, uid, or creationTimestamp fields.

  • If the metadata.deletionTimestamp is set, no new entry can be added to the

metadata.finalizers list.

  • Pod updates may not change fields other than spec.containers[*].image,

spec.initContainers[*].image, spec.activeDeadlineSeconds, spec.terminationGracePeriodSeconds, spec.tolerations or spec.schedulingGates. For spec.tolerations, you can only add new entries.

  • When updating the spec.activeDeadlineSeconds field, two types of updates

are allowed:

  1. setting the unassigned field to a positive number;
  2. updating the field from a positive number to a smaller, non-negative

number.

Pod subresources

The above update rules apply to regular pod updates, but other pod fields can be updated through _subresources_.

  • Resize: The resize subresource allows container resources (spec.containers[*].resources) to be updated.

See Resize Container Resources for more details.

  • Ephemeral Containers: The ephemeralContainers subresource allows

{{< glossary_tooltip text="ephemeral containers" term_id="ephemeral-container" >}} to be added to a Pod. See Ephemeral Containers for more details.

  • Status: The status subresource allows the pod status to be updated.

This is typically only used by the Kubelet and other system controllers.

  • Binding: The binding subresource allows setting the pod's spec.nodeName via a Binding request.

This is typically only used by the {{< glossary_tooltip text="scheduler" term_id="kube-scheduler" >}}.

Pod generation

  • The metadata.generation field is unique. It will be automatically set by the

system such that new pods have a metadata.generation of 1, and every update to mutable fields in the pod's spec will increment the metadata.generation by 1.

{{< feature-state feature_gate_name="PodObservedGenerationTracking" >}}

  • observedGeneration is a field that is captured in the status section of the Pod

object. The Kubelet will set status.observedGeneration to track the pod state to the current pod status. The pod's status.observedGeneration will reflect the metadata.generation of the pod at the point that the pod status is being reported.

{{< note >}} The status.observedGeneration field is managed by the kubelet and external controllers should not modify this field. {{< /note >}}

Different status fields may either be associated with the metadata.generation of the current sync loop, or with the metadata.generation of the previous sync loop. The key distinction is whether a change in the spec is reflected directly in the status or is an indirect result of a running process.

Direct Status Updates

For status fields where the allocated spec is directly reflected, the observedGeneration will be associated with the current metadata.generation (Generation N).

This behavior applies to:

  • Resize Status: The status of a resource resize operation.
  • Allocated Resources: The resources allocated to the Pod after a resize.
  • Ephemeral Containers: When a new ephemeral container is added, and it is in Waiting state.

Indirect Status Updates

For status fields that are an indirect result of running the spec, the observedGeneration will be associated with the metadata.generation of the previous sync loop (Generation N-1).

This behavior applies to:

  • Container Image: The ContainerStatus.ImageID reflects the image from the previous generation until the new image

is pulled and the container is updated.

  • Actual Resources: During an in-progress resize, the actual resources in use still belong to the previous generation's

request.

  • Container state: During an in-progress resize, with require restart policy reflects the previous generation's

request.

  • activeDeadlineSeconds & terminationGracePeriodSeconds & deletionTimestamp: The effects of these fields on the

Pod's status are a result of the previously observed specification.

Resource sharing and communication

Pods enable data sharing and communication among their constituent containers.

Storage in Pods {#pod-storage}

A Pod can specify a set of shared storage {{< glossary_tooltip text="volumes" term_id="volume" >}}. All containers in the Pod can access the shared volumes, allowing those containers to share data. Volumes also allow persistent data in a Pod to survive in case one of the containers within needs to be restarted. See Storage for more information on how Kubernetes implements shared storage and makes it available to Pods.

Pod networking

Each Pod is assigned a unique IP address for each address family. Every container in a Pod shares the network namespace, including the IP address and network ports. Inside a Pod (and only then), the containers that belong to the Pod can communicate with one another using localhost. When containers in a Pod communicate with entities *outside the Pod*, they must coordinate how they use the shared network resources (such as ports). Within a Pod, containers share an IP address and port space, and can find each other via localhost. The containers in a Pod can also communicate with each other using standard inter-process communications like SystemV semaphores or POSIX shared memory. Containers in different Pods have distinct IP addresses and can not communicate by OS-level IPC without special configuration. Containers that want to interact with a container running in a different Pod can use IP networking to communicate.

Containers within the Pod see the system hostname as being the same as the configured name for the Pod. There's more about this in the networking section.

Pod security settings {#pod-security}

To set security constraints on Pods and containers, you use the securityContext field in the Pod specification. This field gives you granular control over what a Pod or individual containers can do. See Advanced Pod Configuration for more details.

For basic security configuration, you should meet the Baseline Pod security standard and run containers as non-root. You can set simple security contexts:

yaml
apiVersion: v1
kind: Pod
metadata:
  name: security-context-demo
spec:
  securityContext:
    runAsUser: 1000
    runAsGroup: 3000
    fsGroup: 2000
  containers:
  - name: sec-ctx-demo
    image: busybox
    command: ["sh", "-c", "sleep 1h"]

For advanced security context configuration including capabilities, seccomp profiles, and detailed security options, see the security concepts section.

  • To learn about kernel-level security constraints that you can use,

see Linux kernel security constraints for Pods and containers.

  • To learn more about the Pod security context, see

Configure a Security Context for a Pod or Container.

Resource requests and limits

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).

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.

CPU limits are enforced by CPU throttling. When a container approaches its CPU limit, the kernel restricts its access to CPU. Memory limits are enforced by the kernel with out-of-memory (OOM) kills when a container exceeds its limit.

{{< note >}} Setting CPU limits involves a trade-off. CPU limits help prevent noisy neighbor problems where a single workload starves others on the same node. This is especially important in multi-tenant environments. However, CPU limits can cause throttling even when the node has spare CPU capacity, potentially degrading latency-sensitive workload performance. Whether to set CPU limits depends on your environment, workload characteristics, and isolation requirements. {{< /note >}}

For details on resource units, enforcement behavior, and configuration examples, see Resource Management for Pods and Containers.

Static Pods

_Static Pods_ are managed directly by the kubelet daemon on a specific node, without the {{< glossary_tooltip text="API server" term_id="kube-apiserver" >}} observing them. Whereas most Pods are managed by the control plane (for example, a {{< glossary_tooltip text="Deployment" term_id="deployment" >}}), for static Pods, the kubelet directly supervises each static Pod (and restarts it if it fails).

Static Pods are always bound to one {{< glossary_tooltip term_id="kubelet" >}} on a specific node. The main use for static Pods is to run a self-hosted control plane: in other words, using the kubelet to supervise the individual control plane components.

For details, see Static Pods.

Pods with multiple containers {#how-pods-manage-multiple-containers}

Pods are designed to support multiple cooperating processes (as containers) that form a cohesive unit of service. The containers in a Pod are automatically co-located and co-scheduled on the same physical or virtual machine in the cluster. The containers can share resources and dependencies, communicate with one another, and coordinate when and how they are terminated.

<!--intentionally repeats some text from earlier in the page, with more detail --> Pods in a Kubernetes cluster are used in two main ways:

  • Pods that run a single container. The "one-container-per-Pod" model is the

most common Kubernetes use case; in this case, you can think of a Pod as a wrapper around a single container; Kubernetes manages Pods rather than managing the containers directly.

  • Pods that run multiple containers that need to work together. A Pod can

encapsulate an application composed of multiple co-located containers that are tightly coupled and need to share resources. These co-located containers form a single cohesive unit of service—for example, one container serving data stored in a shared volume to the public, while a separate {{< glossary_tooltip text="sidecar container" term_id="sidecar-container" >}} refreshes or updates those files. The Pod wraps these containers, storage resources, and an ephemeral network identity together as a single unit.

For example, you might have a container that acts as a web server for files in a shared volume, and a separate sidecar container that updates those files from a remote source, as in the following diagram:

{{< figure src="/images/docs/pod.svg" alt="Pod creation diagram" class="diagram-medium" >}}

Some Pods have {{< glossary_tooltip text="init containers" term_id="init-container" >}} as well as {{< glossary_tooltip text="app containers" term_id="app-container" >}}. By default, init containers run and complete before the app containers are started.

You can also have sidecar containers that provide auxiliary services to the main application Pod (for example: a service mesh).

{{< feature-state feature_gate_name="SidecarContainers" >}}

Enabled by default, the SidecarContainers feature gate allows you to specify restartPolicy: Always for init containers. Setting the Always restart policy ensures that the containers where you set it are treated as _sidecars_ that are kept running during the entire lifetime of the Pod. Containers that you explicitly define as sidecar containers start up before the main application Pod and remain running until the Pod is shut down.

Container probes

A _probe_ is a diagnostic performed periodically by the kubelet on a container. To perform a diagnostic, the kubelet can invoke different actions:

  • ExecAction (performed with the help of the container runtime)
  • TCPSocketAction (checked directly by the kubelet)
  • HTTPGetAction (checked directly by the kubelet)

You can read more about probes in the Pod Lifecycle documentation.

{{% heading "whatsnext" %}}

  • Learn about the lifecycle of a Pod.
  • Read about PodDisruptionBudget

and how you can use it to manage application availability during disruptions.

  • Pod is a top-level resource in the Kubernetes REST API.

The {{< api-reference page="core/pod-v1" >}} object definition describes the object in detail.

  • The Distributed System Toolkit: Patterns for Composite Containers explains common layouts for Pods with more than one container.
  • Read about Pod topology spread constraints
  • Read Advanced Pod Configuration to learn the topic in detail.

That page covers aspects of Pod configuration beyond the essentials, including:

  • PriorityClasses
  • RuntimeClasses
  • advanced ways to configure _scheduling_: the way that Kubernetes decides which node a Pod should run on.

To understand the context for why Kubernetes wraps a common Pod API in other resources (such as {{< glossary_tooltip text="StatefulSets" term_id="statefulset" >}} or {{< glossary_tooltip text="Deployments" term_id="deployment" >}}), you can read about the prior art, including:

  • Aurora
  • Borg
  • Marathon
  • Omega
  • Tupperware.