Primary system and API concepts are documented in the User guide.
Overall API conventions are described in the API conventions doc.
Complete API details are documented via Swagger. The Kubernetes apiserver (aka "master") exports an API that can be used to retrieve the Swagger spec for the Kubernetes API, by default at /swaggerapi
, and a UI you can use to browse the API documentation at /swagger-ui
. We also periodically update a statically generated UI.
Remote access to the API is discussed in the access doc.
The Kubernetes API also serves as the foundation for the declarative configuration schema for the system. The Kubectl command-line tool can be used to create, update, delete, and get API objects.
Kubernetes also stores its serialized state (currently in etcd) in terms of the API resources.
Kubernetes itself is decomposed into multiple components, which interact through its API.
In our experience, any system that is successful needs to grow and change as new use cases emerge or existing ones change. Therefore, we expect the Kubernetes API to continuously change and grow. However, we intend to not break compatibility with existing clients, for an extended period of time. In general, new API resources and new resource fields can be expected to be added frequently. Elimination of resources or fields will require following a deprecation process. The precise deprecation policy for eliminating features is TBD, but once we reach our 1.0 milestone, there will be a specific policy.
What constitutes a compatible change and how to change the API are detailed by the API change document.
Fine-grain resource evolution alone makes it difficult to eliminate fields or restructure resource representations. Therefore, Kubernetes supports multiple API versions, each at a different API path prefix, such as /api/v1beta3
. These are simply different interfaces to read and/or modify the same underlying resources. In general, all API resources are accessible via all API versions, though there may be some cases in the future where that is not true.
Distinct API versions present more clear, consistent views of system resources and behavior than intermingled, independently evolved resources. They also provide a more straightforward mechanism for controlling access to end-of-lifed and/or experimental APIs.
The API and release versioning proposal describes the current thinking on the API version evolution process.
As of June 4, 2015, the Kubernetes v1 API has been enabled by default. The v1beta1 and v1beta2 APIs were deleted on June 1, 2015. v1beta3 is planned to be deleted on July 6, 2015.
We're working to convert all documentation and examples to v1. A simple API conversion tool has been written to simplify the translation process. Use kubectl create --validate
in order to validate your json or yaml against our Swagger spec.
Changes to services are the most significant difference between v1beta3 and v1.
- The
service.spec.portalIP
property is renamed toservice.spec.clusterIP
. - The
service.spec.createExternalLoadBalancer
property is removed. Specifyservice.spec.type: "LoadBalancer"
to create an external load balancer instead. - The
service.spec.publicIPs
property is deprecated and now calledservice.spec.deprecatedPublicIPs
. This property will be removed entirely when v1beta3 is removed. The vast majority of users of this field were using it to expose services on ports on the node. Those users should specifyservice.spec.type: "NodePort"
instead. Read External Services for more info. If this is not sufficient for your use case, please file an issue or contact @thockin.
Some other difference between v1beta3 and v1:
- The
pod.spec.containers[*].privileged
andpod.spec.containers[*].capabilities
properties are now nested under thepod.spec.containers[*].securityContext
property. See Security Contexts. - The
pod.spec.host
property is renamed topod.spec.nodeName
. - The
endpoints.subsets[*].addresses.IP
property is renamed toendpoints.subsets[*].addresses.ip
. - The
pod.status.containerStatuses[*].state.termination
andpod.status.containerStatuses[*].lastState.termination
properties are renamed topod.status.containerStatuses[*].state.terminated
andpod.status.containerStatuses[*].lastState.terminated
respectively. - The
pod.status.Condition
property is renamed topod.status.conditions
. - The
status.details.id
property is renamed tostatus.details.name
.
Some important differences between v1beta1/2 and v1beta3:
- The resource
id
is now calledname
. name
,labels
,annotations
, and other metadata are now nested in a map calledmetadata
desiredState
is now calledspec
, andcurrentState
is now calledstatus
/minions
has been moved to/nodes
, and the resource has kindNode
- The namespace is required (for all namespaced resources) and has moved from a URL parameter to the path:
/api/v1beta3/namespaces/{namespace}/{resource_collection}/{resource_name}
. If you were not using a namespace before, usedefault
here. - The names of all resource collections are now lower cased - instead of
replicationControllers
, usereplicationcontrollers
. - To watch for changes to a resource, open an HTTP or Websocket connection to the collection query and provide the
?watch=true
query parameter along with the desiredresourceVersion
parameter to watch from. - The
labels
query parameter has been renamed tolabelSelector
. - The
fields
query parameter has been renamed tofieldSelector
. - The container
entrypoint
has been renamed tocommand
, andcommand
has been renamed toargs
. - Container, volume, and node resources are expressed as nested maps (e.g.,
resources{cpu:1}
) rather than as individual fields, and resource values support scaling suffixes rather than fixed scales (e.g., milli-cores). - Restart policy is represented simply as a string (e.g.,
"Always"
) rather than as a nested map (always{}
). - Pull policies changed from
PullAlways
,PullNever
, andPullIfNotPresent
toAlways
,Never
, andIfNotPresent
. - The volume
source
is inlined intovolume
rather than nested. - Host volumes have been changed from
hostDir
tohostPath
to better reflect that they can be files or directories.