This document proposes a generic plugin-based networking solution for application containers on Linux, the _Container Networking Interface_, or _CNI_.
It is derived from the [rkt Networking Proposal][rkt-networking-proposal], which aimed to satisfy many of the [design considerations][rkt-networking-design] for networking in [rkt][rkt-github].
For the purposes of this proposal, we define two terms very specifically:
- _container_ can be considered synonymous with a [Linux _network namespace_][namespaces]. What unit this corresponds to depends on a particular container runtime implementation: for example, in implementations of the [App Container Spec][appc-github] like rkt, each _pod_ runs in a unique network namespace. In [Docker][docker], on the other hand, network namespaces generally exist for each separate Docker container.
- _network_ refers to a group of entities that are uniquely addressable that can communicate amongst each other. This could be either an individual container (as specified above), a machine, or some other network device (e.g. a router). Containers can be conceptually _added to_ or _removed from_ one or more networks.
The network configuration allows for fields to change values between invocations. For this purpose there is an optional field "args" which should contain the varying information.
The container runtime sequentially sets up the networks by executing the corresponding plugin for each network.
Upon completion of the container lifecycle, the runtime executes the plugins in reverse order (relative to the order in which they were added) to disconnect them from the networks.
A CNI plugin is responsible for inserting a network interface into the container network namespace (e.g. one end of a veth pair) and making any necessary changes on the host (e.g. attaching other end of veth into a bridge).
It should then assign the IP to the interface and setup the routes consistent with IP Address Management section by invoking appropriate IPAM plugin.
- **Container ID**. This is optional but recommended, and should be unique across an administrative domain while the container is live (it may be reused in the future). For example, an environment with an IPAM system may require that each container is allocated a unique ID and that each IP allocation can thus be correlated back to a particular container. As another example, in appc implementations this would be the _pod ID_.
- **Name of the interface inside the container**. This is the name that should be assigned to the interface created inside the container (network namespace); consequently it must comply with the standard Linux restrictions on interface names.
- Result:
- **IPs assigned to the interface**. This is either an IPv4 address, an IPv6 address, or both.
The executable command-line API uses the type of network (see [Network Configuration](#network-configuration) below) as the name of the executable to invoke.
It will then look for this executable in a list of predefined directories. Once found, it will invoke the executable using the following environment variables for argument passing:
-`CNI_ARGS`: Extra arguments passed in by the user at invocation time. Alphanumeric key-value pairs separated by semicolons; for example, "FOO=BAR;ABC=123"
-`CNI_PATH`: Colon-separated list of paths to search for CNI plugin executables
Network configuration in JSON format is streamed to the plugin through stdin. This means it is not tied to a particular file on disk and can contain information which changes between invocations.
Success is indicated by a return code of zero and the following JSON printed to stdout in the case of the ADD command. This should be the same output as was returned by the IPAM plugin (see [IP Allocation](#ip-allocation) for details).
The network configuration is described in JSON form. The configuration can be stored on disk or generated from other sources by the container runtime. The following fields are well-known and have the following meaning:
-`args` (dictionary): Optional additional arguments provided by the container runtime. For example a dictionary of labels could be passed to CNI plugins by adding them to a labels field under `args`.
-`ipMasq` (boolean): Optional (if supported by the plugin). Set up an IP masquerade on the host for this network. This is necessary if the host will act as a gateway to subnets that are not able to route to the IP assigned to the container.
-`routes` (list): List of subnets (in CIDR notation) that the CNI plugin should ensure are reachable by routing them through the network. Each entry is a dictionary containing:
-`nameservers` (list of strings): list of a priority-ordered list of DNS nameservers that this network is aware of. Each entry in the list is a string containing either an IPv4 or an IPv6 address.
-`domain` (string): the local domain used for short hostname lookups.
-`search` (list of strings): list of priority ordered search domains for short hostname lookups. Will be preferred over `domain` by most resolvers.
-`options` (list of strings): list of options that can be passed to the resolver
Plugins may define additional fields that they accept and may generate an error if called with unknown fields. The exception to this is the `args` field may be used to pass arbitrary data which may be ignored by plugins.
As part of its operation, a CNI plugin is expected to assign (and maintain) an IP address to the interface and install any necessary routes relevant for that interface. This gives the CNI plugin great flexibility but also places a large burden on it. Many CNI plugins would need to have the same code to support several IP management schemes that users may desire (e.g. dhcp, host-local).
To lessen the burden and make IP management strategy be orthogonal to the type of CNI plugin, we define a second type of plugin -- IP Address Management Plugin (IPAM plugin). It is however the responsibility of the CNI plugin to invoke the IPAM plugin at the proper moment in its execution. The IPAM plugin is expected to determine the interface IP/subnet, Gateway and Routes and return this information to the "main" plugin to apply. The IPAM plugin may obtain the information via a protocol (e.g. dhcp), data stored on a local filesystem, the "ipam" section of the Network Configuration file or a combination of the above.
Like CNI plugins, the IPAM plugins are invoked by running an executable. The executable is searched for in a predefined list of paths, indicated to the CNI plugin via `CNI_PATH`. The IPAM Plugin receives all the same environment variables that were passed in to the CNI plugin. Just like the CNI plugin, IPAM receives the network configuration via stdin.
The "dns" field contains a dictionary consisting of common DNS information.
-`nameservers` (list of strings): list of a priority-ordered list of DNS nameservers that this network is aware of. Each entry in the list is a string containing either an IPv4 or an IPv6 address.
-`domain` (string): the local domain used for short hostname lookups.
-`search` (list of strings): list of priority ordered search domains for short hostname lookups. Will be preferred over `domain` by most resolvers.
-`options` (list of strings): list of options that can be passed to the resolver
See [CNI Plugin Result](#result) section for more information.
- **host-local**: Select an unused (by other containers on the same host) IP within the specified range.
- **dhcp**: Use DHCP protocol to acquire and maintain a lease. The DHCP requests will be sent via the created container interface; therefore, the associated network must support broadcast.
#### Notes
- Routes are expected to be added with a 0 metric.
- A default route may be specified via "0.0.0.0/0". Since another network might have already configured the default route, the CNI plugin should be prepared to skip over its default route definition.
