calico/node container is deployed to every node (on Kubernetes, by a DaemonSet), and runs three internal daemons:
- Felix, the Calico daemon that runs on every node and provides endpoints.
- BIRD, the BGP daemon that distributes routing information to other nodes.
- confd, a daemon that watches the Calico datastore for config changes and updates BIRD’s config files.
For manifest-based installations,
calico/node is primarily configured through environment
variables, typically set in the deployment manifest. Individual nodes may also be updated through the Node
calico/node can also be configured through the Calico Operator.
The rest of this page lists the available configuration options, and is followed by specific considerations for various settings.
calico/node does not need to be configured directly when installed by the operator. For a complete operator
configuration reference, see the installation API reference documentation.
Configuring the default IP pool(s)
Calico uses IP pools to configure how addresses are allocated to pods, and how networking works for certain sets of addresses. You can see the full schema for IP pools here.
calico/node can be configured to create a default IP pool for you, but only if none already
exist in the cluster. The following options control the parameters on the created pool.
|CALICO_IPV4POOL_CIDR||The IPv4 Pool to create if none exists at start up. It is invalid to define this variable and NO_DEFAULT_POOLS. [Default: First not used in locally of (192.168.0.0/16, 172.16.0.0/16, .., 172.31.0.0/16) ]||IPv4 CIDR|
|CALICO_IPV4POOL_BLOCK_SIZE||Block size to use for the IPv4 Pool created at startup. Block size for IPv4 should be in the range 20-32 (inclusive) [Default: ||int|
|CALICO_IPV4POOL_IPIP||IPIP Mode to use for the IPv4 Pool created at start up. If set to a value other than ||Always, CrossSubnet, Never ("Off" is also accepted as a synonym for "Never")|
|CALICO_IPV4POOL_VXLAN||VXLAN Mode to use for the IPv4 Pool created at start up. If set to a value other than ||Always, CrossSubnet, Never|
|CALICO_IPV4POOL_NAT_OUTGOING||Controls NAT Outgoing for the IPv4 Pool created at start up. [Default: ||boolean|
|CALICO_IPV4POOL_NODE_SELECTOR||Controls the NodeSelector for the IPv4 Pool created at start up. [Default: ||selector|
|CALICO_IPV6POOL_CIDR||The IPv6 Pool to create if none exists at start up. It is invalid to define this variable and NO_DEFAULT_POOLS. [Default: ||IPv6 CIDR|
|CALICO_IPV6POOL_BLOCK_SIZE||Block size to use for the IPv6 POOL created at startup. Block size for IPv6 should be in the range 116-128 (inclusive) [Default: ||int|
|CALICO_IPV6POOL_VXLAN||VXLAN Mode to use for the IPv6 Pool created at start up. [Default: ||Always, CrossSubnet, Never|
|CALICO_IPV6POOL_NAT_OUTGOING||Controls NAT Outgoing for the IPv6 Pool created at start up. [Default: ||boolean|
|CALICO_IPV6POOL_NODE_SELECTOR||Controls the NodeSelector for the IPv6 Pool created at start up. [Default: ||selector|
|CALICO_IPV4POOL_DISABLE_BGP_EXPORT||Disable exporting routes over BGP for the IPv4 Pool created at start up. [Default: ||boolean|
|CALICO_IPV6POOL_DISABLE_BGP_EXPORT||Disable exporting routes over BGP for the IPv6 Pool created at start up. [Default: ||boolean|
|NO_DEFAULT_POOLS||Prevents Calico from creating a default pool if one does not exist. [Default: ||boolean|
Configuring BGP Networking
BGP configuration for Calico nodes is normally configured through the Node, BGPConfiguration, and BGPPeer resources.
calico/node also exposes some options to allow setting certain fields on these objects, as described
|NODENAME||A unique identifier for this host. See node name determination for more details.||lowercase string|
|IP||The IPv4 address to assign this host or detection behavior at startup. Refer to IP setting for the details of the behavior possible with this field.||IPv4|
|IP6||The IPv6 address to assign this host or detection behavior at startup. Refer to IP setting for the details of the behavior possible with this field.||IPv6|
|IP_AUTODETECTION_METHOD||The method to use to autodetect the IPv4 address for this host. This is only used when the IPv4 address is being autodetected. See IP Autodetection methods for details of the valid methods. [Default: ||string|
|IP6_AUTODETECTION_METHOD||The method to use to autodetect the IPv6 address for this host. This is only used when the IPv6 address is being autodetected. See IP Autodetection methods for details of the valid methods. [Default: ||string|
|AS||The AS number for this node. When specified, the value is saved in the node resource configuration for this host, overriding any previously configured value. When omitted, if an AS number has been previously configured in the node resource, that AS number is used for the peering. When omitted, if an AS number has not yet been configured in the node resource, the node will use the global value (see example modifying Global BGP settings for details.)||int|
|CALICO_ROUTER_ID||Sets the ||string|
|CALICO_K8S_NODE_REF||The name of the corresponding node object in the Kubernetes API. When set, used for correlating this node with events from the Kubernetes API.||string|
Configuring Datastore Access
|DATASTORE_TYPE||Type of datastore. [Default: ||kubernetes, etcdv3|
Configuring Kubernetes Datastore Access
|KUBECONFIG||When using the Kubernetes datastore, the location of a kubeconfig file to use.||string|
|K8S_API_ENDPOINT||Location of the Kubernetes API. Not required if using kubeconfig.||string|
|K8S_CERT_FILE||Location of a client certificate for accessing the Kubernetes API.||string|
|K8S_KEY_FILE||Location of a client key for accessing the Kubernetes API.||string|
|K8S_CA_FILE||Location of a CA for accessing the Kubernetes API.||string|
When Calico is configured to use the Kubernetes API as the datastore, the environments used for BGP configuration are ignored—this includes selection of the node AS number (AS) and all of the IP selection options (IP, IP6, IP_AUTODETECTION_METHOD, IP6_AUTODETECTION_METHOD).
Configuring etcd Datastore Access
|ETCD_ENDPOINTS||A comma separated list of etcd endpoints [Example: ||string|
|ETCD_DISCOVERY_SRV||Domain name to discover etcd endpoints via SRV records. Mutually exclusive with ||string|
|ETCD_KEY_FILE||Path to the file containing the private key matching the ||string|
|ETCD_CERT_FILE||Path to the file containing the client certificate issued to ||string|
|ETCD_CA_CERT_FILE||Path to the file containing the root certificate of the certificate authority (CA) that issued the etcd server certificate. Configures ||string|
|CALICO_DISABLE_FILE_LOGGING||Disables logging to file. [Default: "false"]||string|
|CALICO_STARTUP_LOGLEVEL||The log severity above which startup ||DEBUG, INFO, WARNING, ERROR, CRITICAL, or NONE (case-insensitive)|
Configuring CNI Plugin
calico/node has a few options that are configurable based on the CNI plugin and CNI plugin
configuration used on the cluster.
|USE_POD_CIDR||Use the Kubernetes ||boolean|
|CALICO_MANAGE_CNI||Tells Calico to update the kubeconfig file at /host/etc/cni/net.d/calico-kubeconfig on credentials change. [Default: true]||boolean|
Other Environment Variables
|DISABLE_NODE_IP_CHECK||Skips checks for duplicate Node IPs. This can reduce the load on the cluster when a large number of Nodes are restarting. [Default: ||boolean|
|WAIT_FOR_DATASTORE||Wait for connection to datastore before starting. If a successful connection is not made, node will shutdown. [Default: ||boolean|
|CALICO_NETWORKING_BACKEND||The networking backend to use. In ||bird, vxlan, none|
|CLUSTER_TYPE||Contains comma delimited list of indicators about this cluster. e.g. k8s, mesos, kubeadm, canal, bgp||string|
Node name determination
calico/node must know the name of the node on which it is running. The node name is used to
retrieve the Node resource configured for this node if it exists, or to create a new node resource representing the node if it does not. It is
also used to associate the node with per-node BGP configuration, felix configuration, and endpoints.
When launched, the
calico/node container sets the node name according to the following order of precedence:
- The value specified in the
NODENAMEenvironment variable, if set.
- The value specified in
/var/lib/calico/nodename, if it exists.
- The value specified in the
HOSTNAMEenvironment variable, if set.
- The hostname as returned by the operating system, converted to lowercase.
Once the node has determined its name, the value will be cached in
/var/lib/calico/nodename for future use.
For example, if given the following conditions:
/var/lib/calico/nodenamedoes not exist
- The operating system returns "host-A.internal.myorg.com" for the hostname
calico/node will use "host-a" for its name and will write the value in
/var/lib/calico/nodename. If calico/node
is then restarted, it will use the cached value of "host-a" read from the file on disk.
The IP (for IPv4) and IP6 (for IPv6) environment variables are used to set, force autodetection, or disable auto detection of the address for the appropriate IP version for the node. When the environment variable is set, the address is saved in the node resource configuration for this host, overriding any previously configured value.
calico/node will attempt to detect subnet information from the host, and augment the provided address if possible.
IP setting special case values
There are several special case values that can be set in the IP(6) environment variables, they are:
- Not set or empty string: Any previously set address on the node
resource will be used. If no previous address is set on the node resource
the two versions behave differently:
- IP will do autodetection of the IPv4 address and set it on the node resource.
- IP6 will not do autodetection.
autodetect: Autodetection will always be performed for the IP address and the detected address will overwrite any value configured in the node resource.
none: Autodetection will not be performed (this is useful to disable IPv4).
IP autodetection methods
When Calico is used for routing, each node must be configured with an IPv4
address and/or an IPv6 address that will be used to route between
nodes. To eliminate node specific IP address configuration, the
container can be configured to autodetect these IP addresses. In many systems,
there might be multiple physical interfaces on a host, or possibly multiple IP
addresses configured on a physical interface. In these cases, there are
multiple addresses to choose from and so autodetection of the correct address
can be tricky.
The IP autodetection methods are provided to improve the selection of the correct address, by limiting the selection based on suitable criteria for your deployment.
The following sections describe the available IP autodetection methods.
first-found option enumerates all interface IP addresses and returns the
first valid IP address (based on IP version and type of address) on
the first valid interface. Certain known "local" interfaces
are omitted, such as the docker bridge. The order that both the interfaces
and the IP addresses are listed is system dependent.
This is the default detection method. However, since this method only makes a very simplified guess, it is recommended to either configure the node with a specific IP address, or to use one of the other detection methods.
kubernetes-internal-ip method will select the first internal IP address listed in the Kubernetes node's
can-reach method uses your local routing to determine which IP address
will be used to reach the supplied destination. Both IP addresses and domain
names may be used.
Example using IP addresses:
Example using domain names:
interface method uses the supplied interface regular expression
to enumerate matching interfaces and to return the first IP address on
the first matching interface. The order that both the interfaces
and the IP addresses are listed is system dependent.
Example with valid IP address on interface eth0, eth1, eth2 etc.:
skip-interface method uses the supplied interface regular expression
to exclude interfaces and to return the first IP address on the first
interface that does not match. The order that both the interfaces
and the IP addresses are listed is system dependent.
Example with valid IP address on interface exclude enp6s0f0, eth0, eth1, eth2 etc.:
cidr method will select any IP address from the node that falls within the given CIDRs. For example:
calico/node container supports an exec readiness endpoint.
To access this endpoint, use the following command.
docker exec calico-node /bin/calico-node [flag]
[flag] with one or more of the following.
The BIRD readiness endpoint ensures that the BGP mesh is healthy by verifying that all BGP peers are established and no graceful restart is in progress. If the BIRD readiness check is failing due to unreachable peers that are no longer in the cluster, see decommissioning a node.
CALICO_ROUTER_ID for IPv6 only system
Setting CALICO_ROUTER_ID to value
hash will use a hash of the configured nodename for the router ID. This should only be used in IPv6-only systems with no IPv4 address to use for the router ID. Since each node chooses its own router ID in isolation, it is possible for two nodes to pick the same ID resulting in a clash. The probability of such a clash grows with cluster size so this feature should not be used in a large cluster (500+ nodes).