System requirements

5 MINUTE READ

Node requirements

  • x86-64 processor with at least 2 cores, 8.0GB RAM and 20 GB free disk space

  • Linux kernel 3.10 or later with required dependencies. The following distributions have the required kernel, its dependencies, and are known to work well with Calico Enterprise and Kubernetes.
    • CentOS 8
    • Ubuntu 18.04 and 20.04
    • RHEL 8
    • Debian 10
  • Calico Enterprise must be able to manage cali* interfaces on the host. When IPIP is enabled (the default), Calico Enterprise also needs to be able to manage tunl* interfaces. When VXLAN is enabled, Calico Enterprise also needs to be able to manage the vxlan.calico interface.

    Note: Many Linux distributions, such as most of the above, include NetworkManager. By default, NetworkManager does not allow Calico Enterprise to manage interfaces. If your nodes have NetworkManager, complete the steps in Preventing NetworkManager from controlling Calico Enterprise interfaces before installing Calico Enterprise.

  • If your Linux distribution comes with installed Firewalld or another iptables manager it should be disabled. These may interfere with rules added by Calico Enterprise and result in unexpected behavior.

    Note: If a host firewall is needed, it can be configured by Calico Enterprise HostEndpoint and GlobalNetworkPolicy. More information about configuration at Security for host.

  • In order to properly run Elasticsearch, nodes must be configured according to the Elasticsearch system configuration documentation.

  • The Typha autoscaler requires a minimum number of Linux worker nodes based on total number of schedulable nodes.

    Total schedulable nodes Required Linux nodes for Typha replicas
    1 1
    2 2
    3 3
    up to 250 4
    up to 500 5
    up to 1000 6
    up to 1500 7
    up to 2000 8
    2000 or more 10

Key/value store

Calico Enterprise v3.11 requires a key/value store accessible by all Calico Enterprise components. On Kubernetes, you can configure Calico Enterprise to access an etcdv3 cluster directly or to use the Kubernetes API datastore.

Network requirements

Ensure that your hosts and firewalls allow the necessary traffic based on your configuration. See Component architecture to view the following components.

Configuration Host(s) Port/protocol
Calico Enterprise networking options IP-in-IP (default) Protocol number 4
  BGP TCP 179
  VXLAN UDP 4789
Cluster scaling Any Calico Enterprise networking option above with Typha agents enabled TCP 5473 (default)
APIs Kubernetes API (kube-apiserver) to access Kubernetes API datastore Often TCP 443 or 6443*
  Calico Enterprise API server TCP 8080 and 5443 (default)
Nodes calico-node (Felix, BIRD, confd) TCP 9090 (default)
Component metrics Prometheus metrics TCP 9081 (default)
  Prometheus BGP metrics TCP 9900 (default)
  Prometheus API service TCP 9090 (default)
  Prometheus Alertmanager TCP 9093 (default)
Logs and storage Elasticsearch with fluentd datastore TCP 9200 (default)
  Elasticssearch for cloud (ECK) TCP 9443 (default)
  Elasticsearch gateway TCP 5444 (default)
Visibility and troubleshooting Kibana TCP 5601 (default)
  Packet capture API TCP 8444 (default)
  Calico Enterprise Manager UI TCP 9443 (default)
Intrusion Detection System (IDS) Calico Enterprise intrusion detection TCP 5443 (default)
Compliance Calico Enterprise compliance TCP 5443 (default)
Multi-cluster management Additional port required for Manager UI TCP 9449

* The value passed to kube-apiserver using the --secure-port flag. If you cannot locate this, check the targetPort value returned by kubectl get svc kubernetes -o yaml.

Privileges

Ensure that Calico Enterprise has the CAP_SYS_ADMIN privilege.

The simplest way to provide the necessary privilege is to run Calico Enterprise as root or in a privileged container. When installed as a Kubernetes daemon set, Calico Enterprise meets this requirement by running as a privileged container. This requires that the kubelet be allowed to run privileged containers. There are two ways this can be achieved.

Kubernetes requirements

Supported Kubernetes versions

We test Calico Enterprise v3.11 against the following Kubernetes versions.

  • v1.20
  • v1.21
  • v1.22

Other versions may work, but we do not actively test Calico Enterprise v3.11 against them.

Supported managed Kubernetes versions

We test Calico Enterprise v3.11 against the following managed Kubernetes versions.

  • AKS: 1.21
  • GKE: 1.21
  • EKS: 1.21

Other versions may work, but we do not actively test Calico Enterprise v3.11 against them.

Supported Mirantis Kubernetes Engine (MKE) versions

We test Calico Enterprise v3.11 against the following MKE versions.

Mirantis Kubernetes Engine 3.4.4

Other versions may work, but we do not actively test Calico Enterprise v3.11 against them.

CNI plug-in enabled

Calico Enterprise is installed as a CNI plugin. The kubelet must be configured to use CNI networking by passing the --network-plugin=cni argument. (On kubeadm, this is the default.)

Other network providers

Calico Enterprise must be the only network provider in each cluster. We do not currently support migrating a cluster with another network provider to use Calico Enterprise networking.

Supported kube-proxy modes

Calico Enterprise supports the following kube-proxy modes:

  • iptables (default)

IP pool configuration

The IP range selected for pod IP addresses cannot overlap with any other IP ranges in your network, including:

  • The Kubernetes service cluster IP range
  • The range from which host IPs are allocated

Application layer policy requirements

Note that Kubernetes version 1.16+ requires Istio version 1.2 or greater. Note that Istio version 1.9 requires Kubernetes version 1.17+.

Kernel dependencies

Tip: If you are using one of the recommended distributions, you will already satisfy these.

Due to the large number of distributions and kernel version out there, it’s hard to be precise about the names of the particular kernel modules that are required to run Calico Enterprise. However, in general, you’ll need:

  • The iptables modules (both the “legacy” and “nft” variants are supported). These are typically broken up into many small modules, one for each type of match criteria and one for each type of action. Calico Enterprise requires:

    • The “base” modules (including the IPv6 versions if IPv6 is enabled in your cluster).
    • At least the following match criteria: set, rpfilter, addrtype, comment, conntrack, icmp, tcp, udp, ipvs, icmpv6 (if IPv6 is enabled in your kernel), mark, multiport, rpfilter, sctp, ipvs (if using kube-proxy in IPVS mode).
    • At least the following actions: REJECT, ACCEPT, DROP, LOG.
  • IP sets support.
  • Netfilter Conntrack support compiled in (with SCTP support if using SCTP).
  • IPVS support if using kube-proxy in IPVS mode.
  • IPIP, VXLAN, Wireguard support, if using Calico Enterprise networking in one of those modes.
  • eBPF (including the tc hook support) and XDP (if you want to use the eBPF dataplane).