Masquerading

IPv4 addresses used for pods are typically allocated from RFC1918 private address blocks and thus, not publicly routable. Cilium will automatically masquerade the source IP address of all traffic that is leaving the cluster to the IPv4 address of the node as the node’s IP address is already routable on the network.

This behavior can be disabled with the option enable-ipv4-masquerade: false for IPv4 and enable-ipv6-masquerade: false for IPv6 traffic leaving the host.

Configuration

Setting the routable CIDR: The default behavior is to exclude any destination within the IP allocation CIDR of the local node. If the pod IPs are routable across a wider network, that network can be specified with the option: ipv4-native-routing-cidr: 10.0.0.0/8 (or ipv6-native-routing-cidr: fd00::/100 for IPv6 addresses) in which case all destinations within that CIDR will not be masqueraded.
Setting the masquerading interface: See Implementation Modes for configuring the masquerading interfaces.

Implementation Modes

eBPF-based

Note

IPv6 BPF masquerading is a beta feature. Please provide feedback and file a GitHub issue if you experience any problems. IPv4 BPF masquerading is production-ready.

The eBPF-based implementation is the most efficient implementation. It requires Linux kernel 4.19 and can be enabled with the bpf.masquerade=true helm option.

The current implementation depends on the BPF NodePort feature. The dependency will be removed in the future (GitHub issue 13732).

Masquerading can take place only on those devices which run the eBPF masquerading program. This means that a packet sent from a pod to an outside address will be masqueraded (to an output device IPv4 address), if the output device runs the program. If not specified, the program will be automatically attached to the devices selected by the BPF NodePort device detection mechanism. To manually change this, use the devices helm option. Use cilium status to determine which devices the program is running on:

$ kubectl -n kube-system exec ds/cilium -- cilium-dbg status | grep Masquerading
Masquerading:   BPF (ip-masq-agent)   [eth0, eth1]  10.0.0.0/16

From the output above, the program is running on the eth0 and eth1 devices.

The eBPF-based masquerading can masquerade packets of the following L4 protocols:

TCP
UDP
ICMP

Note

For ICMP, support is limited to Echo request, Echo reply, and the error message “Destination unreachable, fragmentation required, and DF flag set”.

By default, all packets from a pod destined to an IP address outside of the ipv4-native-routing-cidr range are masqueraded, except for packets destined to other cluster nodes (as with ipv6-native-routing-cidr for IPv6). The preceding output shows the exclusion CIDR of cilium status (10.0.0.0/16).

Note

When eBPF-masquerading is enabled, traffic from pods to the External IP of cluster nodes will also not be masqueraded. The eBPF implementation differs from the iptables-based masquerading on that aspect. This limitation is tracked at GitHub issue 17177.

To allow more fine-grained control, Cilium implements ip-masq-agent in eBPF which can be enabled with the ipMasqAgent.enabled=true helm option.

The eBPF-based ip-masq-agent supports the nonMasqueradeCIDRs, masqLinkLocal, and masqLinkLocalIPv6 options set in a configuration file. A packet sent from a pod to a destination which belongs to any CIDR from the nonMasqueradeCIDRs is not going to be masqueraded. If the configuration file is empty, the agent will provision the following non-masquerade CIDRs:

10.0.0.0/8
172.16.0.0/12
192.168.0.0/16
100.64.0.0/10
192.0.0.0/24
192.0.2.0/24
192.88.99.0/24
198.18.0.0/15
198.51.100.0/24
203.0.113.0/24
240.0.0.0/4

In addition, if the masqLinkLocal is not set or set to false, then 169.254.0.0/16 is appended to the non-masquerade CIDRs list. For IPv6, if masqLinkLocalIPv6 is not set or set to false, fe80::/10 is appended.

The agent uses Fsnotify to track updates to the configuration file, so the original resyncInterval option is unnecessary.

The example below shows how to configure the agent via ConfigMap and to verify it:

apiVersion: v1
kind: ConfigMap
metadata:
  name: ip-masq-agent
data:
  config: |
    nonMasqueradeCIDRs:
    - 10.0.0.0/8
    - 172.16.0.0/12
    - 192.168.0.0/16
    masqLinkLocal: true

$ kubectl create -n kube-system -f https://raw.githubusercontent.com/cilium/cilium/1.15.4/examples/kubernetes-ip-masq-agent/rfc1918.yaml

$ # Wait ~60s until the ConfigMap is propagated into the configuration file

$ kubectl -n kube-system exec ds/cilium -- cilium-dbg bpf ipmasq list
IP PREFIX/ADDRESS
10.0.0.0/8
172.16.0.0/12
192.168.0.0/16

Alternatively, you can pass --set ipMasqAgent.config.nonMasqueradeCIDRs='{10.0.0.0/8,172.16.0.0/12,192.168.0.0/16}' and --set ipMasqAgent.config.masqLinkLocal=false (or with the corresponding option, for IPv6) when installing Cilium via Helm to configure the ip-masq-agent as above.

Note

eBPF-based masquerading for IPv6 is not compatible with the host firewall.

iptables-based

This is the legacy implementation that will work on all kernel versions.

The default behavior will masquerade all traffic leaving on a non-Cilium network device. This typically leads to the correct behavior. In order to limit the network interface on which masquerading should be performed, the option egress-masquerade-interfaces: eth0 can be used.

Note

It is possible to specify an interface prefix as well, by specifying eth+, all interfaces matching the prefix eth will be used for masquerading.

For the advanced case where the routing layer would select different source addresses depending on the destination CIDR, the option enable-masquerade-to-route-source: "true" can be used in order to masquerade to the source addresses rather than to the primary interface address. The latter is then only considered as a catch-all fallback, and for the default routes. For these advanced cases the user needs to ensure that there are no overlapping destination CIDRs as routes on the relevant masquerading interfaces.