Skip to content

Iptables Rules

Kube-OVN uses ipset and iptables to implement gateway NAT functionality in the default VPC overlay Subnets.

The ipset used is shown in the following table:

Name(IPv4/IPv6) Type Usage
ovn40services/ovn60services hash:net Service CIDR
ovn40subnets/ovn60subnets hash:net Overlay Subnet CIDR and NodeLocal DNS IP address
ovn40subnets-nat/ovn60subnets-nat hash:net Overlay Subnet CIDRs that enable NatOutgoing
ovn40subnets-distributed-gw/ovn60subnets-distributed-gw hash:net Overlay Subnet CIDRs that use distributed gateway
ovn40other-node/ovn60other-node hash:net Internal IP addresses for other Nodes
ovn40local-pod-ip-nat/ovn60local-pod-ip-nat hash:ip Deprecated
ovn40subnets-nat-policy hash:net All subnet cidrs configured with natOutgoingPolicyRules
ovn40natpr-418e79269dc5-dst hash:net The dstIPs corresponding to the rule in natOutgoingPolicyRules
ovn40natpr-418e79269dc5-src hash:net The srcIPs corresponding to the rule in natOutgoingPolicyRules

The iptables rules (IPv4) used are shown in the following table:

Table Chain Rule Usage Note
filter INPUT -m set --match-set ovn40services src -j ACCEPT Allow k8s service and pod traffic to pass through --
filter INPUT -m set --match-set ovn40services dst -j ACCEPT Allow k8s service and pod traffic to pass through --
filter INPUT -m set --match-set ovn40subnets src -j ACCEPT Allow k8s service and pod traffic to pass through --
filter INPUT -m set --match-set ovn40subnets dst -j ACCEPT Allow k8s service and pod traffic to pass through --
filter FORWARD -m set --match-set ovn40services src -j ACCEPT Allow k8s service and pod traffic to pass through --
filter FORWARD -m set --match-set ovn40services dst -j ACCEPT Allow k8s service and pod traffic to pass through --
filter FORWARD -m set --match-set ovn40subnets src -j ACCEPT Allow k8s service and pod traffic to pass through --
filter FORWARD -m set --match-set ovn40subnets dst -j ACCEPT Allow k8s service and pod traffic to pass through --
filter FORWARD -s 10.16.0.0/16 -m comment --comment "ovn-subnet-gateway,ovn-default" Used to count packets from the subnet to the external network "10.16.0.0/16" is the cidr of the subnet, the "ovn-subnet-gateway" before the "," in comment is used to identify the iptables rule used to count the subnet inbound and outbound gateway packets, and the "ovn-default" after the "," is the name of the subnet
filter FORWARD -d 10.16.0.0/16 -m comment --comment "ovn-subnet-gateway,ovn-default" Used to count packets from the external network accessing the subnet "10.16.0.0/16" is the cidr of the subnet, the "ovn-subnet-gateway" before the "," in comment is used to identify the iptables rule used to count the subnet inbound and outbound gateway packets, and the "ovn-default" after the "," is the name of the subnet
filter OUTPUT -p udp -m udp --dport 6081 -j MARK --set-xmark 0x0 Clear traffic tag to prevent SNAT UDP: bad checksum on VXLAN interface
nat PREROUTING -m comment --comment "kube-ovn prerouting rules" -j OVN-PREROUTING Enter OVN-PREROUTING chain processing --
nat POSTROUTING -m comment --comment "kube-ovn postrouting rules" -j OVN-POSTROUTING Enter OVN-POSTROUTING chain processing --
nat OVN-PREROUTING -i ovn0 -m set --match-set ovn40subnets src -m set --match-set ovn40services dst -j MARK --set-xmark 0x4000/0x4000 Adding masquerade tags to Pod access service traffic Used when the built-in LB is turned off
nat OVN-PREROUTING -p tcp -m addrtype --dst-type LOCAL -m set --match-set KUBE-NODE-PORT-LOCAL-TCP dst -j MARK --set-xmark 0x80000/0x80000 Add specific tags to ExternalTrafficPolicy for Local's Service traffic (TCP) Only used when kube-proxy is using ipvs mode
nat OVN-PREROUTING -p udp -m addrtype --dst-type LOCAL -m set --match-set KUBE-NODE-PORT-LOCAL-UDP dst -j MARK --set-xmark 0x80000/0x80000 Add specific tags to ExternalTrafficPolicy for Local's Service traffic (UDP) Only used when kube-proxy is using ipvs mode
nat OVN-POSTROUTING -m set --match-set ovn40services src -m set --match-set ovn40subnets dst -m mark --mark 0x4000/0x4000 -j SNAT --to-source Use node IP as the source address for access from node to overlay Pods via service IP。 Works only when kube-proxy is using ipvs mode
nat OVN-POSTROUTING -m mark --mark 0x4000/0x4000 -j MASQUERADE Perform SNAT for specific tagged traffic --
nat OVN-POSTROUTING -m set --match-set ovn40subnets src -m set --match-set ovn40subnets dst -j MASQUERADE Perform SNAT for Service traffic between Pods passing through the node --
nat OVN-POSTROUTING -m mark --mark 0x80000/0x80000 -m set --match-set ovn40subnets-distributed-gw dst -j RETURN For Service traffic where ExternalTrafficPolicy is Local, if the Endpoint uses a distributed gateway, SNAT is not required. --
nat OVN-POSTROUTING -m mark --mark 0x80000/0x80000 -j MASQUERADE For Service traffic where ExternalTrafficPolicy is Local, if the Endpoint uses a centralized gateway, SNAT is required. --
nat OVN-POSTROUTING -p tcp -m tcp --tcp-flags SYN NONE -m conntrack --ctstate NEW -j RETURN No SNAT is performed when the Pod IP is exposed to the outside world --
nat OVN-POSTROUTING -s 10.16.0.0/16 -m set ! --match-set ovn40subnets dst -j SNAT --to-source 192.168.0.101 When the Pod accesses the network outside the cluster, if the subnet is NatOutgoing and a centralized gateway with the specified IP is used, perform SNAT 10.16.0.0/16 is the Subnet CIDR,192.168.0.101 is the specified IP of gateway node
nat OVN-POSTROUTING -m set --match-set ovn40subnets-nat src -m set ! --match-set ovn40subnets dst -j MASQUERADE When the Pod accesses the network outside the cluster, if NatOutgoing is enabled on the subnet, perform SNAT --
nat OVN-POSTROUTING -m set --match-set ovn40subnets-nat-policy src -m set ! --match-set ovn40subnets dst -j OVN-NAT-POLICY When Pod accesses the network outside the cluster, if natOutgoingPolicyRules is enabled on the subnet, the packet with the specified policy will perform SNAT ovn40subnets-nat-policy is all subnet segments configured with natOutgoingPolicyRules
nat OVN-POSTROUTING -m mark --mark 0x90001/0x90001 -j MASQUERADE --random-fully When Pod accesses the network outside the cluster, if natOutgoingPolicyRules is enabled on the subnet, the packet with the specified policy will perform SNAT After coming out of OVN-NAT-POLICY, if it is tagged with 0x90001/0x90001, it will do SNAT
nat OVN-POSTROUTING -m mark --mark 0x90002/0x90002 -j RETURN When Pod accesses the network outside the cluster, if natOutgoingPolicyRules is enabled on the subnet, the packet with the specified policy will perform SNAT After coming out of OVN-NAT-POLICY, if it is tagged with 0x90002/0x90002, it will not do SNAT
nat OVN-NAT-POLICY -s 10.0.11.0/24 -m comment --comment natPolicySubnet-net1 -j OVN-NAT-PSUBNET-aa98851157c5 When Pod accesses the network outside the cluster, if natOutgoingPolicyRules is enabled on the subnet, the packet with the specified policy will perform SNAT 10.0.11.0/24 represents the CIDR of the subnet net1, and the rules under the OVN-NAT-PSUBNET-aa98851157c5 chain correspond to the natOutgoingPolicyRules configuration of this subnet
nat OVN-NAT-PSUBNET-xxxxxxxxxxxx -m set --match-set ovn40natpr-418e79269dc5-src src -m set --match-set ovn40natpr-418e79269dc5-dst dst -j MARK --set-xmark 0x90002/0x90002 When Pod accesses the network outside the cluster, if natOutgoingPolicyRules is enabled on the subnet, the packet with the specified policy will perform SNAT 418e79269dc5 indicates the ID of a rule in natOutgoingPolicyRules, which can be viewed through status.natOutgoingPolicyRules[index].RuleID, indicating that srcIPs meets ovn40natpr-418e79269dc5-src, and dstIPS meets ovn40natpr-418e79269dc5- dst will be marked with tag 0x90002
mangle OVN-OUTPUT -d 10.241.39.2/32 -p tcp -m tcp --dport 80 -j MARK --set-xmark 0x90003/0x90003 Introduce kubelet's detection traffic to tproxy with a specific mark
mangle OVN-PREROUTING -d 10.241.39.2/32 -p tcp -m tcp --dport 80 -j TPROXY --on-port 8102 --on-ip 172.18.0.3 --tproxy-mark 0x90004/0x90004 Introduce kubelet's detection traffic to tproxy with a specific mark

微信群 Slack Twitter Support

Comments