Speed up forwarding with a flowtable

Moves established flows to the fast path. Add flags offload for hardware offload on the NIC

nft add flowtable inet filter ft '{ hook ingress priority 0 ; devices = { eth0, eth1 } ; }' && nft add rule inet filter forward ct state established flow add @ft

tool
nftables12 commands
tags
nftablesperformancenetworkkernel

More nftables commands

  • nft add chain inet filter input '{ type filter hook input priority 0 ; policy drop ; }'

    Base chain that denies everything by default. Note: add accept rules first, or you will lock yourself out

  • nft add set inet filter blocklist '{ type ipv4_addr ; flags interval ; }' && nft add element inet filter blocklist '{ 10.0.0.0/8, 192.0.2.5 }'

    A named set you can reuse in many rules. flags interval lets you store CIDR ranges, not only single hosts

  • nft add rule inet filter input tcp dport vmap '{ 22 : accept, 80 : accept, 443 : accept, 3306 : drop }'

    vmap maps keys to verdicts in O(1). It replaces a long chain of dport == rules

  • nft -a list ruleset

    -a prints # handle N for each rule. You need the handle to delete a single rule

  • nft delete rule inet filter input handle 7

    Remove one rule without flushing the chain. Get the handle from nft -a list ruleset

all 12 nftables commands