netwaves netwaves-api-us-central-1-default-us-central-1-0

Global

SettingValueWhy it matters
net.ipv4.ip_forward01 routes IPv4 between interfaces
net.ipv6.conf.all.forwarding01 routes IPv6 between interfaces, and turns off accepting RAs unless accept_ra is 2
net.ipv4.conf.all.rp_filter2reverse-path filter; the effective value per interface is the higher of all and the interface's own
net.ipv4.conf.default.rp_filter2rp_filter for interfaces created later
net.ipv6.conf.all.seg6_enabled0with the interface's own value, accepts SRv6 packets
net.ipv6.seg6_flowlabel0
net.ipv6.fib_multipath_hash_policy0
net.ipv4.conf.all.arp_ignore0
net.ipv4.conf.all.arp_announce0
net.ipv6.conf.all.disable_ipv60
net.netfilter.nf_conntrack_count0connections tracked now
net.netfilter.nf_conntrack_max65536new connections are dropped once the count reaches this
net.core.somaxconn4096upper bound for listen() backlogs
net.ipv4.tcp_max_syn_backlog256
net.core.netdev_max_backlog1000per-CPU input queue; overflow shows as softnet drops
net.core.rmem_max4194304
net.core.wmem_max4194304
net.ipv4.tcp_rmem4096 131072 32081088min, default, max receive buffer
net.ipv4.tcp_wmem4096 16384 4194304min, default, max send buffer
net.ipv4.tcp_congestion_controlbbr
net.ipv4.ip_local_port_range32768 60999
net.ipv4.ping_group_range0 2147483647

Per interface

Interfaceipv4 forwardingrp_filterproxy_arparp_ignoreipv6 forwardingaccept_raseg6_enableddisable_ipv6
all02000100
default02000100
eth002000100
lo00000100

rp_filter: 0 off, 1 strict, 2 loose; the kernel applies the higher of all and the interface's value. accept_ra: 0 off, 1 only when not forwarding, 2 always.