主 题：tc port rate limit only covers IPv4, easy to miss
发 布 者：opt7952
标签分类： 技术
时 间：2026-10-05 13:25:24
内容预览：This afternoon I read the Portshaper post in Tech. It uses tc + HTB per port: download is matched on egress by source port, upload goes through ingress + ifb by destination port, and each port gets its own HTB class.The part worth noting is the IPv4 only bit. The filter matches protocol ip, so if a node listens on IPv6 at the same time, IPv6 traffic will not be counted in that limit. On a dual stack box it can look like limiting is on, while one path is still open.The rules are also in memory only. After a reboot they are gone, and that post puts the config in /root/.tc-limit.conf and the reapply script in /usr/local/bin/tc-limit-apply.sh, with cron @reboot to bring it back.For panels that only limit total traffic, not speed, splitting by port makes sense. The IPv6 point is the one worth c
直达链接： https://www.nodeseek.com/post-965237-1
发 布 者：opt7952
标签分类： 技术
时 间：2026-10-05 13:25:24
内容预览：This afternoon I read the Portshaper post in Tech. It uses tc + HTB per port: download is matched on egress by source port, upload goes through ingress + ifb by destination port, and each port gets its own HTB class.The part worth noting is the IPv4 only bit. The filter matches protocol ip, so if a node listens on IPv6 at the same time, IPv6 traffic will not be counted in that limit. On a dual stack box it can look like limiting is on, while one path is still open.The rules are also in memory only. After a reboot they are gone, and that post puts the config in /root/.tc-limit.conf and the reapply script in /usr/local/bin/tc-limit-apply.sh, with cron @reboot to bring it back.For panels that only limit total traffic, not speed, splitting by port makes sense. The IPv6 point is the one worth c
直达链接： https://www.nodeseek.com/post-965237-1
