Azure Linux虚拟机第二块网卡公网IPv6无法正常通信求助

环境信息 #

网卡配置 #

eth0
Mac地址: 00:22:48:8f:ba:bf
私有IPv4地址: 10.0.0.4
公网IPv4地址: 20.25.226.73
私有IPv6地址: abc:abc:abc:abc::6
公网IPv6地址: 2a01:111:f100:1000::9d37:d42b

eth1
Mac地址: 00:22:48:8f:64:21
私有IPv4地址: 10.0.0.14
公网IPv4地址: 172.183.16.91
私有IPv6地址: abc:abc:abc:abc::16
公网IPv6地址: 2603:1030:603::324

• 两块网卡均归属虚拟网络子网 10.0.0.0/24 和 abc:abc:abc:abc::/64
• 防火墙入站规则已开放22/80/443/3389/ICMP端口

现有Netplan配置 #

当前/etc/netplan/50-cloud-init.yaml配置如下：

network:
    ethernets:
        eth0:
            dhcp4: true
            dhcp4-overrides: &id001
                route-metric: 100
            dhcp6: true
            dhcp6-overrides: *id001
            match:
                driver: hv_netvsc
                macaddress: 00:22:48:8f:ba:bf
            set-name: eth0
            routes:
             - to: 10.0.0.0/24
               via: 10.0.0.1
               metric: 100
               table: 200
             - to: 0.0.0.0/0
               via: 10.0.0.1
               table: 200
            routing-policy:
             - from: 10.0.0.4/32
               table: 200
             - to: 10.0.0.4/32
               table: 200
        eth1:
            dhcp4: true
            dhcp4-overrides: &id002
                route-metric: 200
            dhcp6: true
            dhcp6-overrides: *id002
            match:
                driver: hv_netvsc
                macaddress: 00:22:48:8f:64:21
            set-name: eth1
            routes:
             - to: 10.0.0.0/24
               via: 10.0.0.1
               metric: 200
               table: 201
             - to: 0.0.0.0/0
               via: 10.0.0.1
               table: 201
            routing-policy:
             - from: 10.0.0.14/32
               table: 201
             - to: 10.0.0.14/32
               table: 201
    version: 2

当前问题现象 #

• IPv4通信正常：两块网卡可互相ping通，绑定网站无异常
• IPv6通信异常：
  1. 通过eth0公网IPv4/IPv6登录后，执行telnet -6 ipv6.telnetmyip.com返回eth0的公网IPv6地址，通信正常
  2. 通过eth1公网IPv4登录后，执行上述命令仍返回eth0的公网IPv6地址，未使用eth1的公网IPv6

解决方案：修改Netplan配置 #

问题根源是现有配置仅对IPv4做了策略路由，未针对IPv6配置对应路由表及策略规则。修改后的完整配置如下：

network:
    ethernets:
        eth0:
            dhcp4: true
            dhcp4-overrides: &id001
                route-metric: 100
            dhcp6: true
            dhcp6-overrides: *id001
            match:
                driver: hv_netvsc
                macaddress: 00:22:48:8f:ba:bf
            set-name: eth0
            # IPv4路由及策略
            routes:
             - to: 10.0.0.0/24
               via: 10.0.0.1
               metric: 100
               table: 200
             - to: 0.0.0.0/0
               via: 10.0.0.1
               table: 200
            routing-policy:
             - from: 10.0.0.4/32
               table: 200
             - to: 10.0.0.4/32
               table: 200
             # 添加IPv6策略路由规则
             - from: abc:abc:abc:abc::6/128
               table: 200
             - to: abc:abc:abc:abc::6/128
               table: 200
            # 添加IPv6路由表条目
            routes6:
             - to: abc:abc:abc:abc::/64
               via: abc:abc:abc:abc::1
               metric: 100
               table: 200
             - to: ::/0
               via: abc:abc:abc:abc::1
               table: 200
        eth1:
            dhcp4: true
            dhcp4-overrides: &id002
                route-metric: 200
            dhcp6: true
            dhcp6-overrides: *id002
            match:
                driver: hv_netvsc
                macaddress: 00:22:48:8f:64:21
            set-name: eth1
            # IPv4路由及策略
            routes:
             - to: 10.0.0.0/24
               via: 10.0.0.1
               metric: 200
               table: 201
             - to: 0.0.0.0/0
               via: 10.0.0.1
               table: 201
            routing-policy:
             - from: 10.0.0.14/32
               table: 201
             - to: 10.0.0.14/32
               table: 201
             # 添加IPv6策略路由规则
             - from: abc:abc:abc:abc::16/128
               table: 201
             - to: abc:abc:abc:abc::16/128
               table: 201
            # 添加IPv6路由表条目
            routes6:
             - to: abc:abc:abc:abc::/64
               via: abc:abc:abc:abc::1
               metric: 200
               table: 201
             - to: ::/0
               via: abc:abc:abc:abc::1
               table: 201
    version: 2

配置说明 #

1. 添加IPv6路由表（routes6）：为每块网卡的IPv6配置子网路由和默认路由，指定对应的路由表（eth0用200，eth1用201），网关为子网的IPv6默认网关abc:abc:abc:abc::1
2. 添加IPv6策略路由规则：在routing-policy中补充基于网卡私有IPv6地址的源/目的规则，确保从eth1发出的IPv6流量使用eth1对应的路由表，返回流量也导向eth1

应用配置步骤 #

1. 保存修改后的配置文件
2. 执行命令应用配置：

   sudo netplan apply
   

3. 验证路由规则：

   ip -6 rule show
   ip -6 route show table 200
   ip -6 route show table 201
   

4. 使用telnet -6 ipv6.telnetmyip.com测试eth1的公网IPv6是否正常

内容的提问来源于stack exchange，提问作者Edward Zhou