我有几个 c5.large 实例，每个实例有两个网络接口，运行截至 2018-03 的最新 Amazon Linux 2 AMI。它们以前每个接口有 8 个 IP 地址，现在更改为 7 个。添加 IP 地址和接口时，在重新启动服务/重新启动时会正确更新，如 Amazon 文档中所述。现在，其中一个实例在主网络接口上仍然认为它有 8 个 IP 地址 - 已删除的 IP 地址仍然存在。
我尝试过 systemctl restart network 并重新启动，但无济于事。
以下是 journalctl 输出的相关部分。curlinghttp://xxxx/latest/meta-data/network/interfaces/macs/MAC1/local-ipv4s（具有 eth0 的相关 MAC）返回正确的 7 个 IP 地址，没有过时的 IP 地址，该地址仍然停留在网络接口上。我注意到我们只看到 eth1 的“重写规则”，而不是 eth0。不过，一个实例没问题，而另一个实例则有问题。
network[1829]: Bringing up interface eth0:
dhclient[1982]: DHCPREQUEST on eth0 to x.x.x.x port 67 (xid=0x509915e5)
dhclient[1982]: DHCPACK from x.x.x.x (xid=0x509915e5)
NET[2026]: /usr/sbin/dhclient-script : updated /etc/resolv.conf
dhclient[1982]: bound to x.x.x.x -- renewal in 1364 seconds.
ec2net[2043]: [get_meta] Trying to get http://x.x.x.x/latest/meta-data/network/interfaces/macs/MAC1/local-ipv4s
network[1829]: Determining IP information for eth0... done.
ec2net[2061]: [rewrite_aliases] Rewriting aliases of eth0
systemd[1]: Started AWS CodeDeploy Host Agent.
bash[1826]: /opt/codedeploy-agent/vendor/gems/logging-1.8.2/lib/logging/appender.rb:139: warning: constant ::Fixnum is deprecated
bash[1826]: /opt/codedeploy-agent/vendor/gems/logging-1.8.2/lib/logging/logger.rb:295: warning: constant ::Fixnum is deprecated
network[1829]: Determining IPv6 information for eth0... done.
network[1829]: [ OK ]
network[1829]: Bringing up interface eth1:
dhclient[2217]: DHCPREQUEST on eth1 to x.x.x.x port 67 (xid=0xeeb9077)
dhclient[2217]: DHCPACK from x.x.x.x (xid=0xeeb9077)
ec2net[2262]: [get_meta] Trying to get http://x.x.x.x/latest/meta-data/network/interfaces/macs/MAC2/local-ipv4s
ec2net[2271]: [rewrite_rules] Rewriting rules for eth1
dhclient[2217]: bound to x.x.x.x -- renewal in 1727 seconds.
network[1829]: Determining IP information for eth1... done.
ec2net[2293]: [get_meta] Trying to get http://x.x.x.x/latest/meta-data/network/interfaces/macs/MAC2/local-ipv4s
ec2net[2300]: [rewrite_aliases] Rewriting aliases of eth1
dhclient[2166]: XMT: Solicit on eth0, interval 1060ms.
network[1829]: Determining IPv6 information for eth1... done.
network[1829]: [ OK ]
systemd[1]: Started LSB: Bring up/down networking.
systemd[1]: Starting Initial cloud-init job (metadata service crawler)...
systemd[1]: Reached target Network.
systemd[1]: Starting Network.
systemd[1]: Starting Postfix Mail Transport Agent...
dhclient[2166]: XMT: Solicit on eth0, interval 2050ms.
cloud-init[2477]: Cloud-init v. 0.7.9 running 'init' at Tue, 13 Feb 2018 13:28:18 +0000. Up 9.23 seconds.
dhclient[2431]: XMT: Solicit on eth1, interval 1040ms.
cloud-init[2477]: ci-info: +++++++++++++++++++++++++++Net device info++++++++++++++++++++++++++++
编辑：我期望接口在网络重启时（或至少在实例重启时）从主网络接口中删除未分配的 IP 地址。我该如何实现这一点？
编辑 2：通过手动从接口中删除 IPip addr del x.x.x.x/20 dev eth0暂时将其删除，但当我运行时它会回来service network restart。也许有一些缓存的结果坚持下来了……？