CentOS7中持久化静态路由69.0.0.0/8的配置定位排查问询

Alright, let's tackle this persistent route issue on CentOS 7 step by step. The fact that it survives reboots but doesn't show up in /etc searches tells us it's either managed by a service you haven't checked, or it's an auto-generated kernel route (which is super common here).

Beyond the /etc root you already scanned, these are the most likely spots:

• NetworkManager Connection Files: Most CentOS 7 systems use NetworkManager by default, and it stores per-connection configs in /etc/NetworkManager/system-connections/. Each network connection is a plaintext file (often named after your NIC or connection name) with an [ipv4] section that might define routes via routes= or route1= entries. Run grep -r "69.0.0.0" /etc/NetworkManager/ to check these.
• DHCP-Pushed Routes: If your eth0 interface uses DHCP, the DHCP server might be sending this route along with your IP address. Check the lease file at /var/lib/dhclient/dhclient-eth0.leases for any route-related lines, or look at /etc/dhcp/dhclient.conf to see if custom route requests are configured.
• Post-Up Scripts: Custom scripts that run after the network comes up can add routes. Check:
  • /etc/rc.local (make sure it's executable with chmod +x /etc/rc.local if you're using it)
  • /etc/sysconfig/network-scripts/ifup-local (a script that runs for all interfaces when they come up)
• Systemd Services: Custom systemd services might run ip route add commands on boot. Scan /etc/systemd/system/ and /usr/lib/systemd/system/ for service files that include route-related commands, or use systemctl list-unit-files --type=service to look for suspect services.

Here's how to trace where this route is coming from:

• Check for Kernel-Generated Routes: Your ip route output shows proto kernel—that's a key clue! This means the route was automatically created by the Linux kernel when your eth0 interface was assigned its IP (69.1XX.XXX.X7). If your NIC's subnet mask is set to /8 (255.0.0.0), the kernel will automatically add a route for the entire 69.0.0.0/8 network as a direct attached subnet. Verify this with ip addr show eth0—look for the inet line to confirm the subnet mask is /8.
• Inspect NetworkManager's Active Config: Use nmcli con show to list all active connections, find the one tied to eth0, then run nmcli con show <connection-name> | grep -i route to see if any static routes are defined there. You can also check the device's current state with nmcli device show eth0 | grep route.
• Track Route Changes in Real-Time: Use ip monitor route to watch for route additions/removals as you reboot or restart network services—this can help you narrow down exactly when the route is added.

• NetworkManager vs. Legacy network Service: CentOS 7 defaults to NetworkManager, but if you've ever switched to the legacy network service, configs might be split or conflicting. Check which service is running with systemctl status NetworkManager and systemctl status network.
• Static vs. DHCP Conflicts: If you switched from DHCP to static IP but didn't fully clean up DHCP settings, leftover leases or configs might still be pushing routes.
• Kernel Route Behavior: Remember that direct attached routes (marked proto kernel) are not static routes—they're auto-generated based on your NIC's IP and subnet. You won't find them in any static route config files, and modifying them directly (with ip route del) will only work until the next reboot or network restart.

内容的提问来源于stack exchange，提问作者Frank Phillips