# NDP：eth0上fe80::1的重复地址检测失败，地址已被00:1a:2b:3c:4d:5e使用

- **ID:** `networking/ipv6-ndp-duplicate-address`
- **领域:** networking
- **类别:** layer2
- **验证级别:** ai_generated
- **修复率:** 79%

## 根因

IPv6重复地址检测（DAD）检测到链路本地地址fe80::1已被同一链路上的另一主机声明，阻止接口使用该地址。

## 版本兼容性

| 版本 | 状态 | 引入 | 弃用 |
|------|------|------|------|
| Linux kernel 6.2.0 | active | — | — |
| FreeBSD 13.2 | active | — | — |
| Windows 11 IPv6 stack | active | — | — |

## 解决方案

1. ```
   Change the link-local address on one host by modifying the interface identifier: `ip link set dev eth0 address 02:00:00:00:00:01` (EUI-64 based), then reboot or restart networking.
   ```
2. ```
   Use a static link-local address with a unique suffix on each host: `ip -6 addr add fe80::1/64 dev eth0` but ensure no conflict by checking with `ndp -a`.
   ```
3. ```
   Reconfigure the network to use globally routable IPv6 addresses instead of link-local for communication, and disable DAD for link-local only: `sysctl -w net.ipv6.conf.eth0.dad_transmits=0`.
   ```

## 无效尝试

- **** — Disabling DAD leads to network instability and packet loss as two hosts respond to the same address. (95% 失败率)
- **** — Both hosts will conflict; only one can respond to neighbor solicitations, causing intermittent failures. (90% 失败率)
- **** — Flushing caches does not resolve the underlying address duplication; the conflict persists. (80% 失败率)