某局点无线组网替换网关之后AP掉线问题处理经验案例

传统的AC、AP注册环境

AP的地址通过交换机DHCP server分配；

现网业务因为交换机原因，需要替换为华为交换机，类似的网关IP地址保持不变。

替换前AP的DHCP地址获取租约为24小时，按照协议机制12小时为1/2 DHCP租约会进行续约操作。

在替换之后12小时发现AP陆续与AC脱管。紧急排查发现华为交换机对应的Interface Vlan接口没有配置dhcp select globle命令，导致该接口无法提供DHCP功能。增加该命令之后AP陆续上线。

但是即使华为的交换机配置错误，但是不影响AP的地址在1/2租约时间之后被释放。

因此怀疑是否AP的处理逻辑存在问题，或者网络中有其他的报文发送导致DHCP地址被释放造成AP掉线。

因此实验室模拟验证测试：

设备：WX5540E AP WA2620X 

测试目的：AP 通过AC dhcp获取ip地址之后 在租约过了1/2之后IP地址是否依旧存在，及什么情况下IP地址会释放 

步骤1： FIT AP 零配置上线，AC 开启dhcp server ，租约时长10分钟，当AP获取IP地址之后，关闭AC的dhcp功能 AP在dhcp client verbose显示两个续约时间段T1和T2。

<WA2620X>dis dhcp client  verbose

Vlan-interface1 DHCP client information:

Current machine state: BOUND

Allocated IP: 100.0.0.6 255.255.255.0

Allocated lease: 600 seconds, T1: 300 seconds, T2: 525 seconds

Lease from 2000.04.26 12:00:28   to   2000.04.26 12:10:28

DHCP server: 100.0.0.1

Transaction ID: 0x55a55c93

Default router: 100.0.0.1

Client ID: 0023-89e2-ed80

T1 will timeout in 0 day 0 hour 1 minute 55 seconds.

当T1超时之后，AP获取的IP地址没有释放依旧存在，ping 网关AC设备 操作正常。

步骤2： FIT AP 零配置上线，AC 开启dhcp server ，租约时长10分钟，当AP获取IP地址之后，DHCP server继续工作，

配置dhcp server forbidden-ip 100.0.0.6 模拟DHCP NAK的环境。

<WA2620X>dis dhcp client  verbose

Vlan-interface1 DHCP client information:

Current machine state: BOUND

Allocated IP: 100.0.0.6 255.255.255.0

Allocated lease: 600 seconds, T1: 300 seconds, T2: 525 seconds

Lease from 2000.04.26 12:00:28   to   2000.04.26 12:10:28

DHCP server: 100.0.0.1

Transaction ID: 0x55a55c93

Default router: 100.0.0.1

Client ID: 0023-89e2-ed80

T1 will timeout in 0 day 0 hour 1 minute 12 seconds.

等T1时间超时之后：

*Apr 26 12:05:28:543 2000 WA2620X DHCPC/7/DHCPC_EVENT:

  Vlan-interface1: T1 timer expired. Begin to renew.

*Apr 26 12:05:28:543 2000 WA2620X DHCPC/7/DHCPC_PACKET:

  Vlan-interface1: Send a DHCP packet...

  Head: op (BOOTPREQUEST); htype (ETHERNET); hlen (6); xid (0x55a55c93);

  ciaddr (100.0.0.6); yiaddr (0.0.0.0); chaddr (0023-89e2-ed80);

  Options:

    63 82 53 63 35 01 03 0C 19 57 41 32 36 32 30 58

    2D 30 30 2D 32 33 2D 38 39 2D 65 32 2D 65 64 2D

    38 30 37 08 01 79 03 06 0F 21 2B 8A 39 02 04 80

    3C 10 48 33 43 2E 20 48 33 43 20 57 41 32 36 32

    30 58 3D 07 01 00 23 89 E2 ED 80 FF

*Apr 26 12:05:28:543 2000 WA2620X DHCPC/7/DHCPC_EVENT:

  Vlan-interface1: Sending DHCPREQUEST packet succeeded.    //发送dhcp request报文

*Apr 26 12:05:28:544 2000 WA2620X DHCPC/7/DHCPC_EVENT:

  Vlan-interface1: FSM state transfers (BOUND-->RENEWING) successfully.

*Apr 26 12:05:28:545 2000 WA2620X DHCPC/7/DHCPC_EVENT:

  Vlan-interface1: Receive a packet.

*Apr 26 12:05:28:545 2000 WA2620X DHCPC/7/DHCPC_PACKET:

  Vlan-interface1: Receive a DHCP packet...

  Head: op (BOOTPREPLY); htype (ETHERNET); hlen (6); xid (0x55a55c93);

  ciaddr (0.0.0.0); yiaddr (0.0.0.0); chaddr (0023-89e2-ed80);

  Option: type (DHCPNAK);        //收到服务器的NAK报文

  server (100.0.0.1);

*Apr 26 12:05:28:545 2000 WA2620X DHCPC/7/DHCPC_EVENT:

  Vlan-interface1: Move to INIT state.

*Apr 26 12:05:28:545 2000 WA2620X DHCPC/7/DHCPC_EVENT:

  Vlan-interface1: Notify route module to delete the default gateway: destination (0.0.0.0), mask (0.0.0.0), nexthop (100.0.0.1).

*Apr 26 12:05:28:545 2000 WA2620X DHCPC/7/DHCPC_EVENT:

  Vlan-interface1: FSM state transfers (RENEWING-->INIT) successfully.

*Apr 26 12:05:28:545 2000 WA2620X DHCPC/7/DHCPC_EVENT:

  Vlan-interface1: DHCPDISCOVER will be sent in 10 ms.   //重新发送dhcp discover报文，寻求新的IP地址

*Apr 26 12:05:28:553 2000 WA2620X DHCPC/7/DHCPC_PACKET:

  Vlan-interface1: Send a DHCP packet...

  Head: op (BOOTPREQUEST); htype (ETHERNET); hlen (6); xid (0x376a33d3);

  ciaddr (0.0.0.0); yiaddr (0.0.0.0); chaddr (0023-89e2-ed80);

  Options:

    63 82 53 63 35 01 01 0C 19 57 41 32 36 32 30 58

    2D 30 30 2D 32 33 2D 38 39 2D 65 32 2D 65 64 2D

    38 30 37 08 01 79 03 06 0F 21 2B 8A 39 02 04 80

    3C 10 48 33 43 2E 20 48 33 43 20 57 41 32 36 32

    30 58 3D 07 01 00 23 89 E2 ED 80 FF

*Apr 26 12:05:28:814 2000 WA2620X DHCPC/7/DHCPC_EVENT:

  Vlan-interface1: Sending DHCPDISCOVER packet succeeded.

*Apr 26 12:05:28:935 2000 WA2620X DHCPC/7/DHCPC_EVENT:

  Vlan-interface1: FSM state transfers (INIT-->SELECTING) successfully.

如果在接下来的环境依旧被服务器NAK了，AP将一直无法获取到IP地址；

结合现场的故障现象，如果AP的租约在1/2的时间段内就被释放了，很有可能在1/2发送的dhcp request报文被服务器给NAK了。   

通过上述验证，证明在现网环境租约超过1/2租约之后AP不会主动释放地址，而是发送了Dhcp Request报文，被收到了NAK导致释放。

结合现场环境新开VLAN做验证环境可以证明该结论。