目录
简介
虚拟可扩展局域网(VXLAN)作为数据中心互联(DCI)解决方案越来越流行。VXLAN功能用于在第3层/公共路由域上提供第2层扩展。本文档讨论Cisco IOS XE设备的基本配置和故障排除。
本文档的配置和验证部分涵盖两种场景：
- 场景A描述在组播模式下三个数据中心之间的VXLAN配置。
- 场景B描述单播模式下两个数据中心之间的VXLAN配置。
先决条件
要求
Cisco 建议您了解以下主题：
- 基本了解DCI重叠和组播
使用的组件
本文档中的信息基于以下软件和硬件版本：
- 运行软件03.16.00.S的ASR1004
- 运行软件3.16.03.S的CSR100v(VXE)
本文档中的信息都是基于特定实验室环境中的设备编写的。本文档中使用的所有设备最初均采用原始（默认）配置。如果您的网络处于活动状态，请确保您了解所有命令的潜在影响。
配置
场景 A：在组播模式下配置三个数据中心之间的VXLAN
基本配置
组播模式需要站点之间的单播和组播连接。本配置指南使用开放最短路径优先(OSPF)提供单播连接，使用双向协议独立组播(PIM)提供组播连接。
以下是所有三个数据中心的组播操作模式基本配置：
! DC1#show run | sec ospf router ospf 1 network 1.1.1.1 0.0.0.0 area 0 network 10.10.10.4 0.0.0.3 area 0
!
PIM双向配置：
!
DC1#show run | sec pim ip pim bidir-enable ip pim send-rp-discovery scope 10 ip pim bsr-candidate Loopback1 0 ip pim rp-candidate Loopback1 group-list 10 bidir ! access-list 10 permit 239.0.0.0 0.0.0.255
!
DC1#
!
此外，PIM稀疏模式在所有L3接口下启用，包括环回：
!
DC1#show run interface lo1 Building configuration... Current configuration : 83 bytes ! interface Loopback1 ip address 1.1.1.1 255.255.255.255 ip pim sparse-mode end
另请确保在设备上启用组播路由，并且您看到组播路由表已填充。
网络图
DC1(VTEP1)配置
! ! Vxlan udp port 1024 ! Interface Loopback1 ip address 1.1.1.1 255.255.255.255 ip pim sparse-mode !
在网桥域配置下定义VNI成员和成员接口：
! bridge-domain 1 member vni 6001 member FastEthernet0/1/7 service-instance 1 !
创建网络虚拟接口(NVE)并定义需要通过广域网扩展到其他数据中心的VNI成员：
! interface nve1 no ip address shut member vni 6001 mcast-group 239.0.0.10 ! source-interface Loopback1 !
通过LAN接口（即连接LAN网络的接口）创建服务实例，以覆盖特定VLAN（802.1q标记流量） — 在本例中为VLAN 1:
! interface FastEthernet0/1/7 no ip address negotiation auto cdp enable no shut !
在通过重叠发送流量之前删除VLAN标记，并在返回流量发送到VLAN后推送该标记：
! service instance 1 ethernet encapsulation unagged !
DC2(VTEP2)配置
! ! Vxlan udp port 1024 ! interface Loopback1 ip address 2.2.2.2 255.255.255.255 ip pim sparse-mode ! ! bridge-domain 1 member vni 6001 member FastEthernet0/1/3 service-instance 1 ! ! interface nve1 no ip address member vni 6001 mcast-group 239.0.0.10 ! source-interface Loopback1 shut ! ! interface FastEthernet0/1/3 no ip address negotiation auto cdp enable no shut ! service instance 1 ethernet encapsulation untagged !
DC3(VTEP3)配置
! ! Vxlan udp port 1024 ! interface Loopback1 ip address 3.3.3.3 255.255.255.255 ip pim sparse-mode ! ! bridge-domain 1 member vni 6001 member GigabitEthernet2 service-instance 1 ! interface nve1 no ip address shut member vni 6001 mcast-group 239.0.0.10 ! source-interface Loopback1 ! interface gig2 no ip address negotiation auto cdp enable no shut ! service instance 1 ethernet encapsulation untagged !
场景 B:在单播模式下配置两个数据中心之间的VXLAN
网络图
DC1配置
! interface nve1 no ip address member vni 6001 ! ingress replication shold be configured as peer data centers loopback IP address. ! ingress-replication 2.2.2.2 ! source-interface Loopback1 ! ! interface gig0/2/1 no ip address negotiation auto cdp enable ! service instance 1 ethernet encapsulation untagged ! ! ! bridge-domain 1 member vni 6001 member gig0/2/1 service-instance 1
DC2配置
! interface nve1 no ip address member vni 6001 ingress-replication 1.1.1.1 ! source-interface Loopback1 ! ! interface gig5 no ip address negotiation auto cdp enable ! service instance 1 ethernet encapsulation untagged ! ! bridge-domain 1 member vni 6001 member gig5 service-instance 1
验证
场景 A：在组播模式下配置三个数据中心之间的VXLAN
在您完成场景A的配置后，每个数据中心中连接的主机应该能够在同一广播域内相互访问。
使用这些命令检验配置。场景B下提供了一些示例。
Router#show nve vni Router#show nve vni interface nve1 Router#show nve interface nve1 Router#show nve interface nve1 detail Router#show nve peers
场景 B:在单播模式下配置两个数据中心之间的VXLAN
在DC1上：
DC1#show nve vni Interface VNI Multicast-group VNI state nve1 6001 N/A Up DC1#show nve interface nve1 detail Interface: nve1, State: Admin Up, Oper Up Encapsulation: Vxlan source-interface: Loopback1 (primary:1.1.1.1 vrf:0) Pkts In Bytes In Pkts Out Bytes Out 60129 6593586 55067 5303698 DC1#show nve peers Interface Peer-IP VNI Peer state nve1 2.2.2.2 6000 -
在DC2上：
DC2#show nve vni
Interface VNI Multicast-group VNI state
nve1 6000 N/A Up
DC2#show nve interface nve1 detail
Interface: nve1, State: Admin Up, Oper Up Encapsulation: Vxlan
source-interface: Loopback1 (primary:2.2.2.2 vrf:0)
Pkts In Bytes In Pkts Out Bytes Out
70408 7921636 44840 3950835
DC2#show nve peers
Interface Peer-IP VNI Peer state
nve 1 1.1.1.1 6000 Up
DC2#show bridge-domain 1
Bridge-domain 1 (3 ports in all)
State: UP Mac learning: Enabled
Aging-Timer: 300 second(s)
BDI1 (up)
GigabitEthernet0/2/1 service instance 1
vni 6001
AED MAC address Policy Tag Age Pseudoport
0 7CAD.74FF.2F66 forward dynamic 281 nve1.VNI6001, VxLAN src: 1.1.1.1 dst: 2.2.2.2
0 B838.6130.DA80 forward dynamic 288 nve1.VNI6001, VxLAN src: 1.1.1.1 dst: 2.2.2.2
0 0050.56AD.1AD8 forward dynamic 157 nve1.VNI6001, VxLAN src: 1.1.1.1 dst: 2.2.2.2
故障排除
“验证”部分描述的命令提供基本的故障排除步骤。当系统不工作时，这些附加诊断可能会有所帮助。
注意：其中一些诊断可能导致内存和CPU利用率增加。
调试诊断
#debug nve error
*Jan 4 20:00:54.993: NVE-MGR-PEER ERROR: Intf state force down successful for mcast nodes cast nodes
*Jan 4 20:00:54.993: NVE-MGR-PEER ERROR: Intf state force down successful for mcast nodes cast nodes
*Jan 4 20:00:54.995: NVE-MGR-PEER ERROR: Intf state force down successful for peer nodes eer nodes
*Jan 4 20:00:54.995: NVE-MGR-PEER ERROR: Intf state force down successful for peer nodes
#show nve log error
[01/01/70 00:04:34.130 UTC 1 3] NVE-MGR-STATE ERROR: vni 6001: error in create notification to Tunnel
[01/01/70 00:04:34.314 UTC 2 3] NVE-MGR-PEER ERROR: Intf state force up successful for mcast nodes
[01/01/70 00:04:34.326 UTC 3 3] NVE-MGR-PEER ERROR: Intf state force up successful for peer nodes
[01/01/70 01:50:59.650 UTC 4 3] NVE-MGR-PEER ERROR: Intf state force down successful for mcast nodes
[01/01/70 01:50:59.654 UTC 5 3] NVE-MGR-PEER ERROR: Intf state force down successful for peer nodes
[01/01/70 01:50:59.701 UTC 6 3] NVE-MGR-PEER ERROR: Intf state force up successful for mcast nodes
[01/01/70 01:50:59.705 UTC 7 3] NVE-MGR-PEER ERROR: Intf state force up successful for peer nodes
[01/01/70 01:54:55.166 UTC 8 61] NVE-MGR-PEER ERROR: Intf state force down successful for mcast nodes
[01/01/70 01:54:55.168 UTC 9 61] NVE-MGR-PEER ERROR: Intf state force down successful for peer nodes
[01/01/70 01:55:04.432 UTC A 3] NVE-MGR-PEER ERROR: Intf state force up successful for mcast nodes
[01/01/70 01:55:04.434 UTC B 3] NVE-MGR-PEER ERROR: Intf state force up successful for peer nodes
[01/01/70 01:55:37.670 UTC C 61] NVE-MGR-PEER ERROR: Intf state force down successful for mcast nodes
#show nve log event
[01/04/70 19:48:51.883 UTC 1DD16 68] NVE-MGR-DB: Return vni 6001 for pi_hdl[0x437C9B68]
[01/04/70 19:48:51.884 UTC 1DD17 68] NVE-MGR-DB: Return pd_hdl[0x1020010] for pi_hdl[0x437C9B68]
[01/04/70 19:48:51.884 UTC 1DD18 68] NVE-MGR-DB: Return vni 6001 for pi_hdl[0x437C9B68]
[01/04/70 19:49:01.884 UTC 1DD19 68] NVE-MGR-DB: Return pd_hdl[0x1020010] for pi_hdl[0x437C9B68]
[01/04/70 19:49:01.884 UTC 1DD1A 68] NVE-MGR-DB: Return vni 6001 for pi_hdl[0x437C9B68]
[01/04/70 19:49:01.885 UTC 1DD1B 68] NVE-MGR-DB: Return pd_hdl[0x1020010] for pi_hdl[0x437C9B68]
[01/04/70 19:49:01.885 UTC 1DD1C 68] NVE-MGR-DB: Return vni 6001 for pi_hdl[0x437C9B68]
[01/04/70 19:49:11.886 UTC 1DD1D 68] NVE-MGR-DB: Return pd_hdl[0x1020010] for pi_hdl[0x437C9B68]
[01/04/70 19:49:11.886 UTC 1DD1E 68] NVE-MGR-DB: Return vni 6001 for pi_hdl[0x437C9B68]
[01/04/70 19:49:11.887 UTC 1DD1F 68] NVE-MGR-DB: Return pd_hdl[0x1020010] for pi_hdl[0x437C9B68]
[01/04/70 19:49:11.887 UTC 1DD20 68] NVE-MGR-DB: Return vni 6001 for pi_hdl[0x437C9B68]
[01/04/70 19:49:21.884 UTC 1DD21 68] NVE-MGR-DB: Return pd_hdl[0x1020010] for pi_hdl[0x437C9B68]
嵌入式数据包捕获
Cisco IOS XE软件中提供的嵌入式数据包捕获(EPC)功能可为故障排除提供其他信息。
例如，此捕获解释了VXLAN封装的数据包：
EPC配置（TEST_ACL是用于过滤捕获数据的访问列表）：
#monitor capture TEST access-list TEST_ACL interface gigabitEthernet0/2/0 both #monitor capture TEST buffer size 10 #monitor capture TEST start
以下是结果的数据包转储：
# show monitor capture TEST buffer dump # monitor capture TEST export bootflash:TEST.pcap // with this command you can export the capture in pcap format to the bootflash, which can be downloaded and opened in wireshark.
以下是一个示例，说明简单Internet控制消息协议(ICMP)如何通过VXLAN工作。
通过VXLAN重叠发送的地址解析协议(ARP):
ARP响应：
ICMP请求：
ICMP响应：
其他调试和故障排除命令
本节还介绍一些debug和troubleshooting命令。
在本例中，调试的突出显示部分显示NVE接口无法加入组播组。因此，VXLAN封装未为VNI 6002启用。这些调试结果指向网络上的组播问题。
#debug nve all
*Jan 5 06:13:55.844: NVE-MGR-DB: creating mcast node for 239.0.0.10
*Jan 5 06:13:55.846: NVE-MGR-MCAST: IGMP add for (0.0.0.0,239.0.0.10) was failure
*Jan 5 06:13:55.846: NVE-MGR-DB ERROR: Unable to join mcast core tree
*Jan 5 06:13:55.846: NVE-MGR-DB ERROR: Unable to join mcast core tree
*Jan 5 06:13:55.846: NVE-MGR-STATE ERROR: vni 6002: error in create notification to mcast
*Jan 5 06:13:55.846: NVE-MGR-STATE ERROR: vni 6002: error in create notification to mcast
*Jan 5 06:13:55.849: NVE-MGR-TUNNEL: Tunnel Endpoint 239.0.0.10 added
*Jan 5 06:13:55.849: NVE-MGR-TUNNEL: Endpoint 239.0.0.10 added
*Jan 5 06:13:55.851: NVE-MGR-EI: Notifying BD engine of VNI 6002 create
*Jan 5 06:13:55.857: NVE-MGR-DB: Return vni 6002 for pi_hdl[0x437C9B28]
*Jan 5 06:13:55.857: NVE-MGR-EI: VNI 6002: BD state changed to up, vni state to Down
以下是VNI加入组播组后将发送的互联网组管理协议(IGMP)成员报告：
本示例显示在为组播模式在NVE下配置VNI后，如果组播按预期运行，预期的调试结果：
*Jan 5 06:19:20.335: NVE-MGR-DB: [IF 0x14]VNI node creation
*Jan 5 06:19:20.335: NVE-MGR-DB: VNI Node created [437C9B28]
*Jan 5 06:19:20.336: NVE-MGR-PD: VNI 6002 create notification to PD
*Jan 5 06:19:20.336: NVE-MGR-PD: VNI 6002 Create notif successful, map [pd 0x1020017] to [pi 0x437C9B28]
*Jan 5 06:19:20.336: NVE-MGR-DB: creating mcast node for 239.0.0.10
*Jan 5 06:19:20.342: NVE-MGR-MCAST: IGMP add for (0.0.0.0,239.0.0.10) was successful
*Jan 5 06:19:20.345: NVE-MGR-TUNNEL: Tunnel Endpoint 239.0.0.10 added
*Jan 5 06:19:20.345: NVE-MGR-TUNNEL: Endpoint 239.0.0.10 added
*Jan 5 06:19:20.347: NVE-MGR-EI: Notifying BD engine of VNI 6002 create
*Jan 5 06:19:20.347: NVE-MGR-DB: Return pd_hdl[0x1020017] for pi_hdl[0x437C9B28]
*Jan 5 06:19:20.347: NVE-MGR-DB: Return vni 6002 for pi_hdl[0x437C9B28]
*Jan 5 06:19:20.349: NVE-MGR-DB: Return vni state Create for pi_hdl[0x437C9B28]
*Jan 5 06:19:20.349: NVE-MGR-DB: Return vni state Create for pi_hdl[0x437C9B28]
*Jan 5 06:19:20.349: NVE-MGR-DB: Return vni 6002 for pi_hdl[0x437C9B28]
*Jan 5 06:19:20.351: NVE-MGR-EI: L2FIB query for info 0x437C9B28
*Jan 5 06:19:20.351: NVE-MGR-EI: PP up notification for bd_id 3
*Jan 5 06:19:20.351: NVE-MGR-DB: Return vni 6002 for pi_hdl[0x437C9B28]
*Jan 5 06:19:20.352: NVE-MGR-STATE: vni 6002: Notify clients of state change Create to Up
*Jan 5 06:19:20.352: NVE-MGR-DB: Return vni 6002 for pi_hdl[0x437C9B28]
*Jan 5 06:19:20.353: NVE-MGR-PD: VNI 6002 Create to Up State update to PD successful
*Jan 5 06:19:20.353: NVE-MGR-EI: VNI 6002: BD state changed to up, vni state to Up
*Jan 5 06:19:20.353: NVE-MGR-STATE: vni 6002: No state change Up
*Jan 5 06:19:20.353: NVE-MGR-STATE: vni 6002: New State as a result of create Up