From 47cf2d10c22904f958e8a919ee770e49c9e1602a Mon Sep 17 00:00:00 2001 From: Alexandre Derumier Date: Wed, 22 Aug 2018 08:33:16 +0200 Subject: [PATCH] vxlan: increase mtu to 1550 vxlan encapsulation have 50 bytes overhead,so we need to increase mtu on physical interfaces to avoid fragmentation. --- vxlan-and-evpn.adoc | 31 ++++++++++++++++++++++++++++++- 1 file changed, 30 insertions(+), 1 deletion(-) diff --git a/vxlan-and-evpn.adoc b/vxlan-and-evpn.adoc index 32f748d..5115d78 100644 --- a/vxlan-and-evpn.adoc +++ b/vxlan-and-evpn.adoc @@ -16,6 +16,9 @@ while accommodating a very large number of tenants. It is defined in RFC 7348. Each overlay network is known as a VXLAN Segment and identified by a unique 24-bit segment ID called a VXLAN Network Identifier (VNI). +VXLAN encapsulation add 50bytes overhead, so you need to increase mtu on your host +physical interfaces to 1550 at minimum. (or decrease mtu inside your vms to 1450) + For BUM traffic (broadcast / unknown unicast traffic, multicast), we have 3 differents vxlan setup modes : multicast, unicast, bgp-evpn @@ -36,6 +39,7 @@ remote VTEPs will get the packet and answer accordingly direct to the originatin ---- auto eno1 iface eno1 inet manual + mtu 1550 auto vmbr0 iface vmbr0 inet static @@ -74,6 +78,7 @@ iface vmbr3 inet manual ---- auto eno1 iface eno1 inet manual + mtu 1550 auto vmbr0 iface vmbr0 inet static @@ -113,6 +118,7 @@ iface vmbr3 inet manual ---- auto eno1 iface eno1 inet manual + mtu 1550 auto vmbr0 iface vmbr0 inet static @@ -161,6 +167,7 @@ The VXLAN device will still learn remote addresses automatically using source-ad ---- auto eno1 iface eno1 inet manual + mtu 1550 auto vmbr0 iface vmbr0 inet static @@ -203,6 +210,7 @@ iface vmbr3 inet manual ---- auto eno1 iface eno1 inet manual + mtu 1550 auto vmbr0 iface vmbr0 inet static @@ -244,6 +252,7 @@ iface vmbr3 inet manual ---- auto eno1 iface eno1 inet manual + mtu 1550 auto vmbr0 iface vmbr0 inet static @@ -296,6 +305,7 @@ it's possible to use external bgp route reflector servers. ---- auto eno1 iface eno1 inet manual + mtu 1550 auto vmbr0 iface vmbr0 inet static @@ -363,6 +373,7 @@ line vty ---- auto eno1 iface eno1 inet manual + mtu 1550 auto vmbr0 iface vmbr0 inet static @@ -429,6 +440,7 @@ line vty ---- auto eno1 iface eno1 inet manual + mtu 1550 auto vmbr0 iface vmbr0 inet static @@ -526,6 +538,7 @@ net.ipv6.conf.all.forwarding=1 ---- auto eno1 iface eno1 inet manual + mtu 1550 auto vmbr0 iface vmbr0 inet static @@ -600,6 +613,7 @@ line vty ---- auto eno1 iface eno1 inet manual + mtu 1550 auto vmbr0 iface vmbr0 inet static @@ -674,6 +688,7 @@ line vty ---- auto eno1 iface eno1 inet manual + mtu 1550 auto vmbr0 iface vmbr0 inet static @@ -779,6 +794,7 @@ iface vrf1 auto eno1 iface eno1 inet manual + mtu 1550 auto vmbr0 iface vmbr0 inet static @@ -888,6 +904,7 @@ iface vrf1 auto eno1 iface eno1 inet manual + mtu 1550 auto vmbr0 iface vmbr0 inet static @@ -998,6 +1015,7 @@ iface vrf1 auto eno1 iface eno1 inet manual + mtu 1550 auto vmbr0 iface vmbr0 inet static @@ -1118,6 +1136,7 @@ iface vrf1 auto eno1 iface eno1 inet manual + mtu 1550 auto vmbr0 iface vmbr0 inet static @@ -1237,6 +1256,7 @@ iface vrf1 auto eno1 iface eno1 inet manual + mtu 1550 auto vmbr0 iface vmbr0 inet static @@ -1347,6 +1367,7 @@ iface vrf1 auto eno1 iface eno1 inet manual + mtu 1550 auto vmbr0 iface vmbr0 inet static @@ -1465,6 +1486,7 @@ iface vrf1 auto eno1 iface eno1 inet manual + mtu 1550 auto vmbr0 iface vmbr0 inet static @@ -1479,6 +1501,7 @@ iface eno2 address 172.16.0.1 netmask 255.255.255.0 vrf vrf1 + mtu 1550 post-up ip route add default via 172.16.0.254 dev eno2 vrf vrf1 #if you have multiple external routers, you can use ecmp balancing #post-up route add default nexthop via 172.16.0.253 dev eno2 vrf vrf1 nexthop via 172.16.0.254 dev eno2 vrf vrf1 @@ -1584,6 +1607,7 @@ iface vrf1 auto eno1 iface eno1 inet manual + mtu 1550 auto vmbr0 iface vmbr0 inet static @@ -1598,6 +1622,7 @@ iface eno2 address 172.16.0.3 netmask 255.255.255.0 vrf vrf1 + mtu 1550 post-up ip route add default via 172.16.0.254 dev eno2 vrf vrf1 #if you have multiple external routers, you can use ecmp balancing #post-up route add default nexthop via 172.16.0.253 dev eno2 vrf vrf1 nexthop via 172.16.0.254 dev eno2 vrf vrf1 @@ -1704,7 +1729,8 @@ iface vrf1 auto eno1 iface eno1 inet manual - + mtu 1550 + auto vmbr0 iface vmbr0 inet static address 192.168.0.3 @@ -1718,6 +1744,7 @@ iface eno2 address 172.16.0.3 netmask 255.255.255.0 vrf vrf1 + mtu 1550 post-up ip route add default via 172.16.0.254 dev eno2 vrf vrf1 #if you have multiple external routers, you can use ecmp balancing #post-up route add default nexthop via 172.16.0.253 dev eno2 vrf vrf1 nexthop via 172.16.0.254 dev eno2 vrf vrf1 @@ -1832,6 +1859,7 @@ iface eno2 address 172.16.0.1 netmask 255.255.255.0 vrf vrf1 + mtu 1550 post-up ip route add default via 172.16.0.254 dev eno2 vrf vrf1 vrrp-id 1 vrrp-priority 1 @@ -1845,6 +1873,7 @@ auto eno2 iface eno2 address 172.16.0.2 netmask 255.255.255.0 + mtu 1550 vrf vrf1 post-up ip route add default via 172.16.0.254 dev eno2 vrf vrf1 vrrp-id 1 -- 2.39.2