I'm testing jumbo frames between two Proxmox nodes. Weirdly, using ping -M do -s 8972, it still works when I don't have jumbo enabled on the interface, or the switch's VLAN.
I'm using a VLAN interface with no MTU set, running on a 10 Gb/s interface that has the MTU set to 9000
The servers are connected to an HP 5412zl, and jumbo is off on the VLAN on the switch.
Does the VLAN interface inherit the underlying physical NICs MTU?
Can a reasonable switch such as a HP 5412zl buffer jumbo frames despite jumbo being off on the VLAN I'm using?
I'm using a VLAN interface with no MTU set, running on a 10 Gb/s interface that has the MTU set to 9000
The servers are connected to an HP 5412zl, and jumbo is off on the VLAN on the switch.
Does the VLAN interface inherit the underlying physical NICs MTU?
Can a reasonable switch such as a HP 5412zl buffer jumbo frames despite jumbo being off on the VLAN I'm using?
Code:
auto eno12409np1
iface eno12409np1 inet manual
mtu 9000
...
auto eno12409np1.107
iface eno12409np1.107 inet static
address 10.20.7.141
netmask 255.255.255.0
Code:
root@hn-node-1:~# ping -I eno12409np1.107 -M do -s 8972 10.20.7.142
PING 10.20.7.142 (10.20.7.142) from 10.20.7.141 eno12409np1.107: 8972(9000) bytes of data.
8980 bytes from 10.20.7.142: icmp_seq=1 ttl=64 time=0.116 ms
8980 bytes from 10.20.7.142: icmp_seq=2 ttl=64 time=0.102 ms
8980 bytes from 10.20.7.142: icmp_seq=3 ttl=64 time=0.075 ms
...