Jumbo frames bug?

M_Devil

Hi,

using pfSense 2.2 in a VM (ESXi 5.5) with a couple of virtual NIC's (VMXNET3).
Each NIC is assigned to a interface. For each interface I set the MTU to 9000.
After assignment and even reboot it was not possible to send jumbo frames (for example a ping with size 8000) from one interface to another. Of course a normal ping did work.

I found out that the route still has a MTU setting of 1500 (looking at the Routes option under Diagnostic). When I manualy update the route(CLI) with MTU=9000 the jumbo ping works.

Is this a bug in pfSense 2.2 or freeBSD 10?

cmb

what's the route in question, a static route or your default gateway? v4 or v6?

There were issues along those lines in FreeBSD 8.x which were easily replicable, though I went through and tested a variety of scenarios in 10.1 and they all were fine there. Appears there's some circumstance along those lines that's still an issue that I missed.

M_Devil

Below the partial output of the 'diagnostics -> routes pages'

192.168.10.0/24 link#8 US 78751014 9000 vmx4
192.168.10.2 link#8 UHS 0 16384 lo0
192.168.20.0/24 link#9 US 0 9000 vmx5
192.168.20.2 link#9 UHS 0 16384 lo0
192.168.30.0/24 link#7 US 115225 9000 vmx3
192.168.30.2 link#7 UHS 0 16384 lo0
192.168.40.0/24 link#6 US 302047 9000 vmx2
192.168.40.2 link#6 UHS 0 16384 lo0

As you can see the .0/24 routes are now MTU = 9000. Before script they where MTU = 1500.

The corresponding interfaces are all MTU = 9000. I can see that on the 'Status -> Interfaces page'.

In this case it is IPv4, don't know about IPv6 situation.

Does this answer your questions?

If you like, I can perform more testing.

cmb

Ah, I see what you mean, on the link routes. It's not always including the MTU in ifconfig IP assignments where necessary.
https://redmine.pfsense.org/issues/4397

TECHMAN2000

I am having the same problem, when I setting everything in bridge mode and set the jumbo size on the nic cards. After rebooting the system its show jumbo size on the every nic card but when I do the ifconfig command on shell, to confirm the setting has saved its show on every device MTU 1500 and when I do the benchmark its showing 1Gbes per transfers and these on 10Gb network card.

Its must be a bug...
On the older version of Pfsense never had a problem.