Change Interface MTU Value on cisco Nexus3000

[bandararmc]

Hello Friends, 

I want to change MTU value to accept jumbo frames on a Nexus3000 to fix a giants error issue. Please advise the exact we way to do it. 

Below is the current config and the interface output. 

**********************************

SW-1# show int e1/22 | include MTU
MTU 1500 bytes, BW 10000000 Kbit, DLY 10 usec

SW-1# show run all 

class-map type network-qos class-default
match qos-group 0
policy-map type network-qos default-nq-policy
class type network-qos class-default
mtu 1500
system qos
service-policy type queuing input default-in-policy
service-policy type queuing output default-out-policy
service-policy type qos input default-in-policy
service-policy type network-qos default-nq-policy

**********************************

I tried to use below the configuration to change the MTU value but that caused to increase outbound packet discards on all interfaces. So I reverted it back. 

***********************************

policy-map type network-qos jumbo
class type network-qos class-default
mtu 9216
system qos
service-policy type network-qos jumbo

***************************************

Please somebody advise me on this.... 

Best Regards~~~

[M02@rt37]

Hello @bandararmc,

To apply to the switch use these commands:

policy-map type network-qos jumbo
class type network-qos class-default
mtu 9216
system qos
service-policy type network-qos jumbo

see link:

http://www.cisco.com/c/en/us/td/docs/switches/datacenter/nexus3000/sw/qos/503_u1_1/b_Cisco_nexus3000_qos_config_gd_503_u1_1/b_Cisco_nexus3000_qos_config_gd_503_u1_1_chapter_010.html

Verification of the change can be validated by running a ‘show queuing interface’ command on one of the 3k interfaces (int e1/22

#show queuing interface ethernet 1/22

If you run the ‘show interface’ command then a value of 1500 will be displayed despite the system wide change that was made.

As you confirmed that Nexus "work" in a jumbo frame "mode", ensure that all devices in the network path, including connected devices (servers, storage devices, etc.), support jumbo frames and have their MTU values properly configured.

[bandararmc]

Hi M02@rt37, 

Thank you very much for your reply. 
As I mentioned in the ticket, I tried the configuration you have proposed. But When I applied the changes on the Switch, I noticed a huge spike on "outbound packets discarded". So I reverted back the changes. 

Best Regards~~~

[MHM Cisco World]

you must config same MTU value in ingress and egress of traffic 
here you config some port and you see high drop. 
in other words either all port use jumbo or not.

[bandararmc]

hi @MHM Cisco World, 

Thank you for your reply. 

Here I changed the system config and It will affected all ports. Yes this SW is in the middle and some SWs and severs which are connected this SW don't have jumbo frame enabled. Still this SW should support normal frames as well ? 
I just want understand the exact reason for these "outbound packets discards". So that means MTU path negotiation is not happening if I enable jumbo frame in the middle SW? 

I just want to clear this please.. 

Best Regards~~~
 

[MHM Cisco World]

Firend 
either from Server to client all Path support Jumbo or you will see issue like this. 
Server send jumbo some SW forward it (jumbo enable) other drop.
the only way this work is the traffic inter-vlan so the VLAN (IP L3 layer) can fragment this jumbo. even this I am not so sure it work.
thanks 
MHM

[bandararmc]

Hi @MHM Cisco World, 

Thank you for your reply. 
But from the server side or client side, jumbo is not enable. I just want to change the MTU value for a SW in the middle for a new connection to support Jumbo. I don't want to change the existing setup for the old connections. 
So SW should be able handle both jumbo and normal frames? that is the point I am not understanding. 

Best Regards~~~