Cisco Nexus 3548 MTU issue

[lmqtechnology]

I have two cisco Nexus3500 C3548P-10GX Chassis connected to one another via L3 port-channel. MTU on the system and interface are both set to 9000, however I am unable to ping over link more that 1478 bytes?

[MHM Cisco World]

there are two MTU 
the MTU of interface and MTU of VLAN SVI, 
so check each port member of PO MTU 
and check the MTU of SVI 

[lmqtechnology]

Thanks for your reply but Its a L3 port-channel not L2 with VLAN.

[MHM Cisco World]

can you share the show interface of one port member of L3 PO?

[lmqtechnology]

absolutely!

interface port-channel2
no switchport
mtu 9000
ip address 10.51.222.89/30
ip ospf network point-to-point
ip router ospf 1 area 0.0.0.51
ip pim sparse-mode

[MHM Cisco World]

port member not the PO, 
i meaning for example g0/0 and g0/1 is bound to form PO-2 
PO-2 have MTU 9000 but are g0/0 and g0/1 have same ?

[lmqtechnology]

its the same (MTU 9000), as once in the Po all config from Po.

[MHM Cisco World]

As I know all L3 port (include the L3 PO) for all NSK platform support per-port MTU and you do right 
https://www.cisco.com/c/en/us/support/docs/switches/nexus-9000-series-switches/118994-config-nexus-00.html

BUT 
still one thing 
you config MTU 9000 not 9216 
so can you try ping 8700 not 9000 ?

[lmqtechnology]

anything above 1478 does not ping.

[MHM Cisco World]

Switch#show system internal ethpm info interface eth <PO port member> | egrep -i mtu

can you share this  

[lmqtechnology]

show system internal ethpm info interface Po2 | egrep -i mtu
medium(broadcast), snmp trap(on), MTU(9000),
admin_state(1), admin_layer(1), admin_router_mac(0) admin_mtu(1)
ethpm_ im_get_ifindex_mtu_list 9000
ethpm_ im_get_ifindex_mtu_usr_conf_flag_list 1
ethpm_ im_get_ifindex_vl_mtu_list No function pointer
ethpm_ im_get_ifindex_system_def_class_mtu_list No function pointer

[MHM Cisco World]

show system internal ethpm info interface <port member not PO> | egrep -i mtu

[lmqtechnology]

show system internal ethpm info interface eth1/23 | egrep -i mtu
medium(broadcast), snmp trap(on), MTU(9000),
admin_state(1), admin_layer(1), admin_router_mac(0) admin_mtu(1)
ethpm_ im_get_ifindex_mtu_list 9000
ethpm_ im_get_ifindex_mtu_usr_conf_flag_list 1
ethpm_ im_get_ifindex_vl_mtu_list No function pointer
ethpm_ im_get_ifindex_system_def_class_mtu_list No function pointer

[MHM Cisco World]

https://www.cisco.com/c/en/us/support/docs/lan-switching/ethernet/215451-padding-and-different-mtu-behavior.html

all your config is right and PO and port member show 9000 as jumbo frame size, 
I think may be this is not issue of NSK but issue of other SW connect both side (ping source and ping destination) 
if L2 SW interconnect both NSK then even if NSK support 9000 the L2 SW with 1500 will drop the frame. 
check above link

[lmqtechnology]

issue with config, MTU had to be set in QoS policy