Header Only - DO NOT REMOVE - Extreme Networks

Jumbo frames configured though dell devices will not respond to a mtu size set for more than 8976

  • 1 September 2017
  • 2 replies

Userlevel 3
Jumbo fames enabled for L2 and L3.

If I ping between two extreme switches with an MTU of 9000 , I get an appropriate response, basically it works as expected.
ping vr "VR-Default" start-size 9000

Ping(ICMP) 4 packets, 9000 data bytes, interval 1 second(s).

9008 bytes from icmp_seq=0 ttl=64 time=7.777 ms

9008 bytes from icmp_seq=1 ttl=64 time=3.028 ms

9008 bytes from icmp_seq=2 ttl=64 time=1.771 ms

9008 bytes from icmp_seq=3 ttl=64 time=2.984 ms

If I ping from the switch to a dell server , the max mtu size I can achieve is :
# ping vr "VR-CS-SERVER" start-size 8976

Ping(ICMP) 4 packets, 8976 data bytes, interval 1 second(s).

8984 bytes from icmp_seq=0 ttl=64 time=0.487 ms

8984 bytes from icmp_seq=1 ttl=64 time=0.470 ms

8984 bytes from icmp_seq=2 ttl=64 time=2.732 ms

8984 bytes from icmp_seq=3 ttl=64 time=2.656 ms

ping vr "VR-CS-SERVER" start-size 8977
Ping(ICMP) 4 packets, 8977 data bytes, interval 1 second(s).

--- ping statistics ---
4 packets transmitted, 0 packets received, 100% loss
round-trip min/avg/max = 0/0/0 ms

IS this a limitation on Jumbo frames ?
Am I missing some overhead that is added to teh ping ?

Any help would be appreciated

2 replies

Userlevel 4
Hi Rod,

Have you checked the MTU size of the Dell server port?
It looks like the Dell server can support up to 8976 MTU size or it is configured with 8976 MTU size, so it can't reply with 8977 packet size as it exceeds the maximum size.
Userlevel 3

From the customer :

I have change the storage device ( to 9000 and when we ping –s9000 from the server ( to the storage it does fail thus:

# vmkping -d -I vmk1 -s 9000

PING ( 9000 data bytes

sendto() failed (Message too long)

sendto() failed (Message too long)

sendto() failed (Message too long)

However there might be a simple explanation for this.

This link http://rickardnobel.se/troubleshoot-jumbo-frames-with-vmkping/ provides the answers. Basically when we do a ping test we should do a test with –s 8972 to allow for the overheads in the ping command itself.

So this works a treat:

# vmkping -d -I vmk1 -s 8972

PING ( 8972 data bytes

8980 bytes from icmp_seq=0 ttl=64 time=0.151 ms

8980 bytes from icmp_seq=1 ttl=64 time=0.133 ms

8980 bytes from icmp_seq=2 ttl=64 time=0.135 ms

This seems to have working in this scenario

Something to lookout for in the future.