Unsolved
This post is more than 5 years old
11 Posts
0
6564
How to check that Jumbo frames are enabled
Hi, I'm hoping someone might be able to help me with this Jumbo frame query!
We have an FCIP connection between 2 sites. I've been advised by our network team that they have enabled Jumbo frames end to end. But, is there anyway I can check this from a Cisco MDS9513 switch running NXOS 4.2.3?
I have tried the ping command from the switch and seem to be getting....
# ping
Target IP address: xxx.xxx.xxx.130
Repeat count [5]:
Datagram size [100]: 2300
Timeout in seconds [1]: 2
Extended commands : y
Source address or interface: yyy.yyy.yyy.34
Type of service [0]:
Set DF bit in IP header : y
Data pattern [in hex (without leading 0x)]:
Sweep range of sizes :
PATTERN: 0xabcd
PING xxx.xxx.xxx.130 (xxx.xxx.xxx.130) from yyy.yyy.yyy.34 : 2300(2328) bytes of data.
From yyy.yyy.yyy.34 icmp_seq=1 Frag needed and DF set (mtu = 0)
From yyy.yyy.yyy.34 icmp_seq=1 Frag needed and DF set (mtu = 0)
From yyy.yyy.yyy.34 icmp_seq=1 Frag needed and DF set (mtu = 0)
From yyy.yyy.yyy.34 icmp_seq=1 Frag needed and DF set (mtu = 0)
From yyy.yyy.yyy.34 icmp_seq=1 Frag needed and DF set (mtu = 0)
--- xxx.xxx.xxx.130 ping statistics ---
0 packets transmitted, 0 received, +5 errors
I've no idea what this error means... Any suggestions? Or ideas as to how I can confirm Jumbo frames are enabled on the network side?
Many thanks
Mark
heaton1
11 Posts
0
October 28th, 2010 08:00
Thanks Hersh,
I did as you suggested - and the largest data packet I can use is 2272 - as below. I presume this is because I have my MTU set to 2300, and the ping command seems to add additional size.
# ping
Target IP address: xxx.xxx.xxx.130
Repeat count [5]:
Datagram size [100]: 2272
Timeout in seconds [1]: 2
Extended commands : y
Source address or interface: yyy.yyy.yyy.34
Type of service [0]:
Set DF bit in IP header : y
Data pattern [in hex (without leading 0x)]:
Sweep range of sizes :
PATTERN: 0xabcd
PING xxx.xxx.xxx.130 (xxx.xxx.xxx.130) from yyy.yyy.yyy.34 : 2272(2300) bytes of data.
2280 bytes from xxx.xxx.xxx.130: icmp_seq=1 ttl=251 time=7.07 ms
2280 bytes from xxx.xxx.xxx.130: icmp_seq=2 ttl=251 time=6.99 ms
2280 bytes from xxx.xxx.xxx.130: icmp_seq=3 ttl=251 time=7.02 ms
2280 bytes from xxx.xxx.xxx.130: icmp_seq=4 ttl=251 time=6.98 ms
2280 bytes from xxx.xxx.xxx.130: icmp_seq=5 ttl=251 time=6.97 ms
--- xxx.xxx.xxx.130 ping statistics ---
5 packets transmitted, 5 received, 0% packet loss, time 4046ms
rtt min/avg/max/mdev = 6.974/7.011/7.079/0.064 ms
So - it looks like Jumbo frames are enabled end to end. Thanks!
Mark
hersh1
197 Posts
1
October 28th, 2010 08:00
From your output it looks like something is set to less than 2328 bytes. You could adjust the datagram size to determine what the largest possible packet and what they have set. If the size was incorrect on your side you would get a message like this ( 3.3(4a) code):
ping: local error: Message too long, mtu=2300
kelleg
4.5K Posts
0
October 28th, 2010 12:00
From Primus emc151113:
Perform some ping tests the ensure that traffic is getting through consistently. The test consist of executing pings between the server and the array, varying the packet size (-l xxxx) with the non-fragmentable option (-f) set to determine the maximum MTU size that can be sent between the server and the array. Sizes that should be tried include: 1448, 1500 for small packet sizes and 4470, 5000, 7000, 8000, 8900, and 8972.
ping -t -f -l 8972 <IP address>
Note: MTU is the maximum amount of data in a frame, in this case an Ethernet frame not the maximum amount in an ICMP or TCP data. For a ping, an ICMP packet, you have to take into account the IP header and the ICMP header. There are 20 bytes and 8 bytes respectively. So 9000 - 20 - 8 = 8972 - this will be the maximum size that will work correctly when using ping - 8972.
glen