How to use PING to check for proper MTU level for fragmentation for storage networks
book
Article ID: CTX236751
calendar_today
Updated On:
Description
There are many times in XenServer environments where the customer will have implemented MTU of 9000 to be able to enjoy the benefits of Jumbo Frames. While this is 100% supported, it is not 100% recommended.
For the best customer experience we need to be able to suggest, test, and implement the correct MTU for the customer environment.
If the blanket 9000 is used, the Cx may see network errors with iSCSI such as 1020 and 1010, or have intermittent connection issues in with NFS\SMB.
Instructions
The base level of this is that we must ping FROM the storage network interface TO the SAN IP. We do this with the -I flag to be able to choose the interface that is used.
The command structure would look like this:
ping -I eth<port of network> -M do -s <MTU TO TEST> <SAN_IP>
example: ping -I eth4 -M do -s 7900 10.254.6.24
The following are examples of responses you would see in CLI. These are set for If you see this (or similar): From <host> (IP) icmp_seq=1 Frag needed and DF set (mtu = <MTU USED>), it means the MTU you are testing is fragging out and needs to be lowered.
If you get this (or similar): PING <destination> (destination_ip) from <host_ip> eth0: <MTU USED(ACTUAL_MTU_WITH_OVERHEAD)> bytes of data, it means you have hit the sweet spot for non fragmentation.
Issue/Introduction
To be able to set correct MTU level for best customer experiences
Was this article helpful?
thumb_up
Yes
thumb_down
No