This post is more than 5 years old
8 Posts
0
55830
Enabling Jumbo Frames on MD3220i SAN
I am trying to setup Jumbo Frames on my Virtualised System.
To do this I have done the following:
1) Enable Jumbo Frames on my Dell MD3220i SAN by checking the "Enable jumbo frames" box for each port and setting the MTU value to 9000.
2) Enable Jumbo Frames on my iSCSI switch (Dell PowerConnect 7024) by setting the "Maximum Frame Size" to 9216 on all ports. (I then power cycled the Switch and confirmed that the settings had saved)
3) Enable Jumbo Frames on my Host (Hyper-V Server 2008 R2 Core OS) by using the following command on each NIC:
netsh interface ipv4 set subinterface "Adapter Name" mtu=9000 store=persistent
I also edited the registry to update the *JumboPacket key to 9014 on all relevant NICs
I then rebooted the host for the changes to take effect.
To test my configuration - I used the standard ping command to attempt to ping the SAN (192.168.130.101) from the Host (192.168.130.3).
The results I got are:
ping -f -l 1472 192.168.130.101 (This is ok - and I get a standard reply)
ping -f -l 1476 192.168.130.101 (This is ok - and I get a standard reply)
ping -f -l 1477 192.168.130.101 (This is bad - I get a "Request timed out.")
ping -f -l 8972 192.168.130.101 (This is bad - I get a "Request timed out.")
ping -f -l 8973 192.168.130.101 (For this I get "Packet needs to be fragmented but DF set.")
I am pretty sure this means that the Jumbo Frames config is not working - but I am fresh out of ideas.
Any help will be much appreciated.
MrMoo
8 Posts
1
February 12th, 2014 01:00
Hi Sam,
I have resolved my issue!
The NICs on the Host are unable to take a value of > 9000 for the JumboPackets in the Registry. Setting them to 9014 did not work - as they must have reverted to using 1500 (even though the registry value remains at 9014)
Therefore if I set the registry value to 9000 for eah NIC - the Ping command works.
Regards
Simon
DELL-Sam L
Moderator
Moderator
•
7.1K Posts
0
February 11th, 2014 12:00
Hello MrMoo,
How many nic’s do you have connected on your server to the MD3220i? Also how many different subnets do you have for your ports on the MD3220i? I am not sure if you used our deployment guide when setting up your MD3220i but if not then here is a link to check it and make sure that you haven’t missed anything. ftp://ftp.dell.com/Manuals/Common/powervault-md3200i_Deployment%20Guide_en-us.pdf
Please let us know if you have any other questions.
MrMoo
8 Posts
0
February 11th, 2014 14:00
Hi Sam,
I am using the default IP addresses for each of the iSCSI ports on both Controller 0 and Controller 1.
i have setup my host (Hyper-V Server 2008 R2) and I am using 2 NICs on this host for SAN traffic, the IP Addresses that I have assigned are 192.168.130.3 (255.255.255.0) and 192.168.132.3 (255.255.255.0). Each of these NICs is connected through a separate Dell iSCSI switch to the SAN.
I can successfully ping each of the 8 SAN ip addresses, 4 on primary switch and 4 on secondary switch. All subnet masks on the SAN ports are 255.255.255.0.
Only when I try to ping large packets do I get errors.
Regards
Simon
Dev Mgr
4 Operator
4 Operator
•
9.3K Posts
0
February 12th, 2014 05:00
This isn't a solution, but may help rule out (or identify) the switch for being the cause of this issue:
Take the NIC in the 192.168.130.x subnet and direct-connect it to the top controller iSCSI port 0 and then try pinging again with 8972 (or any value larger than 1476 and smaller than 8973).