MTU settings in vSphere 4.1, ESXi

[VMLost]

Just want to know if anyone can confirm the following.

trying  to setup multipathing with the DELL EQ mutipathing plugin. After setting  the MTU to 9000, after the script runs, it errors out and no vmnics are  added. Specific error is as follows :

     " Operation failed,  diagnostics report: Unable to add uplink! The MTU setting used by this  virtual switch is not supported by the NIC being added."

Now,  just to verify : DOES MTU/jumbo frames need to infact be set to 900 on  the ISCSI switch first, in order for me to enable it on the vswitch?  Currently my iSCSI switch is set to 1500 only.

thanks in advance

[jsanabria]

Hi J-D, I have been searching quite a while, people like me with HP NC375T NICs in their ESXi Virtualization environment. Actually I have three DL380 G7 servers with ESXi 4.1.0, and with integrated Broadcom NIC's. Apart of them, I have three quad-port HP NC375T NICs, one for each server. Testing the environment, I have detected some virtual machine connectivity drops suddenly. When I check the networking tab in configuration of the ESXi machines, I see no connectivity at all on HP NC375T NICs. And there are a lot of links with problems with these NICs of HP in Internet.

Now, I'm thinking of upgrading the firmware & driver of these hosts. Actually I have firmware version 4.0.544 & driver version 4.0.550. If I have understood your post, you have upgraded to lastest firmware version 4.0.598 & latest driver version 4.0.556. Are these upgrades stable, did you notice anything? Thanks:

JS.

[J-D]

Hey, we only have the built-in 375i's and currently these are the driver and firmware versions:

[root@sevmw3 ~]# ethtool -i vmnic0

driver: nx_nic

version: 4.0.598

firmware-version: 4.0.556

bus-info: 0000:04:00.0

This is on 4.1u1 (not u2)

Before that upgrade we had VMotion issues. After the upgrade VMotion went fine (just 1 ping lost instead of 6-8).

We have 2 DL580 G7's and 1 has suddenly disconnected from vCenter while the other has been running for a month now without any issues.

We currently have a call open at HP. I wonder if there is a bad batch of servers. The one without problems was bought about 7 months later.

HP has escalated this to VMware. It's very strange our ESX got disconnected as we have made the service console redundant by creating a vswitch with 2 vmnic's, 1 tied to the onboard 375i and one to a dualport nic which has never shown issues. It's as if the port shows as being alive but suddenly doesn't pass any traffic. It happened about 14 days after being under load so quite difficult to troubleshoot. The other DL580 G7 however seems to run stable...

I think you should definitely upgrade firmware and driver.

hope this info helps...

[jsanabria]

Hi J-D,

your post does help, for sure, thanks. Although your NICs are not exactly the same, indeed, my hosts are 4.1 U1 (not U2) also, and we have suffered also, of "alive" like ports on these NICs, while there were not sending or receiving any traffic. Therefore virtual machines depending on these ports, were totally isolated from the network. Definitively I will upgrade the hosts. Thanks for your feedback, JS.