We are having a very similar issue with our Intel X710 based NIC's, that seemed to start happening out of the blue. We are running a Hyper-V cluster based on Dell R730 servers, and recently as we have started to increase the load in the environment we are getting flooded with the same events from Windows (2012 R2 Datacenter Core). After reading some other threads we have disabled tcp checksum offload, and so far today thing seem ok, but last week it started to cause issues with our guests. We have our network adapters configured with LACP trunking connected to a pair of Avaya VSP switches. Would it be possible to look into this further for us if we provide additional necessary information?
Thank you for the post. Can you provide more information to better assist on this.
What is the exact X710 NIC model?
NVM version :
Driver version :
Dell bios :
I also send a PM message.
Will reply to your PM, but for the meantime this is what our servers currently look like
Dell BIOS: 2.1.7 (for Dell R730 servers)
Switch FW: Avaya Virtual Services Platform 9000 switch with 3.4.5 code, and we have ports trunks with LACP between two VSP switches in an SMLT configuration.
Should note that per reading some other things we have tried disabling checksum offload in Windows, however, this doesn't seem to alleviate the issue. It does however seem that the Hyper-V guests that incur heavy network load induce this issue, as we seem to find that the event of disconnects will follow those gusts from Hyper-V server to Hyper-V server within the cluster.
One of the main reasons we are seeming to point our finger towards the Intel network adapters is that we have other Hyper-V servers in a separate cluster, but configured the same way in the same data centers, that contain Emulex branded 10Gb network interfaces, which are not displaying this issue.
Good day. Please help provide the serial number of your X710 network adapter for better investigation, the serial number can be found at the white sticker on the NIC, format of the serial #
e.g. 15 alpanumerics + 6 alphanumeric + 6 alphanumeric - 3 digits.
Is there anyway to obtain the serial number from a command... since the servers are in production at a location different from where I work it's a bit time consuming to be able to open these servers up to read this off.
Please send me the SSU log, U can download the SSU tool at https://downloadcenter.intel.com/download/25293/Intel-System-Support-Utility, run the tool and attach the log generated for me to check.
I am also having the same issue with Intel x710 bNDCs on Dell FC630 blades. When the network load increases on our servers, we see the error more often. I currently have two ports in a switch independent team, using dynamic load balancing with both adapters active.
I get an Event 27, i40ea, Network link is disconnected error.
Two seconds later I get a Event 64, i40ea, This network adapter's etrack ID is: 800024C3.
At the same time, as the Event 64, I get a Event 32, i40ea, Network link has been established at 1Gbps full duplex.
On the switch port, I see no errors and the port does not go down.
What information do you need from the ssu tool? My servers are on disconnected networks and I can only share certain information.
Part Number: Intel Ethernet 10G 4P X710-k bNDC
NVM version : 5.4
Driver version : 188.8.131.52
Dell bios : 2.1.6
switch FW: Brocade ICX7450 running FastIron 8030j
Thank you for the post. Just to double check is this an onboard NIC on the Dell system? If yes, have you check with Dell support?
It is a network daughter card. I have not yet involved Dell support. I saw many posts on this message board about X710 NICs with disconnect issues. Have Intel engineers been made aware of the issue?
Thank you for the reply. I am gather information from Eric but no further update yet. Are you using the network driver from Dell*?
Yes, we are using the latest Dell Drivers and Firmware. As others have stated, turning off hardware offload features seems to fix the disconnect problem. I am going to slowly enable offload features to figure out if there is a specific feature causing problems.
These servers are Hyper-V hosts. Virtual Machine Queuing has been disabled from the start.