I have been experiencing NIC drops on multiple servers across our organization for over 6 months. I have had number cases open with Dell and Microsoft unsure if our 10gb switches were causing the issue or the OS or the NIC hardware. I can confidently say that the common factor is the X710 network card in all cases as we have switched out our 10gb switches to Cisco now and we are still seeing this issue.
The system event log shows an event 27 - Intel(R) Ethernet Converged Network Adapter X710 #X Network link is disconnected
There is no rhyme or reason as to when this happens, but I can guarantee at least 1 time per week or more I see this event.
I have a case open with Dell, again, and they had me update the firmware and drivers on the cards to: Firmware family: 17.5.11 and Driver Provider Intel, driver date 3/22/2016 22.214.171.124
These are 2 PCIe cards installed on a Dell PowerEdge R430.
During troubleshooting I disabled the following nic features:
* Encapsulated Task Offload
* IPv4 Checksum Offload
* Large Send Offload V2 (IPv4)
* Large Send Offload V2 (IPv6)
* TCP Checksum Offload (IPv4)
* TCP Checksum Offload (IPv6)
* UDP Checksum Offload (IPv4)
* UDP Checksum Offload (IPv6)
* Virtual machine Queues
Since making the changes to the 4 10gb nics on 1 of my 2012R2 Hyper-V hosts I have gone 3 weeks without an Event 27 NIC drop.
I did NOT make the change on my other 2 hosts, 1 hasn't dropped in 3 weeks and the other was good for 2 and now started dropping the last 4 days in a row, sometimes multiple times a day.
I have done hours and hours of research and just don't know if I need all these offloads disabled and what will happen to my virtual machines performance in my cluster if we turn off offloading.
Is there a driver update for these cards so we can enable offloading again or is this the only way these cards work? Does the disconnect happen because of load on the NIC and that is why it is so random, it shuts itself down if it gets too much traffic?
Thanks in advance for any insight into this problem.