Data Domain: Network Interfaces May Go Down Due to Intel Workaround Configuration
Summary: Network interfaces may go down, become intermittent, or fail to recover properly following events such as node reboots, HA failovers, link renegotiation, or other network transitions on systems equipped with Intel NICs. This issue can occur when the Intel workaround for M14MK dual-rate transceivers remains configured after the interfaces have been migrated to W4GPP 25Gb transceivers. The resulting speed mismatch can cause interface connectivity loss and intermittent network behavior. ...
Symptoms
Network interfaces unexpectedly go down.
Interfaces may fail to recover following a reboot, failover, or link transition event.
Connectivity may be temporarily restored by reseating optics or resetting switch ports.
Intermittent network connectivity alerts may be reported.
Repeated link up/down events may be observed in logs.
Alerts similar to the following may be present:
EVT-NETM-00001: Network interface connectivity is down
EVT-NETM-00002: Network interface connectivity is intermittent
Cause
The Intel workaround intended for M14MK dual-rate 10Gb/25Gb transceivers remained enabled after W4GPP 25Gb transceivers were installed.
The affected interfaces were configured to operate at 25Gb while the Intel workaround continued to apply a 10Gb setting, creating a speed mismatch during link initialization and renegotiation events.
Resolution
Review the Autosupport to verify both the installed transceiver type and the configured Intel workaround.
Verify the transceiver type
From the Autosupport, review Transceiver Details for Interface and confirm the installed optic part number.
Example:
Before
Transceiver Details for Interface "eth8a":
Vendor PN : M14MK
Transceiver Details for Interface "eth8b":
Vendor PN : M14MKAfter
Transceiver Details for Interface "eth8a"
Vendor PN : W4GPP
Transceiver Details for Interface "eth8b"
Vendor PN : W4GPP
Verify the Intel workaround configuration
From the Autosupport configuration section, look for entries similar to:
config_master.net.options.intel_workaround.eth8a = 10000
config_master.net.options.intel_workaround.eth8b = 10000
If W4GPP transceivers are installed and the Intel workaround is still configured, remove the workaround from the affected interfaces:
net option reset intel_workaround.eth8a
net option reset intel_workaround.eth8b
After the workaround is removed, the interfaces can negotiate and operate at the configured speed without the conflict.
Additional Information
The Intel workaround referenced in kb Data Domain auto negotiating to 10 Gbps may not be possible with certain switches is intended for specific M14MK transceiver scenarios. If M14MK optics are later replaced with W4GPP 25Gb optics, the workaround may no longer be required and can lead to link-speed conflicts.