The bare-metal edge's eth0 interface is down after the reboot.

Products

• VMware NSX
• VMware NSX-T Data Center

Issue/Introduction

• The bare-metal edge eth0 interface is down after the reboot.
• The following log has been recorded in the edge's syslog: the eth0 interface shows a DOWN status.

2024-08-11T11:00:00.111Z Edge NSX 5191 - [nsx@6876 comp="nsx-edge" subcomp="nsd" tid="2291" level="INFO"] Updating for iface kni-lrport-0 ends. Record runtime state:#012List of Namespaces:#012{Namespace underlay:#012 [Interface uplink-260: type: vlan admin_status: DOWN op_status: 0 UNKNOWN vlan_id: 1 if_index: 14 parent_if_index: 11 parent_ifname: if_flags: 69699 mtu: 1700 mac: ##:##:##:##:##:## ip: ##.##.##.##/## ]#012 [Interface lo: type: admin_status: UP op_status: 0 UNKNOWN if_index: 1 if_flags: 65609 mtu: 65536 mac: ##:##:##:##:##:## ip: ##.##.##.##/## ip: ::1/128 ]#012}#012{Namespace root:#012 [Interface kni-lrport-0: type: admin_status: UP op_status: 0 UNKNOWN if_index: 11 if_flags: 69699 mtu: 9190 mac: ##:##:##:##:##:## ]#012 [Interface kni-fp-eth0: type: admin_status: UP op_status: 0 UNKNOWN if_index: 10 if_flags: 69699 mtu: 1500 mac: ##:##:##:##:##:## ]#012 [Interface docker0: type: bridge admin_status: UP op_status: 6 UP if_index: 8 if_flags: 4163 mtu: 1500 mac: ##:##:##:##:##:## ip: ##.##.##.##/## ]#012 [Interface bond0: type: bond admin_status: UP op_status: 2 DOWN if_index: 7 if_flags: 5123 mtu: 1500 mac: ##:##:##:##:##:## ]#012 [Interface kni-fp-eth1: type: admin_status: UP op_status: 2 DOWN if_index: 9 if_flags: 69635 mtu: 1500 mac: ##:##:##:##:##:## ]#012 [Interface idrac: type: admin_status: DOWN op_status: 2 DOWN if_index: 6 if_flags: 4098 mtu: 1500 mac: ##:##:##:##:##:## ]#012 [Interface eth1: type: admin_status: DOWN op_status: 2 DOWN if_index: 3 if_flags: 4098 mtu: 1500 mac: ##:##:##:##:##:## ]#012 [Interface eth0: type: admin_status: DOWN op_status: 2 DOWN if_index: 2 if_flags: 4098 mtu: 1500 mac: ##:##:##:##:##:## ]#012 [Interface lo: type: admin_status: UP op_status: 0 UNKNOWN if_index: 1 if_flags: 65609 mtu: 65536 mac: ##:##:##:##:##:## ip: ##.##.##.##/## ]#012 [ARP ##.##.##.## mac: ##:##:##:##:##:## dev_index: 1 state: 64 type: 1 ]#012}

Environment

VMware NSX-T Data Center

VMware NSX

Cause

• When the Edge kernel boots, it utilizes the "Predictable Interface Names" method, which expects interface names like "ens8303" or "ens1f1" to be used.
• We found some Broadcom NIC ports are automatically initialized as "eth0" and "eth1". During the Edge boot process, there is a mechanism that attempts to convert interfaces named using the Predictable Interface Names method into the traditional "eth0" and "eth1" naming convention. However, since "eth0" and "eth1" already exist in your environment, the renaming process initially fails.
• Afterward, the Edge's periodically running detection function identifies the incorrect interface names and renames them back to "eth0". However, in the current version, the interface retains its previous status, which means that "eth0" remains in the DOWN state after the renaming.
• In future versions, this will be modified to ensure that the interface is set to UP after renaming.

Resolution

In the current version, we do not have a workaround for this issue. You can use the "ifup eth0" up command every time the edge reboots.