When the "get interface" command is executed, some interface names are changed to "renameXX."
> get interface
bond0 Network interface
docker0 Network interface
eth0 Network interface
eth1 Network interface
eth10 Network interface
eth11 Network interface
eth2 Network interface
eth3 Network interface
eth4 Network interface
eth5 Network interface
eth6 Network interface
eth7 Network interface
eth8 Network interface
eth9 Network interface
lo Network interface
renameXX Network interface <---!
renameXX Network interface <---!
VMware NSX-T Data Center 3.0
VMware NSX-T Data Center 3.1
VMware NSX-T Data Center 3.2
VMware NSX 4.0
VMware NSX 4.1
The udev rules files will get regenerated from scratch on every reboot which may lead to some instability of interface naming. The kernel will initially give the interface a name based on PCI bus discovery timing, and the order of interface discovery could potentially change between reboots.
The "renameXX" naming is a side effect of systemd-udev renaming the interfaces to align with the udev rules file. The version of systemd used in a NSX-T release uses an intermediate name while renaming to avoid a clash with another interface already having the same destination name.
This issue is resolved in VMware NSX-T Data Center 3.2.4
This issue is resolved in VMware NSX 4.1.2
Workaround
Reboot the NSX Edge.
Note:
To prevent recurrence, there is a systemd service that does the udev rules regeneration called 'nsx-edge-net-rules.service', which can be disabled and masked to prevent it from running.
systemctl disable nsx-edge-net-rules.service
systemctl mask nsx-edge-net-rules.service
When adding or removing physical NICs, it is necessary to enable the 'nsx-edge-net-rules.service'. After completing the work, please disable the 'nsx-edge-net-rules.service' again.
systemctl enable nsx-edge-net-rules.service
systemctl unmask nsx-edge-net-rules.service