Nodes don't become "online" after install due to network interface renaming
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
StarlingX |
Fix Released
|
Medium
|
Steven Webster |
Bug Description
Brief Description
-----------------
It seems that the installer (anaconda) uses interfaces eno1 and eno2 after PXE-boot. The installer will then create the network ifcfg scripts accordingly...
After first reboot the interfaces eno1/eno2 don't exist anymore, but got renamed to enp25s0f0/enps0f1.
This second rename is not visible in the installer.
This affects Mellanox w/ OFED version 5.4+ only, as a 2nd udev rule (82-net-
This also only affects systems with 'onboard' Mellanox devices. Most systems use a separate MLX PCI card and would not experience this issue as the renaming in the two udev rules file would be consistent.
Severity
--------
Critical: System/Feature is not usable due to the defect
Steps to Reproduce
------------------
Must have a system with onboard Mellanox devices using the mlx5_core driver.
PXE boot (install).
After the first reboot, the /etc/sysconfig/
Expected Behavior
------------------
The renaming should be consistent between install time and runtime.
Actual Behavior
----------------
The renaming is not consistent between install time and runtime.
Reproducibility
---------------
100%
System Configuration
-------
N/A, just requires onboard MLX devices
Branch/Pull Time/Commit
-------
master
Last Pass
---------
N/A I have never seen a system with onboard MLX devices.
Timestamp/Logs
--------------
N/A not needed
Test Activity
-------------
Evaluation
Workaround
----------
Change the PHYSDEV in the relevant /etc/sysconfig/
Changed in starlingx: | |
importance: | Undecided → Medium |
tags: | added: stx.8.0 |
Fix proposed to branch: master /review. opendev. org/c/starlingx /kernel/ +/850023
Review: https:/