Hello,
what are current best practices for managing network interfaces on AlmaLinux 9 nodes? Alma9/RHEL9/… do not include static network using /etc/sysconfig/network-scripts/ifcfg-ethX files anymore, so NetworkManager should be used instead. This leads to interface naming differences between nodes, while OpenNebula expects the same PHYSDEV=... for each node in a cluster. So, my questions are:
-
what approach do you use for stable network interface naming across your cluster? Something like
net.ifnames=0 biosdevname=0boot parameters, and then/etc/udev/rules.d/60-network-interfaces.rules? -
for one of my older VNets, I use
PHYSDEV=eth1. Is it possible to change this to another name without disrupting the VMs which currently use this VNet? -
if I want to provision the nodes using Ansible on a minimal install of AlmaLinux 9, I would probably need to set up at least one interface with its final name, so that Ansible does not cut the network connection under itself. Which approach do you use for this?
Thanks,
-Yenya
This is OpenNebula 6.4, CentOS 8 stream, which I want to gradually migrate to 6.6 and AlmaLinux 9.