Fix rename interface

[schlichtanders]

works around #1560

This makes the interface renaming failure tolerant. Worked for me with most recent autoscaler.
Still it would be even better if the network interface eth0 wouldn't be disconnecting and reconnecting initially

[schlichtanders]

after some test runs, this unfortunatly is not enough - sometimes even after 3 minutes the network is not available... there is hopefully a solution to this somewhere

[M4t7e]

I haven't encountered this personally, but I've read about others facing similar issues with the internal network interface connections to Hetzner's network. Are the issues occurring with both eth0 and eth1, or just one?

To ensure the configuration is as fault-tolerant as possible, I suggest waiting for the network interfaces to be fully up before proceeding with nmcli after this:

terraform-hcloud-kube-hetzner/locals.tf

Line 848 in d2d4423

ip link set eth1 up

Example snippet to wait for up state:

until [ "$(cat /sys/class/net/eth0/operstate)" = "up" ]; do
    echo "Waiting for eth0 to be up"
    sleep 2
done

until [ "$(cat /sys/class/net/eth1/operstate)" = "up" ]; do
    echo "Waiting for eth1 to be up"
    sleep 2
done

If the problem persists after this, there may be an underlying issue or race condition with nmcli.