Post by Boris Marinov

81d

I might be doing #incus networking wrong. Network forwards work great, but since the created container gets a dynamic IP, the forward breaks on restart.

My aim is to expose the system container on the external network with its own address. For that I am
- creating a network forward
- setting the listen address to the one configured on the physical interface
- grabbing target address from the ipv4_address attribute

Using terraform and a small module I wrote to spin these up, the relevant bits:
```
resource "incus_network_forward" "ipv4_forward" {
network = var.bridge_net
listen_address = var.net_forward_lxc_listen_addr

config = {
target_address = incus_instance.lxc.ipv4_address
}
}
```

I am using something wrong obviously, but I am a bit lost on what else there is which will let me configure the container networking in this manner.

1 0 0 View Post & Replies See Original

81d

@bmarinov container IP addresses should be mostly stable as dnsmasq tries to assign them based on a hash of the MAC. But you can also set a static DHCP lease by setting ipv4.address on the NIC device (typically eth0).

1 0 0 View Post & Replies See Original

81d

@stgraber
Thanks for chiming in, so network forward and setting up the target (internal) address is the way to go.

Actually something different might be happening here - it's not the internal IP randomly changing. On consequent runs of tf apply I get a zeroed out network address. And if I confirm the apply, this breaks the network forward. When I rerun terraform apply a few times, eventually the IP attribute reappears with the correct value set. I missed this last night as it was getting late.
It seems though that I might be experiencing some weird issue with the terraform provider occasionally failing to output dynamic values from the resource attributes. I can't figure out what is triggering it, but it is happening very often.

Edited 81d ago

0 0 0 View Post & Replies See Original