Trouble getting Talos set up in Proxmox on IPv6-only Network

[jordangarside]

I've spent quite a bit of time now trying to get a single Talos control plane node or worker installed correctly in Proxmox on my IPv6 only network.
v1.8.0

I figured out I can put DNS server, Gateway, and Static IP in the network config page on the installer ISO.

So I do that:

DNS: fd01:10:100::1:1:1:1
Interface: enx921536f1f4fd
Mode: Static
Addresses: fd01:10:100:0:3::20/48
Gateway: fd01:10:100::1

And then it seems like it's working, and I can ping it.

I go from seeing these errors:

To seeing this (looks good I think?):

But then I start seeing a new error

Not sure if it matters? I can connect to the VM over the network at this point.

If I try to apply the config then this happens:

Gives me some kind of "no disks matched for volume" error.
I've checked multiple times and /dev/sda exists, and I've also tried using diskSelector: model: "QEMU HARDDISK" instead and get the same error.

➜ talosctl -n fd01:10:100:0:3::20 disks --insecure
DEV          MODEL           SERIAL   TYPE      UUID   WWID   MODALIAS      NAME   SIZE     BUS_PATH                                                                  SUBSYSTEM          READ_ONLY   SYSTEM_DISK
/dev/loop0   -               -        UNKNOWN   -      -      -             -      76 MB    /virtual                                                                  /sys/class/block   *           
/dev/sda     QEMU HARDDISK   -        HDD       -      -      scsi:t-0x00   -      69 GB    /pci0000:00/0000:00:05.0/0000:01:01.0/virtio2/host2/target2:0:0/2:0:0:0   /sys/class/block               
/dev/sr0     QEMU DVD-ROM    -        CD        -      -      scsi:t-0x05   -      106 MB   /pci0000:00/0000:00:01.1/ata2/host1/target1:0:0/1:0:0:0                   /sys/class/block 

It says it starts pulling the image and then nothing happens after that (last line in previous screenshot).

Also after the 1st apply I can no longer apply any config changes:

➜ talosctl apply-config --insecure --nodes fd01:10:100:0:3::20 --file worker.yaml
error applying new configuration: rpc error: code = Unavailable desc = connection error: desc = "transport: Error while dialing: dial tcp [fd01:10:100:0:3::20]:50000: connect: connection refused"

I start getting 2 network errors after the apply as well (compared to 1 before):

I assume the new one is because of my network config not specifying priority in the route:

network:
    nameservers:
        - fd01:10:100::1:1:1:1
    interfaces:
        - interface: enx921536f1f4fd
        dhcp: false
        addresses:
            - fd01:10:100:0:3::20/48
        routes:
            - network: ::/0
            gateway: fd01:10:100::1

Ultimately I'd like to do managed DHCPv6, but I've been trying to get static IPs working first just to test Talos out.

[smira]

The error about adding route is certainly duplicate routes coming in, as you have one set in the machine config, and another via the dashboard - you don't need two of them.

The error about pulling is due to the fact that ghcr.io doesn't have IPv6 connectivity. We can't fix it unfortunately, but you can host a registry mirror on IPv4/6 mixed environment and use that instead.

You can configure Talos to use that mirror instead of ghcr.io.

[jordangarside]

I get the duplicate route error before I apply any machine config (it happens after I use the dashboard). Maybe it’s from the RAs from my router after IPv6 is enabled in the dashboard? I have nat64 and dns64 so there is no public IPv6 traffic. I actually don't have a public IPv6 address. Did I get a pull error? I didn't see any error specifically, it just hangs. Is the volume error a red herring? Thanks!

[smira]

Might be RAs, you know it better than me, but in the end it's not the problem if it's same route.

I don't have a way to guess why pull would fail, but IPv6-only Talos machine won't be able to connect to IPv4 address I guess. The error should be printed eventually.

[jordangarside]

It's very possible the download fails/stalls, I was having some issues earlier with my nat64 dropping packets. Is there any timeout on the download, or would it just hang forever?

[jordangarside]

Confirmed it was packet loss issues related to my nat64 gateway.

I mostly got confused by all the other messages going on and didn't realize that the download was still being attempted.