Re: [PATCH proxmox-ve-rs v4 12/31] ve-config: fabrics: wireguard add validation for wireguard config

[Stefan Hanreich <s.hanreich@proxmox.com>]

On 5/11/26 3:35 PM, Arthur Bied-Charreton wrote:
>> +        if let FabricEntry::WireGuard(entry) = self {
>> +            // check if all interfaces referenced by the peer definitions exist inside the
>> +            // fabric
>> +            let mut all_interfaces = HashSet::new();
>> +            let mut internal_peers = HashSet::new();
>> +
>> +            for node_id in entry.nodes.keys() {
>> +                let node_section = entry.node_section(node_id)?;
>> +
>> +                if let WireGuardNode::Internal(node) = node_section.properties() {
>> +                    all_interfaces.extend(
>> +                        node.interfaces()
>> +                            .map(|interface| (&node_section.id.node_id, &interface.name)),
>> +                    );
>> +
>> +                    internal_peers.extend(node.peers().filter_map(|peer| {
>> +                        if let WireGuardNodePeer::Internal(peer) = peer {
>> +                            return Some((&peer.node, &peer.node_iface));
>> +                        }
>> +
>> +                        None
>> +                    }));
>> +                }
>> +            }
>> +
>> +            for (node_id, interface) in internal_peers.difference(&all_interfaces) {
>> +                return Err(FabricConfigError::InvalidRemoteInterfaceReference(
>> +                    interface.to_string(),
>> +                    node_id.to_string(),
>> +                ));
>> +            }
> Might make sense to also check for interface IPs that are already set on
> other nodes to prevent having 2 nodes with the same IP in a WG fabric.

see my reply to the top-level response

[snip]

>> +impl Validatable for InternalWireGuardNode {
>> +    type Error = FabricConfigError;
>> +
>> +    /// Validates the [FabricSection<WireGuardNodeProperties>].
>> +    ///
>> +    /// Checks if we have either an IPv4 or an IPv6 address. If neither is set, return an error.
> Looks like this doc comment is the result of a copy-paste error

indeed, have removed it!