On Wed, Jul 02, 2025 at 04:49:58PM +0200, Gabriel Goller wrote:
> Add OSPF-specific FRR types. This also reuses the types from
> proxmox-network-types.
>
> The NetworkType FRR option is implemented here, but not exposed to the
> interface, as we want to keep it simple for the users. If they do not
> set an IP, then the interface is considered to be unnumbered and uses
> the Point-to-Point network type.
>
> Signed-off-by: Gabriel Goller <g.gol...@proxmox.com>
> ---
> proxmox-frr/src/lib.rs | 25 ++++++
> proxmox-frr/src/ospf.rs | 179 ++++++++++++++++++++++++++++++++++++++++
> 2 files changed, 204 insertions(+)
> create mode 100644 proxmox-frr/src/ospf.rs
>
> diff --git a/proxmox-frr/src/lib.rs b/proxmox-frr/src/lib.rs
> index ba9eedfb4549..0d94aef5a3cd 100644
> --- a/proxmox-frr/src/lib.rs
> +++ b/proxmox-frr/src/lib.rs
> @@ -1,4 +1,5 @@
> pub mod openfabric;
> +pub mod ospf;
> use std::{fmt::Display, str::FromStr};
>
> use serde::{Deserialize, Serialize};
> @@ -31,6 +32,30 @@ impl Display for InterfaceName {
> }
> }
>
> +/// Generic FRR Interface.
> +///
> +/// In FRR config it looks like this:
> +/// ```text
> +/// interface <name>
> +/// ! ...
> +#[derive(Clone, Debug, PartialEq, Eq, Deserialize, Serialize, PartialOrd,
> Ord)]
> +pub enum Interface {
> + Openfabric(openfabric::OpenfabricInterface),
> + Ospf(ospf::OspfInterface),
> +}
> +
> +impl From<openfabric::OpenfabricInterface> for Interface {
> + fn from(value: openfabric::OpenfabricInterface) -> Self {
> + Self::Openfabric(value)
> + }
> +}
> +
> +impl From<ospf::OspfInterface> for Interface {
> + fn from(value: ospf::OspfInterface) -> Self {
> + Self::Ospf(value)
> + }
> +}
> +
> #[derive(Error, Debug)]
> pub enum FrrWordError {
> #[error("word is empty")]
> diff --git a/proxmox-frr/src/ospf.rs b/proxmox-frr/src/ospf.rs
> new file mode 100644
> index 000000000000..f7e63fbe4e55
> --- /dev/null
> +++ b/proxmox-frr/src/ospf.rs
> @@ -0,0 +1,179 @@
> +use std::fmt::Debug;
> +use std::fmt::Display;
> +use std::net::Ipv4Addr;
> +
> +use serde::{Deserialize, Serialize};
> +
> +use thiserror::Error;
> +
> +use crate::{FrrWord, FrrWordError};
> +
> +/// The name of the ospf frr router.
> +///
> +/// We can only have a single ospf router (ignoring multiple invocations of
> the ospfd daemon)
> +/// because the router-id needs to be the same between different routers on
> a single node.
> +/// We can still have multiple fabrics by separating them using areas.
> Still, different areas have
> +/// the same frr router, so the name of the router is just "ospf" in "router
> ospf".
> +///
> +/// This serializes roughly to:
> +/// ```text
> +/// router ospf
> +/// !...
> +/// ```
> +#[derive(Clone, Debug, PartialEq, Eq, Hash, Deserialize, Serialize,
> PartialOrd, Ord)]
> +pub struct OspfRouterName;
> +
> +impl Display for OspfRouterName {
> + fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
> + write!(f, "ospf")
> + }
> +}
> +
> +#[derive(Error, Debug)]
> +pub enum AreaParsingError {
> + #[error("Invalid area idenitifier. Area must be a number or an ipv4
> address.")]
> + InvalidArea,
> + #[error("Invalid area idenitifier. Missing 'area' prefix.")]
> + MissingPrefix,
> + #[error("Error parsing to FrrWord")]
> + FrrWordError(#[from] FrrWordError),
> +}
> +
> +/// The OSPF Area.
> +///
> +/// The OSPF area is a pseud-ipaddress (so it looks like an ip-address but
> isn't set on any
> +/// interface or even pingable), but can also be specified by a simple
> number. So you can use "5"
> +/// or "0" as an area, which then gets translated to "0.0.0.5" and "0.0.0.0"
> by FRR. We allow both
> +/// a number or an ip-address. Note that the area "0" (or "0.0.0.0") is a
> special area - it creates
> +/// a OSPF "backbone" area.
> +#[derive(Clone, Debug, PartialEq, Eq, Hash, Deserialize, Serialize,
> PartialOrd, Ord)]
> +pub struct Area(FrrWord);
> +
> +impl TryFrom<FrrWord> for Area {
> + type Error = AreaParsingError;
> +
> + fn try_from(value: FrrWord) -> Result<Self, Self::Error> {
> + Area::new(value)
> + }
> +}
> +
> +impl Area {
> + pub fn new(name: FrrWord) -> Result<Self, AreaParsingError> {
> + if name.as_ref().parse::<u32>().is_ok() ||
> name.as_ref().parse::<Ipv4Addr>().is_ok() {
> + Ok(Self(name))
> + } else {
> + Err(AreaParsingError::InvalidArea)
> + }
> + }
> +}
> +
> +impl Display for Area {
> + fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
> + write!(f, "area {}", self.0)
> + }
> +}
> +
> +/// The OSPF router properties.
> +///
> +/// Currently the only property of a OSPF router is the router_id. The
> router_id is used to
> +/// differentiate between nodes and every node in the same area must have a
> different router_id.
> +/// The router_id must also be the same on the different fabrics on the same
> node. The OSPFv2
> +/// daemon only supports IPv4.
> +/// Note that these properties also serialize with a space prefix (" ") as
> they are inside the OSPF
> +/// router block. It serializes roughly to:
> +///
> +/// ```text
> +/// router ospf
> +/// router-id <ipv4-address>
> +/// ```
> +#[derive(Clone, Debug, PartialEq, Eq, Hash, Deserialize, Serialize,
> PartialOrd, Ord)]
> +pub struct OspfRouter {
> + pub router_id: Ipv4Addr,
> +}
> +
> +impl OspfRouter {
> + pub fn new(router_id: Ipv4Addr) -> Self {
> + Self { router_id }
> + }
> +
> + pub fn router_id(&self) -> &Ipv4Addr {
> + &self.router_id
> + }
> +}
> +
> +#[derive(Error, Debug)]
> +pub enum OspfInterfaceError {
> + #[error("Error parsing area")]
> + AreaParsingError(#[from] AreaParsingError),
> + #[error("Error parsing frr word")]
> + FrrWordParse(#[from] FrrWordError),
> +}
> +
> +/// The NetworkType of the interface.
> +///
> +/// The most important options here are Broadcast (which is the default) and
> PointToPoint.
> +/// When PointToPoint is set, then the interface has to have a /32 address
> and will be treated as
> +/// unnumbered.
> +///
> +/// This roughly serializes to:
> +/// ```text
> +/// ip ospf network point-to-point
> +/// ! or
> +/// ip ospf network broadcast
> +/// ```
> +#[derive(Clone, Debug, PartialEq, Eq, Hash, Deserialize, Serialize,
> PartialOrd, Ord)]
> +pub enum NetworkType {
> + Broadcast,
> + NonBroadcast,
> + /// If the interface is unnumbered (i.e. the router-id /32 ip-address is
> set on the interface).
> + ///
> + /// If OSPF is used in an unnumbered way, you don't need to configure
> peer-to-peer (e.g. /31)
> + /// addresses at every interface, but you just need to set the router-id
> at the interface
> + /// (/32). You also need to configure the `ip ospf network
> point-to-point` FRR option.
> + PointToPoint,
> + PointToMultipoint,
Will there be larger entries? Could consider `Copy` here maybe.
> +}
> +
> +impl Display for NetworkType {
> + fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
> + match self {
> + NetworkType::Broadcast => write!(f, "broadcast"),
> + NetworkType::NonBroadcast => write!(f, "non-broadcast"),
> + NetworkType::PointToPoint => write!(f, "point-to-point"),
(I see a `t` where some people don't use a `t` 🤪 /hj)
> + NetworkType::PointToMultipoint => write!(f,
> "point-to-multicast"),
> + }
> + }
> +}
> +
> +/// The OSPF interface properties.
> +///
> +/// The interface gets tied to its fabric by the area property and the FRR
> `ip ospf area <area>`
> +/// command.
> +///
> +/// This serializes to:
> +///
> +/// ```text
> +/// router ospf
> +/// ip ospf area <area>
> +/// ip ospf passive <value>
> +/// ip ospf network <value>
> +/// ```
> +#[derive(Clone, Debug, PartialEq, Eq, Hash, Deserialize, Serialize,
> PartialOrd, Ord)]
> +pub struct OspfInterface {
> + // Note: an interface can only be a part of a single area(so no vec
> needed here)
> + pub area: Area,
> + pub passive: Option<bool>,
> + pub network_type: Option<NetworkType>,
> +}
> +
> +impl OspfInterface {
> + pub fn area(&self) -> &Area {
> + &self.area
> + }
> + pub fn passive(&self) -> &Option<bool> {
> + &self.passive
> + }
> + pub fn network_type(&self) -> &Option<NetworkType> {
> + &self.network_type
> + }
^ like in the previous patch - pub fields vs getters
> +}
> --
> 2.39.5
_______________________________________________
pve-devel mailing list
pve-devel@lists.proxmox.com
https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-devel
