Draft PR for 1.5 https://github.com/vyos/vyos-1x/pull/2508

Tested in VyOS 1.4-rolling-202311100309:
Configurations:

This PR LGTM as it addresses my main need, although I think it'd be better to add support for the rest of options from RFC 2132 (eventually).
When talking about request as an option I'm talking about adding a way to add a request (or also request) statement to dhclient.conf. I need to requst subnet-mask, routers, rfc3442-classless-static-routes from my ISP, and the current implementation doesn't allow me to do that.

Tested on VyOS 1.4-rolling-202311100309 and VyOS 1.5-rolling-202311160736 - L-Time shows 0. But supposed to show 3600 according to the configuration.

dhcp-server's raw parameters, which would pass options to dhclient as-is

oh okay. I don't know what else to do. this is difficult to replicate and it's scary if this happens when i am not at home and primary wan is also unavailable.

@GurliGebis definately

kea-dhcp will only be the server, not the DHCP client that you use here, I am unaware of a Kea based DHCP client.

This has happened 2-3 times since my last comment. I'll keep this open until kea-dhcp is merged and then check if this bug remains.

Can this get backported to 1.4?

PR https://github.com/vyos/vyos-1x/pull/2503

I agree, even if its "odd" at first sight I like that all interfaces are named ethX within VyOS and then its a matter to map each to physical interface by hw-id (which is done automagically during first install but can be remapped if wanted).

I am against this change, why not use eth0 everywhere? If it's a WAN or LAN port should be user decidable. Please use udev rules to rename those interfaces to eth0...ethXX

ahh.. looking at it now it looks like the new upstream DTB uses lanX for all "lan" switchports..

Enumaration order and names are specified here:
DSA switch ports: https://github.com/torvalds/linux/blob/master/arch/arm64/boot/dts/mediatek/mt7986a-bananapi-bpi-r3.dts#L408-L459
CPU Ethernet ports: https://github.com/torvalds/linux/blob/master/arch/arm64/boot/dts/mediatek/mt7986a-bananapi-bpi-r3.dts#L171-L198

@Apachez no, they are not mapped by hw-id, all mapping of DSA ports are done by the DTB file (Hardware Device Tree) loaded into the kernel. also the enumeration order of cpu ethX interfaces are specified by the DTB file

Does all the interfaces at bananapi represent a hw-id which can be used to map to the ethX syntax of VyOS?

Hey @fernando - yes, I tested it with two routers in a test environment, with the following setup: https://docs.vyos.io/en/latest/configuration/vpn/site2site_ipsec.html

Do you tested it ? using our current rolling-release

PR https://github.com/vyos/vyos-1x/pull/2500

This is on a virtio interface:

Simple reproducer - doesn't need to be an upgrade, just apply this to 1.4

PR https://github.com/vyos/vyos-1x/pull/2499

PR: https://github.com/vyos/vyos-build/pull/457