I use bgp confederation. The problem with BGP med is that the priority of as path len is higher than BGP med, which means some upstream with longer as path would never be selected. Even though it has the best ping. Douglas Fischer <fischerdouglas@gmail.com> 于 2023年10月10日周二 上午2:59写道：

It makes a lot sense! On a scenario of an Off-Path RR, aigp works perfectly.

But... You are probably using this to ebgp, right? Why not set med as aigp, and then test med?

Em seg., 9 de out. de 2023 às 15:34, Brandon Zhi <Brandon@huize.asia> escreveu:

...

Hi all,

I plan to use bgp_aigp instead of bgp_med to find the optimal route based on incremental pings per node. For example, if the delay between one node and another node is 40, then bgp_aigp increases by 40 after passing this ibgp.

void bgp_aigp [O] This attribute contains accumulated IGP metric, which is a total distance to the destination through multiple autonomous systems. Currently, the attribute is not accessible from filters.

But it seems I can't modify this value?

Best Regard, *Brandon Zhi* HUIZE LTD www.huize.asia <https://huize.asia/>| www.ixp.su | Twitter

This e-mail and any attachments or any reproduction of this e-mail in whatever manner are confidential and for the use of the addressee(s) only. HUIZE LTD can’t take any liability and guarantee of the text of the email message and virus.

-- Douglas Fernando Fischer Engº de Controle e Automação

Hi Ondrej, as.path.len = 2 as.path.len = 1 Upstream1 ---- Node1 ---(10ms) Node2 --(20ms)-- Upstream2 | Node3 For this network situation, I need to have node 3 go to node 2 and connect to the internet through upstream 1. Even if its as.path is longer. Therefore, a parameter with a priority greater than as.path.len is required to achieve this. Best, *Brandon Zhi* HUIZE LTD www.huize.asia <https://huize.asia/>| www.ixp.su | Twitter This e-mail and any attachments or any reproduction of this e-mail in whatever manner are confidential and for the use of the addressee(s) only. HUIZE LTD can’t take any liability and guarantee of the text of the email message and virus. On Tue, 10 Oct 2023 at 21:43, Ondrej Zajicek <santiago@crfreenet.org> wrote:

On Mon, Oct 09, 2023 at 11:29:33PM +0800, Brandon Zhi wrote:

...

Hi all,

I plan to use bgp_aigp instead of bgp_med to find the optimal route based on incremental pings per node. For example, if the delay between one node and another node is 40, then bgp_aigp increases by 40 after passing this ibgp.

void bgp_aigp [O] This attribute contains accumulated IGP metric, which is a total distance to the destination through multiple autonomous systems. Currently, the attribute is not accessible from filters.

But it seems I can't modify this value?

Hi

You cannot modify bgp_aigp - total distance, but you can modify per-hop distance, which is added to it automatically in each (inter-AS) step.

It is either regular IGP metric (e.g. the attribute 'ospf_metric1' in case of OSPF), or manually configured one (the attribute 'igp_metric'), or configured cost of inter-AS link [0].

[0] https://bird.network.cz/?get_doc&v=20&f=bird-6.html#bgp-cost

I think that using bgp_aigp is the right tool for your use case, but details depend on specific topology of your confederation - whether each internal AS in the confederation is nontrivial network with IGP, flat network, or just one router.

-- Elen sila lumenn' omentielvo

Ondrej 'Santiago' Zajicek (email: santiago@crfreenet.org) OpenPGP encrypted e-mails preferred (KeyID 0x11DEADC3, wwwkeys.pgp.net) "To err is human -- to blame it on a computer is even more so."