[prev in list] [next in list] [prev in thread] [next in thread] 

List:       bird-users
Subject:    Re: [PATCH] [RFC] Babel: Replace internal route selection by bird nest
From:       dxld () darkboxed ! org
Date:       2023-01-30 21:31:17
Message-ID: 20230130213117.trvx67rrgyr6i77a () House ! clients ! dxld ! at
[Download RAW message or body]

Hi Ondrej,

On Mon, Jan 30, 2023 at 03:52:51PM +0100, Ondrej Zajicek wrote:
> I was confused before, but is this really a breaking change? As i
> understand it, both Babel and RIP keep list of received routes, select
> the best, push the best to main (nest) table, but it must be explicitly
> exported back to the protocol in order to be announced to neighbors.

As was I it seems. I didn't look at the update_send code that closely and
must have just started assuming *selected would be announced somewhere
along the way.

> There is babel_select_route() -> babel_announce_rte() -> nest table
> change -> babel_rt_notify() -> babel_trigger_update() cascade,
> babel_entry.selected is the best route as selected by Babel, but it
> is babel_entry.seqno/metric/router_id that is filled in babel_rt_notify()
> and used for sending updates in babel_send_update_().

Right I agree.

In that case can you think of any way suddenly seeing more routes can break
a filter policy? I don't think filters can keep state across individual
routes, right?

I just fixed the unreachable route announcement too I'll send and updated
series.

--Daniel
[prev in list] [next in list] [prev in thread] [next in thread]