[Babel-users] not routing

[Juliusz Chroboczek]

> What if you used the logic to check only in the table babel was
> started with (the -t arg)?  For example, using -t 42, If a viable
> route exists in table 42, don't create another default route, however,
> if there is no default route in table 42, it is ok to add one.

Good idea.

It's a little more work than that, though -- there's at least
flushupdates and handle_update that currently assume that for a given
destination, there is either an external route or a Babel route, but not
both.

I'll see what I can do.

> Here, if we look at this another way, babeld is being handed a route
> which it is throwing away (or not using) because it thinks there is
> already a viable route.  Shouldn't this decision be left to the
> routing policy db?

The external route is not under Babel's control -- it's there whether
Babel agrees or not.  So Babel's route selection policy has nothing to
do with whether that route is used.

The proper solution would be to have whoever installed the external
route remove it when Babel routes are available.

> What would happen if you put all routes that came in into the routing
> table except those which cause a routing loop?

The kernel would fail with EEXIST.

                                        Juliusz