Thanks, Tore! I would like to add that I'm not averse to the idea of coding the LOCAL_OUT trick, and in fact might be inevitable once Jool is merged into nftables. Or impossible; I haven't really analyzed it.
Just a bit of patience, please. I'm hoping to finish MAP-T this year. On Tue, Jun 30, 2020 at 12:35 AM Tore Anderson <[email protected]> wrote: > > * Nico Schottelius > > > You might be right. The use case I see is: > > > > - computer/system x has legacy apps installed > > - computer/system x just installs jool and maps v4 addresses into > > port range > > - legacy apps now can run on IPv6 only hosts > > > > So very similar operation to https://github.com/toreanderson/clatd. > > Which, for no surprise, currently relies on tayga. > > clatd predates Jool, or at least the necessary SIIT bits within Jool. > If I was to re-implement clatd today, I would do it using Jool as the > translation engine instead of TAYGA. As Alberto has pointed out > already, you can do this with Jool no problem. > > All you need to do is to enclose Jool in a private network namespace. > Then you point the IPv4 default route to the veth device connecting to > the Jool network namespace (in exactly the same manner that clatd > points the IPv4 default route to the tun device connecting to the TAYGA > process). > > https://jool.mx/en/node-based-translation.html > https://github.com/NICMx/Jool/issues/177#issuecomment-144648229 > > Tore > _______________________________________________ Jool-list mailing list [email protected] https://mail-lists.nic.mx/listas/listinfo/jool-list
