Re: happy eyesockets
Andy Ruhl <[email protected]>
| Newsgroups | gmane.os.netbsd.devel.network |
|---|---|
| Message-ID | <CAJcb3fpFyTGs+FkWzANNnqR+DEhyhe5S+4+MWZggzY9Bcd6obQ@mail.gmail.com> |
On Sat, Mar 22, 2025 at 5:10 PM Jonathan A. Kollasch <[email protected]> wrote: > > Attached is a patch to actually treat ICMP(v6) host/net/port unreachables > during TCP establishment in a way that doesn't cause ~75 second delays > waiting for multiple retransmits to all be reported as unreachable before > actually letting getaddrinfo(3) fall back to the next address. > > Linux seems to handle this like this already. > > Commitable? Just rambling here. It's been some years since I studied "happy eyeballs". There appears to be a newer-ish RFC on it: https://datatracker.ietf.org/doc/html/rfc8305 Apparently OpenBSD is doing something around it: https://undeadly.org/cgi?action=article;sid=20191128040647 It would be nice to have a variously broken host or hosts to test against for multiple failure scenarios... Maybe it exists. It would also be nice to have some user adjustability around this. Many IPv6 hosts have fewer hops than IPv4 for me, so I'm willing to be a little more patient for IPv6. I might be able to help with some testing. Andy