Session Date/Time: 24 Jul 2025 07:30

happy

Summary

The HAPPY working group meeting at IETF 123 focused primarily on resolving open issues against the main Happy Eyeballs document, with significant discussion around asynchronous DNS resolution, address sorting, IPv6-mostly networks, ECH prioritization, and downgrade attacks. The meeting also included an overview of Happy Eyeballs testing and a discussion about reporting network errors.

Key Discussion Points

Asynchronous DNS Resolution: Extensive discussion around handling multiple record types (A, AAAA, SVCB, HTTPS), determining when to move on from the resolution step, and defining "connectivity" in the context of IPv4/IPv6 networks and VPNs.
Address Sorting: Explored the different sources of preferences (client, SVCB, application requirements) and the proposed three-level sorting approach based on application/protocol/security requirements, service priority, and destination address preference. Debate around prioritizing ECH versus server-provided priorities.
IPv6-Mostly Networks: Discussion regarding handling synthesized IPv6 addresses and native IPv4 paths in IPv6-mostly networks, and determining if the implementation knows if an address is synthesized or not
ECH Prioritization: Debate about whether clients should override server-provided priorities to prefer ECH, considering the collaborative nature of connection security.
Downgrade Attacks: Explored potential downgrade attacks in secure DNS scenarios where attackers could drop SVCB/HTTPS information, preventing ECH/TLS usage. The proposal is that we might need to block until we receive the SVCB/HTTPS record. Mentioned attacks could happen over DOH.
Retransmission vs. Next Connection Attempt: Discussion around the timing of retransmissions versus starting the next connection attempt, acknowledging the difficulty of precise coordination.
Service Priorities: Consideration of whether to raise connection attempts across multiple service priorities or wait until all higher priority endpoints fail, favoring the current approach of allowing lower priority attempts to proceed.
Target Fixation: Discussed how prefering previously used addresses causes stickiness.
Happy Eyeballs Testing: Dave Planck presented research and tooling for testing Happy Eyeballs implementations, including a website that allows users to test their own browsers.
Reporting Network Errors: Discussed the scope, target, and potential privacy implications of reporting network errors related to Happy Eyeballs, with different perspectives on where the reporting should be implemented (browser, OS, etc.).

Decisions and Action Items

Written Proposal on Connectivity: A written proposal defining "connectivity" in the context of IPv4/IPv6 networks and VPNs is needed.
Issue or PR for Clarification: An issue or PR should be created to clarify that moving on from the DNS resolution step doesn't mean abandoning less preferred options that may arrive later.
Issue/PR for Alias Mode: File and Issue/PR for what to do if the only record returned is an SVCB Alias Record.
Update Text for Hard ECH requirement: File a PR to update the text to reflect what to do if there is a hard ECH requirement.
New Issue: Lorenzo to file a new issue regarding if we can do anything to try to help SVCB queries survive attack scenarios.
Discuss more on the List: There will be additional discussion and collaboration on the HAPPY list.

Next Steps

Interim Meeting: The working group will schedule an interim meeting to close on some of the open issues related to the main Happy Eyeballs document. The group is planning two interims, one for the main document and another for the reporting discussions.
Test Tooling Deployment: Dave Planck will work to deploy a Happy Eyeballs test site in North America and encourages participation in the Montreal hackathon.
Network Error Reporting Discussion: Continue discussion on reporting network errors on the mailing list to discuss specific breakage and what issues the draft is attempting to fix.