Markdown Version | Session Recording

Session Date/Time: 11 Jun 2024 18:00

TOOLS

Summary

This TOOLS Working Group session focused primarily on the ongoing IETF infrastructure transition, particularly the upcoming migration of core services like www.ietf.org, the data tracker, and the mail archive. Significant discussion centered on the timing and impact of the required downtime. The group also discussed methods for gathering broader community feedback on TOOLS priorities and received updates on Medico client enhancements and the deployment of Cloudflare's managed challenge service.

Key Discussion Points

Infrastructure Transition Update

• Mailman 3 Transition Recovery: The team successfully recovered from issues related to messages not being delivered promptly to mailing lists during the Mailman 3 transition in May.
• Application Migrations:
  • The RSC production center's working machine has been moved to Digital Ocean.
  • The older ioc website (now static HTML) is being served directly from Cloudflare R2 blob storage.
• Clustered Application Migration (Primary Focus): Extensive effort is underway to move a cluster of applications that share a common file system, requiring close coordination and near-simultaneous migration. This includes:
  • www.ietf.org (especially non-Wagtail content like exposed file system directories).
  • The data tracker.
  • The mail archive.
  • The IMAP server.
  • The rsync server.
• Staging Progress: A version of www.ietf.org is in staging, with only minor details remaining. The mail archive is approximately 95% complete in staging, with ongoing work on authentication, CORS, and Elastic Search integration.
• Mail Processing Integration:
  • A Mailman archiver plugin (being developed by Steve at Serius) will post messages to the new mail archive's HTTP API.
  • Jennifer has developed a service that accepts messages via LMTP for the data tracker, posting them to its API.
  • Efforts are underway to reduce email ingestion methods for the data tracker, replacing them with API posts where possible, though nits email ingestion will likely continue.
• File System Management Complexity: Untangling decades of varied file serving and duplication methods, previously handled by cron jobs, is a significant effort. The data tracker now writes files to four disk locations, with scripts handling additional synchronizations. This untangling is nearing completion.

Discussion on Core Services Downtime

• Initial Target: The intent was to move the primary cluster of applications (www, data tracker, mail archive) to Digital Ocean early next week (initially Monday).
• Anticipated Downtime: Expected to be measured in hours, not tens of minutes, due to final file system syncs and database migrations.
• Community Notice: Concern was raised about giving only "a few days" notice for significant data tracker downtime.
• Impact on Agenda Work: Liz noted that Monday is a critical day for working on the meeting agenda, which would be affected.
• Deadline Pressure: The end-of-June deadline for moving off IETFA is firm due to budget implications (three months of delays already) and the need to avoid disruptions during IETF 120 (July). Delaying would likely push the move to August.
• ISG/IAB Perspective: Roman and Eric acknowledged the tight constraints, effectively stating that there was no palatable alternative to proceeding with a short-notice downtime, accepting the need for "damage control" regarding community communication.
• Evolving Schedule:
  • Initially discussed moving from Monday to Wednesday to allow more testing time and better notice.
  • Cindy raised concerns about existing interim meetings scheduled for Wednesday that would be dramatically disrupted if data tracker and notes.ietf.org were unavailable (some lasting 7 hours).
  • A sense of those present indicated that moving the downtime to Thursday after the Telchat would be preferable to avoid interim meeting conflicts and provide additional buffer time. The IETF coordination call (which uses Meetecho) could be moved to WebEx for that day.

Community Feedback on TOOLS Priorities

• Roman's Request: Roman asked for a broader way to vet TOOLS priorities and the roadmap beyond current calls and GitHub issues, noting that current feedback mechanisms might represent a narrow slice of the community.
• Canny.io Discussion: Nick demonstrated Canny.io (ietf.canny.io), a service providing an up/downvote system for issues, integrated with GitHub.
• Concerns with Canny.io:
  • Requires creating an account (via Facebook, GitHub, or email).
  • Might still suffer from limited community engagement, similar to calls.
  • Could lead to distorted priorities if only a few individuals use it to upvote pet projects.
• Alternative Suggestion: Russ suggested a mailing list discussion would be superior for gathering qualitative feedback and broader engagement.
• Roadmap Communication: Mixed feedback on the current GitHub card-style roadmap's effectiveness for new users, though some long-time followers found it useful in context.
• IAB Perspective: Miriam noted that figuring out priority and timing of tools development has been challenging, but calls and transparent minutes help.

Other Updates (FYI)

• Medico Client Enhancements: Work is ongoing to add features allowing session chairs/delegates to place disruptive participants into an observer-only mode. These changes will be available by the end of June, with training provided before IETF 120. Permissions will extend beyond just chairs.
• Cloudflare Managed Challenge Service: A captcha replacement and defense against automated abuse has been deployed on www.ietf.org forms. This is expected to significantly reduce hassle for the secretariat.

Decisions and Action Items

• Decision: The main infrastructure transition (involving www.ietf.org, data tracker, mail archive, IMAP/rsync servers) will be scheduled for Thursday after the upcoming Telchat (targeting the week following this meeting). This revised timing is to allow for further testing, provide more community notice, and avoid conflicts with scheduled interim meetings.
• Action Item: The team will provide an estimated downtime duration and issue a community announcement today (or as soon as confident in the Thursday target).
• Action Item: Cindy will coordinate moving the IETF Coordination Call scheduled for that Thursday from Meetecho to WebEx.
• Action Item: Further discussion regarding methods for gathering broader community feedback on TOOLS priorities (e.g., Canny.io, mailing list) will be moved to the mailing list.

Next Steps

• Complete testing for the main infrastructure transition throughout the remainder of this week.
• Execute the main infrastructure transition on the agreed-upon Thursday.
• The mail processing system is expected to be moved in the week following the main cluster migration (last week of June).
• The registration system will be temporarily moved to a Digital Ocean droplet, followed by conversion to run in the Kubernetes cluster.
• Develop and roll out training for the new Medico client features for IETF 120.
• Continue the discussion on tools priority feedback mechanisms on the mailing list.
• The next TOOLS session is anticipated in July, with a plan to meet in person at IETF 120 in Vancouver.