Discord Incident

· systems · Source ↗

Discord’s API systems went down May 8, 2026 (~12:08–13:19 PDT), blocking logins and message sending globally before traffic was metered back in during recovery.

What Matters

  • Outage window: ~71 minutes from first investigation (12:08 PDT) to declared recovery (13:19 PDT).
  • Root cause identified by 12:24 PDT; Discord cited session-start failures affecting broad service availability.
  • Recovery used traffic metering on reconnect, suggesting load-shedding to prevent thundering-herd collapse.
  • [HN: @coreylane] A concurrent AWS outage (single-AZ failure) went largely unreported the same day.
  • [HN: @jeffwask] Discord’s single-AZ sensitivity surprised commenters: “Kinda surprised those guys are built in a way where a single AZ failure takes them down.”
  • [HN: @rweichler] Developer mid-bug-report when Discord dropped, underscoring how community infrastructure with no async fallback (email, mailing list) creates hard blockers.

Original | Discuss on HN