Production Supabase down 12+ hours (us-east-2) and nobody responding: Postgres never restarts after resize, API says ACTIVE_HEALTHY

[jdanfossi]

Our production project swveoevinqgmndqypwfg (us-east-2, Pro) has been completely down since 01:56 UTC on Oct 1, 2026, and two Urgent support tickets (SU-490754, SU-490831) have had no human reply in 12+ hours.

Short version:

• Postgres has not written a single log line since a Nano -> Small compute resize at 02:20 UTC. A restart (02:55) and a second resize to Medium (03:50) did nothing.
• REST/Auth return Cloudflare 522; the pooler returns connection to database not available; Supabase's own mgmt-api gets 522 from the instance.
• The Management API reports the project as ACTIVE_HEALTHY the whole time, so the platform does not seem to see the outage.
• A "Restore to new project" (lsnpjwdvhxtbrhlhiome) has been stuck at COMING_UP for 11+ hours with an empty cluster, created on Postgres image 17.11.0.002 from a physical backup taken on 17.6.1.104.
• Other users report the same symptoms today on Discord (stuck Checking/Unhealthy, relation "realtime.subscription" does not exist). Same pattern as Database unreachable after disk-full — 522 on all connections, won't recover after disk resize + restarts (project qmrbuyicrdafqzbpjzqi, us-east-2) #50009 (us-east-2, closed for inactivity without a fix) and discussion Production Supabase down and nobody responding #43795.

Full details and timeline: #51134

This is a medical clinic's production system in the middle of our biggest campaign of the year. We have stopped touching both projects so as not to interfere. If anyone at Supabase can get this to the infrastructure on-call, we would be extremely grateful.

[burggraf]

I'm escalating this with our engineering team.

[burggraf]

This is resolved -- was a hardware failure, migrating to new hardware fixed it.