Supabase's Management API started overwriting Postgres config changes with only the latest value sent, losing past updates in the process. The fault sat in Supabase's own API layer and lasted 2h 11m before a fix went out, with nothing for you to have done differently on your end.
Started
04:56 UTC
Duration
Lasted 2h 11m
Source
IsDown
Next time
Supabase is degraded. Is your product working?
An incident on their side does not always mean an outage on yours, and the only way to know is to already be watching your own endpoints. Set that up in under a minute and you will be notified the next time your own site is affected. And there will be a next time.
50 monitors for your own websites. No card.
What happened?
Supabase's status page opened the incident at 04:56 UTC, reporting that Postgres Config changes made through the Management API or the Database Settings Dashboard were only saving the latest values sent, with past changes being overwritten rather than retained. Within the first update, Supabase told users to refrain from making Postgres Config changes via either route while they worked on it. By 05:16 UTC they had gone further and disabled certain Postgres config changes on the dashboard outright, rather than let the bug keep silently dropping settings. At 06:17 UTC Supabase said they had identified the cause and were deploying a fix, with the disabled dashboard setting to be re-enabled shortly after. At 06:24 UTC that fix landed and the specific setting named, Connection Logging, was confirmed available again. The incident closed at 07:04 UTC with a note that Connection Logging was working fine for all customers. No components are listed as affected on Supabase's status page for this incident, and no user reports were filed during the window. The whole episode ran its course in under two and a half hours, with five updates published across that span. Supabase's post did not name the underlying cause beyond "identified," so what specifically triggered the overwrite behaviour in the Management API is not on the record. This was Supabase's 52nd incident in the last 90 days, with a median duration of 2 hours 5 minutes, so this one tracked close to typical for them.
Learning
This incident lived entirely inside Supabase's Management API, a control-plane layer that governs configuration rather than your database's live queries or connections. A standard uptime check against your own endpoints would not have caught this, because your database kept running and serving traffic throughout: the damage was to settings being silently overwritten, not to availability. That is the general shape of this kind of failure, it is invisible from outside unless something is specifically watching the provider's control-plane status, because your own health checks are built to answer "is my app up," not "is my provider's config API behaving correctly." UptimeRobot's 30-day data for Supabase shows 19 incidents, all resolved, with a median duration of 2h 11m, matching almost exactly what happened here. Supabase's own status page is the only source for what went wrong and when, since nothing about this surfaces in typical endpoint monitoring. Your monitoring covers your half, UptimeRobot now watches the provider's half too, so next time you know which side broke without guessing.