Investigate persistent PostgREST 503 PGRST002 on Mars Data API
PostgREST Data API returning 503 PGRST002 (schema-cache introspection fault) on every /rest/v1/* GET across the visible Supabase log window (hours), predates MARS-416 migration and is unrelated to it. /auth/v1/* healthy throughout (200s from real production traffic), authenticator role has live idle DB connections — reads as a genuine sustained fault, not a deliberate Data-API-disabled toggle (disabled would 404 at gateway, not proxy to this PostgREST error). No Management-API/dashboard access from agent side to confirm the Data API on/off setting directly — needs Supabase Studio > Settings > Data API check to settle definitively. Flagged by db-mars-cc during MARS-416 closure verification; audit-mars-ca called it a separate incident, not evidence either way about the grant fix.
Questions
Activity
-
Resolved: the 503 PGRST002 traffic was self-generated — every /rest/v1/* hit in the visible log window carries UA evolutiva-internal-probe/mars-ptd (db-mars-cc's own MARS-416 verification probing), zero external/unknown callers. Data API is confirmed deliberately disabled fleet-wide (Elazar) and unused — 503 is expected behavior on a disabled Data API being probed, not a fault. No incident.