429 Too Many Requests on Realtime connections with Project Alerts
0 reputation · 12 Jul 2022, 19:06 UTC
0 reputation · 12 Jul 2022, 19:06 UTC
Goal is to configure Supabase Project Alerts to signal real capacity issues without generating excessive notification noise, with particular attention to Realtime WebSocket behavior.
Project Alerts support thresholds for API usage, storage and database queries, but the dashboard does not expose fine-grained mute controls per metric. Evaluation is tied to a fixed 5-minute window without custom aggregation, and there is no documented auto-silence after manual acknowledgment. The alerting surface also lacks a REST API for programmatic suppression. The Realtime API documents a per-project connection limit and a 429 Too Many Requests response when the limit is exceeded, without clarifying whether the 429 is per-client or aggregated across all connections.
Is the 429 Too Many Requests response for Realtime emitted per client or aggregated across the project connection pool? Can a specific metric be muted or silenced within Project Alerts while keeping other metrics active? Is there a documented acknowledgment or suppression mechanism that prevents re-triggering on each evaluation window?
35275 reputation · 13 Jul 2022, 04:33 UTC
The 429 Too Many Requests for Supabase Realtime is aggregated across the project‑wide WebSocket connection pool; it is not emitted per individual client.
When the pool reaches its configured concurrent‑connection limit, new handshake attempts receive HTTP 429 while existing clients stay connected, so bursts of 429s appear for later clients until capacity frees.
Do you need to keep the alert active for other metrics while silencing this specific metric?
Use comments to ask for clarification. Post a solution as an answer.
No question comments on this page.