Which alerting signals should a Supabase project route to an external sink before dashboard notifications become noise?

Answer to the Question

For a production Supabase project, only high‑impact, low‑frequency signals should be forwarded to an external sink (Slack, PagerDuty, etc.) before the native dashboard starts to flood you with noise. In practice that means:

• Authentication failures – repeated failed logins, token revocations, or sudden spikes in auth errors.
• Critical database errors – query timeouts, deadlocks, failed migrations, or any error that can corrupt data integrity.
• Storage bucket failures – failed uploads, permission denials, or bucket quota exhaustion.
• Edge function crashes or deployment failures – custom logic that goes down.
• Connection‑pool saturation or high latency – indicates that the Postgres pool is a bottleneck.
• Infrastructure health thresholds – CPU, memory, or network metrics that cross a hard limit.
• Quota thresholds – compute, bandwidth, or database size approaching the plan limit.

Everything else – routine slow queries, minor latency spikes, or normal auth activity – can stay in the Supabase dashboard where you can fine‑tune severity and grouping.

Why This Split Makes Sense

Supabase’s native alerts are easy to enable but offer limited control over severity tiers and deduplication. If you push every event to an external sink you risk:

• Over‑loading your channel with low‑value noise.
• Increasing cost (many plans charge per webhook hit).
• Duplicating alerts that are already visible in the dashboard.

By filtering first – only the high‑impact events – you keep the external sink useful and the dashboard manageable.

Confirmed Facts (Based on Supabase Feature Set)

• Supabase offers per‑alert thresholds in the Project Settings > Alerts page.
• Webhooks and log drains are available on the Pro and Enterprise tiers; the free tier only supports basic email alerts.
• Each alert can be assigned a severity level (Low, Medium, High, Critical) which can be mapped to different external actions.
• The dashboard already aggregates slow‑query statistics via pg_stat_statements and displays them in the “Performance” tab.

Steps to Route Critical Signals

1. Identify the tier – confirm you’re on Pro or Enterprise to use webhooks.
2. Navigate to Project Settings > Alerts – review the list of available alerts.
3. Enable the high‑impact alerts – toggle on Auth failures, DB errors, Storage errors, Edge function failures, and Infrastructure health.
4. Set thresholds – e.g., 5 failed logins in 10 minutes, 3 deadlocks in 5 minutes, or CPU > 80% for 2 minutes.
5. Configure the webhook – paste the URL of your external sink (Slack webhook, PagerDuty routing key, etc.).
6. Map severity to action – set High/Critical alerts to trigger a PagerDuty incident, Medium to a Slack #alerts channel.
7. Test the flow – simulate a token revocation or a failed migration and verify the external sink receives the message.
8. Fine‑tune – monitor the first week, adjust thresholds to eliminate false positives.

What to Keep in the Dashboard

• Slow‑query outliers – you can use Grafana or Supabase’s built‑in charts to drill down.
• Normal auth activity – useful for audit logs but not for immediate alerts.
• General usage metrics – compute, bandwidth, and size – to spot trends.

Missing Diagnostic Detail

Could you confirm which Supabase plan you’re on (Free, Pro, Enterprise)? This determines whether you have access to webhooks and log drains, which are essential for routing the high‑impact alerts described above.