Platform
Alerts & ops
Built-in detectors that watch your streams 24/7, with Slack routing when something breaks.
Built-in detectors
| Detector | Default threshold | Severity |
|---|---|---|
| Rebuffer ratio spike | > 3% sustained 5 min | critical |
| Startup time regression | p95 > 3s sustained 10 min | warn |
| Error rate spike | > 2% sustained 5 min | critical |
| Ingest quota 80% | monthly | info |
| Ingest quota exceeded | monthly | warn |
| Payment gateway p95 | > 1s sustained 5 min | warn |
| Churn spike | > 2× 7d baseline | warn |
| Content completion drop | > 20% vs 7d baseline | info |
| New release detected | first play_start for contentId | info |
Custom alert rules
Go to Alerts → Rules → New rule. Pick a metric, a condition, a time window, and a severity. Optional: scope to a specific region, device, or content id.
Routing
Each severity can be routed to different destinations. Common setup:
- critical → PagerDuty + #incidents Slack channel
- warn → #ops Slack channel
- info → admin panel only
Slack format
The Slack adapter renders alerts as attachment-formatted messages with severity color, metric values, and a deep link back to the admin panel.
[CRITICAL] Rebuffer spike on Mumbai edge
Region: Maharashtra (ap-south-1)
Rebuffer: 3.1% -> 8.4%
Affected: 24,000 sessions
[View in dashboard]Acknowledging and snoozing
Acknowledge an alert to remove it from the realtime ops feed. Snooze mutes all firings of the same rule for a period (15 min, 1h, or custom). Snoozes auto-expire — you can’t accidentally mute a rule forever.
Incident runbooks
Attach a runbook URL to each rule. The URL appears in the Slack message and on the alert detail page, so the responder always has the playbook one click away.