Monitoring production flows
Monitor sessions, webhooks, retries, reconciliation, and support signals for live 0Bit integrations.
Production monitoring should tell you whether customers can start flows, complete hosted actions, receive backend events, and see accurate status after completion or failure.
Monitor your contract, not internal systems
Public integration monitoring should focus on partner-visible outcomes: API responses, session states, webhook delivery, event processing, reconciliation, and support identifiers.
Monitoring map
Metrics to track
| Signal | Alert when |
|---|---|
| Product-object creation failures | Error rate or 5xx rate rises. |
| Origin or credential errors | Any production spike appears. |
| Widget unavailable callbacks | Above normal baseline. |
| Webhook dead letters | Any sustained or repeated delivery failure. |
| Event processing lag | Backlog grows or terminal events are not processed. |
| Duplicate fulfillment blocked | Non-zero occurrences need review. |
| Reconciliation backlog | Attempts remain unresolved beyond your policy. |
| Support escalations | Multiple users hit the same status or request id family. |
Event timeline
type FlowTimeline = {
attemptId: string;
productObjectId: string;
createdAt: string;
browserSignals: Array<{ type: string; at: string }>;
webhookEvents: Array<{ eventId: string; type: string; at: string }>;
finalStatus: 'processing' | 'fulfilled' | 'failed' | 'expired' | 'cancelled' | 'manual_review';
};Guardrails
- Alert on symptoms partners can act on.
- Keep dashboards redacted by default.
- Track event ids and request ids for support.
- Separate sandbox noise from production alerts.
- Do not expose provider, treasury, reserve, or internal worker details in public dashboards.