Monitoring & troubleshooting

Day-two operations: where to look when something fails or goes quiet. Complements Production readiness.

What execution logs exist

• Per agent: GET /api/v1/agents/:id/executions, .../logs, .../metrics — timestamps, status, payloads (as exposed by your tenant).
• Workflows: use test mode for dry runs; production behavior surfaces through downstream effects and any logging steps you added.
• Webhooks: GET /api/v1/webhooks/:id/deliveries for outbound attempt history.

How to inspect failed runs

1. Identify whether the failure is agent, workflow trigger, or integration/webhook.
2. Open the agent execution row; read error text and correlation identifiers.
3. For outbound webhooks, check HTTP status from the delivery log and fix the receiver.
4. Replay with care: use idempotency_key for agent calls so retries stay safe.

Metrics exposed in the product

Dashboard monitoring plus API routes under /api/v1/monitoring/* and agent-level .../metrics (counts, timing where available). Interpret alongside your own APM.

Recommended alerts

• Spike in failed agent executions vs. baseline.
• Webhook delivery failure rate above threshold.
• Scheduled workflow produces no runs in an expected window (missed schedule or disabled flow).
• Health checks: GET /health for load balancer or uptime monitors.

Webhook verification sequence (outbound)

  DripPulse                    Your endpoint
      |                              |
      |-------- signed POST -------->|
      |                              | verify signature / secret
      |<------- 2xx ack -------------|
      |                              |
      | retry on transient errors    | idempotent handler

Agent execution flow (simplified)

  Request --> Auth / org scope
                |
                v
           Load agent policy
                |
                v
           Run execution (record idempotency_key)
                |
                v
           Persist result + metrics

Support: support@drippulse.io