Monitoring & Logs

Effective monitoring helps you understand workflow behavior, identify issues, and optimize performance. This guide covers the monitoring capabilities available in Agents.

Execution Status

Status Lifecycle

Every workflow execution progresses through these states:

                 ┌→ completed
queued → running ┼→ failed
                 └→ canceled

Status Definitions

Status Timestamps

Each execution tracks key timestamps:

Viewing Executions

Execution List

View all executions with filtering:

Available Filters:

• Status (queued, running, completed, failed, canceled)
• Workflow
• Date range
• Trigger type

Execution Detail

Each execution shows:

1. Summary - Status, duration, trigger info
2. Timeline - Visual execution flow
3. Node States - Status of each node
4. Inputs/Outputs - Data at each step
5. Errors - Any error details

Node-Level Monitoring

Node States

Each node in an execution has its own state:

Node Execution Details

For each node, view:

• Input Data - What data the node received
• Output Data - What data the node produced
• Duration - How long execution took
• Attempts - Retry count if applicable
• Error - Error message if failed

Port State Tracking

For complex nodes with multiple ports:

Input Ports:
├── in[0]: 3 items ready
└── in[1]: 2 items ready

Output Ports:
├── success[0]: 5 items produced
└── error[0]: 0 items

Execution Logs

Log Levels

Log Structure

Each log entry includes:

{
  "timestamp": "2024-01-15T10:30:45.123Z",
  "level": "info",
  "message": "Node execution completed",
  "context": {
    "executionId": "exec-123",
    "nodeId": "node-456",
    "duration": 1250
  }
}

Viewing Logs

Access logs through:

1. Execution Detail - Logs for specific execution
2. Workflow Logs - All logs for a workflow
3. System Logs - Platform-wide logs

Monitoring Patterns

1. Execution Success Rate

Track workflow reliability:

Success Rate = Completed / (Completed + Failed) × 100

Healthy: > 95% Warning: 90-95% Critical: < 90%

2. Execution Duration

Monitor performance over time:

3. Queue Depth

Track pending executions:

Healthy: < 10 queued Warning: 10-50 queued Critical: > 50 queued (processing backlog)

4. Error Patterns

Identify recurring issues:

• Same error message across executions
• Errors at specific times
• Errors for specific trigger types

Error Tracking

Error Information

When a node fails, capture:

Error Categories

Error Notifications

Set up alerts for:

1. Execution Failures - Immediate notification
2. Error Rate Threshold - Alert when rate exceeds limit
3. DLQ Entries - New messages in dead letter queue

Performance Monitoring

Key Metrics

Identifying Bottlenecks

Look for:

1. Slow Nodes - Nodes taking longer than expected
2. High Retry Counts - Nodes requiring multiple attempts
3. Queue Buildup - Executions waiting to start

Optimization Opportunities

Based on monitoring data:

Dashboard Views

Workflow Dashboard

Overview of all workflow activity:

• Total executions (24h, 7d, 30d)
• Success/failure breakdown
• Average duration
• Recent executions

Execution Timeline

Visual representation of execution flow:

[Trigger] → [Load] → [AI] → [Send]
   0s        1.2s    4.5s    5.1s

Error Summary

Aggregated error information:

• Error counts by type
• Most common errors
• Error trends over time

Best Practices

1. Monitor Proactively

Don't wait for user reports:

• Check dashboards regularly
• Set up automated alerts
• Review DLQ daily

2. Track Baselines

Know what "normal" looks like:

• Typical execution duration
• Expected success rate
• Normal queue depth

3. Investigate Anomalies

When metrics deviate:

1. Check for system issues
2. Review recent changes
3. Examine error patterns

4. Document Incidents

Keep records of:

• What happened
• Root cause
• Resolution
• Prevention steps

5. Use Log Filters Effectively

Filter logs to find issues quickly:

• Filter by execution ID for specific issues
• Filter by error level for all errors
• Filter by time range for incidents

Troubleshooting with Monitoring

Execution Not Starting

Check:

• Is workflow released and active?
• Is trigger configured correctly?
• Any queued executions?

Execution Stuck in Running

Check:

• Which node is executing?
• Has it exceeded timeout?
• External service issues?

High Failure Rate

Check:

• Common error message?
• Specific node failing?
• Integration issues?

Slow Executions

Check:

• Which nodes are slow?
• AI model tier?
• External API latency?

Related

• Testing & Debugging
• Error Handling
• Dead Letter Queue
• Performance Optimization