Monitoring
Ultralytics Platform provides comprehensive monitoring for deployed endpoints. Track request metrics, view logs, and analyze performance in real-time.
Monitoring Dashboard
Access the global monitoring dashboard from the sidebar:
- Click Monitoring in the sidebar
- View all deployments at a glance
- Click individual endpoints for details
Overview Cards
| Metric | Description |
|---|---|
| Total Requests | Requests across all endpoints (24h) |
| Active Deployments | Currently running endpoints |
| Error Rate | Percentage of failed requests |
| Avg Latency | Mean response time |
Deployments Table
View all deployments with key metrics:
| Column | Description |
|---|---|
| Model | Model name with link |
| Region | Deployed region with flag |
| Status | Running/Stopped indicator |
| Requests | Request count (24h) |
| Latency | P50 response time |
| Errors | Error count (24h) |
| Sparkline | Traffic trend visualization |
Real-Time Updates
The dashboard polls every 30 seconds. Click refresh for immediate updates.
Endpoint Metrics
View detailed metrics for individual endpoints:
- Navigate to your model's Deploy tab
- Click on an endpoint
- View the metrics panel
Available Metrics
| Metric | Description | Unit |
|---|---|---|
| Request Count | Total requests over time | count |
| Request Latency | Response time distribution | ms |
| Error Rate | Failed request percentage | % |
| Instance Count | Active container instances | count |
| CPU Utilization | Processor usage | % |
| Memory Usage | RAM consumption | MB |
Time Ranges
Select time range for metrics:
| Range | Description |
|---|---|
| 1h | Last hour |
| 6h | Last 6 hours |
| 24h | Last 24 hours (default) |
| 7d | Last 7 days |
Metric Charts
Interactive charts show:
- Line graphs for trends over time
- Hover for exact values
- Zoom to analyze specific periods
Logs
View request logs for debugging:
Log Entries
Each log entry shows:
| Field | Description |
|---|---|
| Timestamp | Request time |
| Severity | INFO, WARNING, ERROR |
| Message | Log content |
| Request ID | Unique identifier |
Severity Levels
Filter logs by severity:
| Level | Color | Description |
|---|---|---|
| INFO | Blue | Normal requests |
| WARNING | Yellow | Non-critical issues |
| ERROR | Red | Failed requests |
Log Filtering
Filter logs to find issues:
- Select severity level
- Search by keyword
- Filter by time range
Alerts
Set up alerts for endpoint issues (coming soon):
| Alert Type | Trigger |
|---|---|
| High Error Rate | Error rate > threshold |
| High Latency | P95 latency > threshold |
| No Requests | Zero requests for period |
| Scaling | Instances at max capacity |
Performance Optimization
Use monitoring data to optimize:
High Latency
If latency is too high:
- Check instance count (may need more)
- Verify model size is appropriate
- Consider closer region
- Check image sizes being sent
High Error Rate
If errors are occurring:
- Review error logs for details
- Check request format
- Verify API key is valid
- Check rate limits
Scaling Issues
If hitting capacity:
- Increase max instances
- Set min instances > 0
- Consider multiple regions
- Optimize request batching
Export Data
Export monitoring data for analysis:
- Select time range
- Click Export
- Download CSV file
Export includes:
- Timestamp
- Request count
- Latency metrics
- Error counts
- Instance metrics
FAQ
How long is data retained?
| Data Type | Retention |
|---|---|
| Metrics | 30 days |
| Logs | 7 days |
| Alerts | 90 days |
Can I set up external monitoring?
Yes, endpoint URLs work with external monitoring tools:
- Uptime monitoring (Pingdom, UptimeRobot)
- APM tools (Datadog, New Relic)
- Custom health checks
How accurate are the latency numbers?
Latency metrics measure:
- P50: Median response time
- P95: 95th percentile
- P99: 99th percentile
These represent server-side processing time, not including network latency to your users.
Why are my metrics delayed?
Metrics have a ~2 minute delay due to:
- Metrics aggregation pipeline
- Aggregation windows
- Dashboard caching
For real-time debugging, check logs which are near-instant.
Can I monitor multiple endpoints together?
Yes, the global monitoring dashboard shows all endpoints. Use the table to compare performance across deployments.
