text
stringlengths
0
59.1k
| ------------------- | ------ | ----------------- | ----------------------------------------------- |
| Status | select | eq, neq | Trace status: `error`, `success`, `in_progress` |
| Latency (ms) | number | gt, lt | Trace duration in milliseconds |
| Model | text | eq, neq | LLM model name |
| User ID | text | eq, neq | User identifier from trace |
| Input | text | contains | Trace input content |
| Output | text | contains | Trace output content |
| Error Message | text | eq, contains | Error message from failed traces |
| Agent/Workflow Name | text | eq, neq, contains | Name of the root span |
| Entity Type | select | eq, neq | `agent` or `workflow` |
| Metadata | text | eq, neq, contains | Custom metadata key-value pairs |
### Filter Operators
| Operator | Description |
| ---------- | ------------------------------------- |
| `eq` | Equals |
| `neq` | Not equals |
| `gt` | Greater than |
| `lt` | Less than |
| `contains` | Contains substring (case-insensitive) |
### Metadata Filters
Filter by custom metadata fields using dot notation:
- `metadata.environment` - Direct metadata key
- `metadata.context.region` - Nested context key
## Notification Channels
### Webhook
Send HTTP POST requests to any URL when an alert triggers.
**Configuration:**
- **URL**: Endpoint to receive the webhook
- **Headers**: Optional HTTP headers (JSON format)
- **Body**: Optional custom payload (JSON format)
**Default Payload:**
```json
{
"alert_id": "uuid",
"alert_name": "High Error Rate",
"metric": "error_rate",
"value": 15.5,
"threshold": 10,
"timestamp": "2024-01-15T10:30:00Z"
}
```
### Slack
Send notifications to a Slack channel using Incoming Webhooks.
**Setup:**
1. Create an Incoming Webhook in your Slack workspace ([Slack documentation](https://docs.slack.dev/messaging/sending-messages-using-incoming-webhooks/))
2. Copy the webhook URL (format: `https://hooks.slack.com/services/...`)
3. Paste the URL in the Slack channel configuration
Slack notifications include:
- Alert name and metric
- Current value and threshold
- Link to view the incident
- Link to view a sample trace (when available)
### Testing Notifications
Click "Send Test Notification" to verify your channel configuration. Test webhooks include an `is_test: true` field.
## Incidents
When an alert triggers, VoltOps creates an incident. Incidents track the lifecycle of an alert from trigger to resolution.
### Incident Statuses
| Status | Description |
| ------------ | ---------------------------------------- |
| Open | Alert triggered, awaiting response |
| Acknowledged | Team member is investigating |
| Snoozed | Temporarily muted until a specified time |
| Resolved | Issue addressed, incident closed |
### Incident Workflow
1. Alert triggers → Incident created with status `open`
2. Team member takes ownership → Status changes to `acknowledged`
3. Investigation complete → Status changes to `resolved`
Alternatively:
- **Snooze**: Temporarily silence the incident. When the snooze period expires, the incident reopens if conditions still trigger.
### Incident Details