Managing Collectors and Sources
Monitor collector health, configure data sources, trigger manual collections, and understand the circuit breaker pattern that keeps discovery resilient
Collectors are the engines that discover your infrastructure. Each collector connects to a specific platform (Kubernetes, Proxmox, Ceph, etc.) and periodically scans it to discover configuration items. The collector management page lets you monitor health, configure sources, and trigger manual collections.
Viewing Collectors
Navigate to Settings --> Collectors to see all registered collectors.
Each collector card shows:
| Field | Description |
|---|---|
| Name | The collector identifier (e.g., k8s-collector, ceph-collector) |
| Status | Whether the collector is running and healthy |
| Sources | How many data sources this collector manages |
| Last Collection | When the collector last completed a collection cycle |
| CI Count | Number of CIs discovered by this collector |
Understanding Sources
A single collector can manage multiple instances of its target system. For example, the Kubernetes collector might monitor three clusters, each configured as a separate "source."
Each source has:
| Field | Description |
|---|---|
| Name/Label | Friendly identifier for this instance |
| Source Label | Identifies which instance of the source system this represents (e.g., production-cluster, eu-west-ceph) |
| Health State | Current health status (see Health States below) |
| Enabled/Disabled | Whether collection runs for this source |
Health States
Collectors and their sources use a circuit breaker pattern for resilience. When a source starts failing, the system progressively backs off rather than hammering an unreachable target.
| State | Meaning | Action Needed |
|---|---|---|
| Healthy | Normal operation, collections succeeding | None |
| Degraded | Collections succeeding but slow (latency greater than 10s) | Monitor -- may indicate target system under load |
| Unhealthy | Recent failures detected, still retrying | Check target system connectivity and credentials |
| Circuit Open | 3 consecutive failures, skipping collections for 60s | Investigate target system -- likely down or unreachable |
Triggering Manual Collection
Click the Collect Now button on a collector to trigger an immediate collection cycle outside the normal schedule. This is useful for:
- Verifying a newly configured source works
- Getting fresh data before an investigation
- Testing connectivity after credential changes
Collection Modes
| Mode | Description |
|---|---|
| Periodic | Collects on a fixed interval (e.g., every 60 seconds) |
| Watch | Listens for real-time events (Kubernetes only) |
Most collectors use periodic mode. The Kubernetes collector supports both -- periodic collection for full reconciliation and watch mode for immediate change detection between collection intervals.
Source Enable/Disable
You can temporarily disable collection for a specific source without removing its configuration. Disabled sources:
- Are skipped during collection cycles
- Retain their configuration for re-enabling later
- Show as "Disabled" in the health status
Monitoring Collector Health
The collector health page at Settings --> Collectors provides:
- Real-time health state for each collector and source
- Collection timing (last run, next scheduled)
- Error counts and recent failure messages
- Circuit breaker state transitions
Related Documentation
- Collector Reference: Infrastructure -- What each infrastructure collector discovers
- Collector Reference: Platform Services -- Platform and service collectors
- Collector Reference: Discovery -- OS and SNMP discovery collectors
- Managing Credentials -- Credential setup for collectors