Monitoring Qdrant Cloud Clusters
Qdrant Cloud gives you three ways to watch a cluster:
| Where | What you get | Use it for |
|---|---|---|
| Cloud Console | Metrics, requests, and logs on the Cluster Details page | A quick look at a cluster’s health |
| Email alerts | Automatic alerts to the account owner | Being told about capacity, health, and version problems |
| Metrics endpoints | Prometheus-compatible /metrics and /sys_metrics, plus /telemetry | Your own dashboards and alerting |
Telemetry
The Metrics and Requests sections of the Cluster Details page show the health of the database cluster in the Qdrant Cloud Console.
Logs
The Logs section of the Cluster Details page shows the logs of the database cluster. For a record of API calls, turn on audit logging.
Alerts
The account owner receives automatic emails when a cluster has a capacity, health, version, key, or load problem. Examples are memory or disk use above 80% for more than 5 minutes, a node that ran out of memory, and a Database API key that is about to expire. The Cluster Alerts reference lists every alert with its trigger, meaning, and fix. To read alerts from automation, use qcloud cluster alert list in the Cloud CLI or the ListClusterAlerts method of the Cloud API.
Qdrant Database Metrics and Telemetry
You can also read the metrics and telemetry that the database nodes provide. Every request needs a Database API key. A read-only key is enough. Send it in the api-key header, or as a Bearer token, which some monitoring providers require.
| Endpoint | Where to call it | What it reports |
|---|---|---|
/metrics | Each node endpoint, so that you scrape every node | Database metrics for that node only, in Prometheus format. See Qdrant monitoring |
/sys_metrics | The cluster endpoint, once per cluster | The database metrics from all nodes, plus infrastructure data such as load balancer traffic and container CPU and memory. Managed Cloud only. See the metrics reference |
/telemetry | The cluster endpoint | The current state of the database, including the number of vectors and shards. See the API reference |
Scraping /sys_metrics once is enough for a cluster-wide view. Scraping it on every node returns the same data each time.
Grafana Dashboard
If you scrape your cluster metrics into your own monitoring system and use Grafana, the Qdrant Cloud Grafana dashboard visualizes them. For a full Prometheus and Grafana setup, see Scrape Metrics with Prometheus and Grafana.
Ready to try Qdrant?