Prometheus#
How it works#
Unryo has the ability to treat your Prometheus server(s) as datasources, querying them to discover dependencies and surface problems.
Telemetry data stays in your data source(s) and is not duplicated in Unryo (except if you want to). By doing so, Unryo acts smoothly as a correlation and consolidation layer (MoM approach) on top of your monitoring stack.
Connecting your Prometheus#
You can add one or multiple Prometheuses as data sources, so that Unryo starts correlating and provide insights to your operation teams.
To add a Prometheus server as a data source, follow these steps:
- Go in Configuration Management > Connectors
- Click +
- Select the "Prometheus connector" template
- Enter the necessary settings (note: Unryo stores safely configurations and credentials encrypted in a secure vault stored on-premise in your infrastructure).
The connector configuration tells Unryo how to a connect to a specific Prometheus instance. Furthermore, it allows you to activate or deactivate Prometheus pollings provided by Unryo, such as the "alerts" polling, which retrieves alerts from the Prometheus AlertManager. Lastly, it allows you to define your pollings, allowing you to write custom queries and processing.
Built-in Configurations#
Once Unryo has access to your Prometheus, it looks for key data:
- metric metadata, to get context and find dependencies
- metric values, to spot anomalies
- events when available and configured
This is done via configurations that inform Unryo about how your data is structured, so that it knows how to query key data and establish dependencies.
The following configurations are enabled by default:
| Configuration | Comments |
|---|---|
| Kubernetes Topology | Inspects metric metadata to discover all Kubernetes resources (Pods, Nodes, Services, …) to determine their dependencies and status. These configurations are responsible for building and maintaining the topology model that feeds the correlation engine. |
| Node Exporter | Hardware and OS metrics from Linux servers. |
| Windows Exporter | Hardware and OS metrics from Windows servers. |
| Prometheus HealthCheck | Monitor Prometheus reachability and global health. |
| Istio | Discover traffic between services: request count, errors, latency. |
Custom Configurations#
If you have custom metrics, or if you use a Prometheus exporter that is not supported, you can add your own topology configuration.