Edge Node Observability Stack#
The Edge Node Observability stack is intended for Edge & Host Managers and Host Operators to monitor the following aspects of Edge Nodes:
Hosts (Edge Node) - telemetry from the hosts of the Edge Node (including provisioning).
Clusters (Edge Node) - telemetry from the cluster running on the Edge Node.
Applications (Edge Node) - telemetry from the applications running on the Edge Node.
Key components of observability stack that allow collection, storage, and consumption of metrics, logs, and traces are:
Grafana UI
- provides the frontend for visualization.OpenTelemetry Collector
(Customized):Provides logs collection pipelines with log transformation.
Collects supplementary metrics from Edge Orchestrator services.
Routes logs and metrics to appropriate tenants based on metadata.
Grafana Mimir
- used as metrics backend.Grafana Loki
- used as logs backend.
Note
To restrict access to data associated with individual tenants, a
grafana-proxy
custom extension is used while accessing datasources.
Refer to Multitenancy for more details on how
multitenancy in observability is handled.
Deployment#
The stack is separated into following ArgoCD applications:
edgenode-observability
- provides core services of the observability stack (for example, Grafana UI, Grafana Mimir, Grafana Loki, OpenTelemetry Collector).edgenode-dashboards
- provides Grafana UI dashboards for Edge Node Hosts, Clusters, and Applications.
Storage#
Grafana Mimir
and Grafana Loki
are configured to use:
AWS S3 for cloud-based deployments
S3-compatible storage (MinIO) for on-prem deployments