Loki

Loki is a log aggregation system inspired by Prometheus.

Loki K8s is a Juju charm that deploys and manages Grafana Loki in monolithic mode on Kubernetes, designed for small to medium-sized deployments. It is a core component of COS Lite, the lightweight version of the Canonical Observability Stack (COS): https://documentation.ubuntu.com/observability

Any charm that supports the loki_push_api relation can send logs to Loki automatically. Collected logs can then be explored through Grafana and used to trigger alerts via Alertmanager.

Key features:

• Automatic log ingestion from related charms via the Loki Push API relation.
• Seamless integration with Grafana, Prometheus, and Alertmanager through Juju relations.
• Built-in Grafana dashboards and Prometheus alert rules for self-monitoring.
• Configurable ingestion rate limits, burst sizes, and log retention periods.
• Per-unit ingress support for cross-model log collection through reverse proxies such as Traefik.
• TLS support via the certificates relation.
• Distributed tracing support via integration with Tempo.
• Persistent storage for log index and chunks across container restarts.
• Configurable Kubernetes resource limits for CPU and memory.

Known limitations:

• Each unit runs as an independent Loki instance with its own local storage and no data replication. Ingress is per unit, and each unit registers as a separate Grafana datasource. For horizontally scalable, highly available log aggregation, consider the Loki Coordinator and Loki Worker charms.