Lightweight Kubernetes Log Collection with Loki: Deployment and Configuration Guide

The article introduces Loki, a lightweight log aggregation system from Grafana, as an alternative to resource‑heavy ELK stacks, explaining why it is suitable for Kubernetes environments.

Key advantages include support for multiple clients (Promtail, Fluentbit, Fluentd, Vector, Logstash, Grafana Agent), no strict log format requirements, PromQL‑style queries, dynamic log slicing, cloud‑native compatibility, and minimal indexing overhead.

Loki consists of five micro‑service components (distributor, ingester, querier, ruler, compactor) that can be deployed in various modes: an all‑in‑one binary, a micro‑service architecture with separate roles, or on bare metal.

For the all‑in‑one mode, a sample configuration file is provided. Important sections include auth_enabled: false , target: all , server ports, and detailed settings for distributor, querier, ingester, storage, and limits. The configuration can be saved as loki-all.yaml and loaded via a Kubernetes ConfigMap:

kubectl create configmap --from-file ./loki-all.yaml loki-all

Persistent storage is set up with a PersistentVolume and PersistentVolumeClaim using hostPath (or other supported back‑ends). Example YAML snippets are shown for both resources.

The Loki service itself is deployed as a StatefulSet with appropriate container arguments, probes, resource limits, and volume mounts. A corresponding Service exposes ports 3100 (HTTP) and 7946 (memberlist).

Promtail, the recommended log‑shipping agent, is configured with a YAML file that defines the server, client endpoint, positions file, and scrape_configs for Kubernetes pods. The Promtail DaemonSet mounts the config, host paths for container logs, and includes readiness probes. Example DaemonSet YAML is included.

For quicker installation, Helm charts can be used. The steps are adding the Grafana repo, updating it, and installing Loki with helm upgrade --install loki grafana/loki-simple-scalable , optionally specifying a namespace or custom values.

Common troubleshooting tips cover 502 errors (incorrect service URL), ingester readiness delays, replication factor mismatches, and Promtail‑Loki connectivity issues, with guidance on adjusting memberlist and replication_factor settings.