demo/k8s/helm/logsys

logsys Helm chart

This chart deploys a complete log collection and storage stack:

• Loki — Log storage engine (StatefulSet with PersistentVolume)
• Vector — Log collector agent (DaemonSet)

The Vector agent collects container JSON logs from the shop-recycle namespace and ships parsed labels to Loki and metrics to a Prometheus exporter. Configuration and parsing rules are adapted from the Log.md monitoring guide.

Install

helm upgrade --install vector ./k8s/logsys -n shoprecycle --create-namespace

This single command deploys:

• Loki StatefulSet with persistent storage (10Gi default)
• Vector DaemonSet for log collection
• All ConfigMaps for configuration

Configuration

Loki:

• loki.enabled — Enable/disable Loki deployment (default: true)
• loki.namespace — Namespace (default: shoprecycle)
• loki.replicas — Number of Loki replicas (default: 1)
• loki.persistence.size — Storage size (default: 10Gi)
• loki.retention.days — Log retention in days (default: 30)

Vector:

• vector.enabled — Enable/disable Vector deployment (default: true)
• vector.loki.endpoint — Loki push endpoint (default: http://loki:3100)
• vector.logSelector — List of app names to collect (defaults: gateway/order/payment/web)

Usage

After deployment, access Loki via:

kubectl port-forward -n shoprecycle svc/loki 3100:3100

Then configure Grafana to use datasource http://loki:3100.

Notes

• This chart mounts /var/log from the host on each node.
• Loki uses filesystem storage with BoltDB for production-style log indexing.
• Follow the migration steps in Log.md: start with staging dual-write, then canary, then full roll-out.