Skip to main content
Think about the difference between cooking from scratch and using a meal kit. From scratch, you source every ingredient, measure each one, and hope the proportions are right. A meal kit arrives pre-measured with a recipe card so you just combine a few things and go—there’s far less to get wrong. Deploying applications into Kubernetes follows the same two paths. Manually you write every Deployment, Service, and ConfigMap for Loki, Alloy, and Grafana. That’s a lot of YAML. Helm is the meal kit: a package that contains the Kubernetes manifests and sensible defaults, letting you override only the values you care about.
A simple diagram depicting Helm as a meal kit: an open box with the Helm logo and a "defaults" tag, a separate values.yaml card, and a Kubernetes icon above.
A Helm chart packages all the bits for an application and exposes a values.yaml file where you override defaults such as image tags, resource requests/limits, and storage backend settings. Historically, there was an loki-stack all-in-one chart that bundled Loki, Promtail, and related components; that chart is now deprecated. The community maintains separate charts so each component can be managed independently. Today you’ll find Loki, Grafana, and Alloy charts published across the Grafana and grafana-community Helm repositories.
A slide titled "Three current charts, two repos" showing three boxed package icons on shelves labeled "loki", "grafana", and "alloy." A "grafana-community" tag appears under the loki and grafana boxes.
Add the required Helm repositories and update your local index:
Recommended charts and repositories Loki’s Helm chart supports multiple deployment topologies to match scale and operational needs:
Slide titled "Loki's deployment modes" showing three deployment options in boxed cards: SingleBinary ("all of Loki, one process" — learning/small volumes), SimpleScalable ("read, write, and backend targets" — mid-size production), and Distributed ("each component its own service" — large-scale production).
Deployment modes at a glance Loki requires durable storage for chunks and indexes. In a lab or on-prem cluster, an S3-compatible store such as MinIO works well; in cloud environments use the managed S3 (or GCS/Azure Blob) backend. The storage backend, endpoint, bucket, credentials, and the deployment mode are all configured in the chart’s values.yaml.
For this lesson we will use the Single Binary mode. It’s the simplest to get started with and sufficient for demos and low-volume clusters.
Next steps (quick pointers)
  • Edit a local values.yaml to set loki.mode: singlebinary, configure the storage section, and tune resource requests.
  • Install Loki with helm install loki grafana/loki -f values.yaml (replace loki with a release name you prefer).
  • Install Promtail or another log shipper to push logs into Loki and add Grafana to visualize logs.
Links and references

Watch Video