Kubernetes (Helm)

IMPORTANT

Data Safety Warning: For production deployments, we strongly advise using a dedicated SQL database (PostgreSQL or MySQL) for metadata, and enabling persistence (PVC) or using a remote backend (S3, GCS, Swift) for file storage to ensure no data is lost.

Plik provides a Helm chart to deploy it on Kubernetes.

The chart source is located in charts/plik/.

Installation

From Repository

helm repo add plik https://root-gg.github.io/plik
helm repo update
helm install plik plik/plik

From Source

git clone https://github.com/root-gg/plik.git
cd plik/charts/plik
helm install plik .

Configuration

The chart supports several modes of deployment. All Plik configuration options are available under the plikd key in values.yaml.

Values Reference

A complete table of all configurable values with types, defaults, and descriptions is available in the chart's README.

The README is auto-generated with helm-docs. To regenerate after editing values.yaml:

make helm-docs

Persistence

Plik requires two types of persistent storage when using the file data backend:

PVC	values.yaml key	Default mount path	Stores
File data	persistence	/home/plik/server/files	Uploaded files
Database	dbPersistence	/home/plik/server/db	SQLite metadata (plik.db)

By default both use emptyDir (data is lost on pod restart). For production, enable both.

WARNING

If you only enable persistence (file data PVC) but not dbPersistence, the SQLite database will still be on emptyDir. Restarting the pod will delete all metadata — uploaded files will survive on the PV but Plik will not know about them.

StatefulSet (Recommended when using File Data Backend)

StatefulSet is recommended for the file backend as each pod gets its own stable PVCs via volumeClaimTemplates.

kind: StatefulSet

persistence:
  size: 10Gi

dbPersistence:
  size: 1Gi

Deployment (Recommended when using Cloud Data Backends)

If you use S3, GCS, or Swift for file storage, a standard Deployment is more appropriate. Enable dbPersistence to preserve the SQLite database across pod restarts.

kind: Deployment

dbPersistence:
  enabled: true
  size: 1Gi

plikd:
  DataBackend: s3
  DataBackendConfig:
    Endpoint: "s3.amazonaws.com"
    Bucket: "my-bucket"
    # Credentials should be provided via secret

Custom Database Connection String

The default MetadataBackendConfig.ConnectionString is /home/plik/server/db/plik.db (inside the dbPersistence volume). Override it to use PostgreSQL or MySQL instead:

plikd:
  MetadataBackendConfig:
    Driver: "postgres"
    ConnectionString: "host=pg.example.com user=plik dbname=plik sslmode=require"

When using a remote database, dbPersistence is not needed.

Secrets Management

Sensitive values (API secrets, database passwords) are managed via a Kubernetes Secret. You can either provide them in values.yaml (which will be automatically extracted into a Secret and injected as environment variables) or specify an existingSecret.

# Use an existing secret containing PLIKD_GOOGLE_API_SECRET, etc.
existingSecret: "my-custom-plik-secret"

The following environment variables are supported for secret injection:

Environment Variable	Values Key
PLIKD_GOOGLE_API_SECRET	plikd.GoogleApiSecret
PLIKD_OVH_API_SECRET	plikd.OvhApiSecret
PLIKD_OVH_API_KEY	plikd.OvhApiKey
PLIKD_OIDC_CLIENT_SECRET	plikd.OIDCClientSecret
PLIKD_DATA_BACKEND_CONFIG	plikd.DataBackendConfig (sensitive keys as JSON)
PLIKD_METADATA_BACKEND_CONFIG	plikd.MetadataBackendConfig (sensitive keys as JSON)

Service and Ingress

The service is exposed via a ClusterIP by default. You can configure an Ingress in values.yaml.

ingress:
  enabled: true
  className: nginx
  hosts:
    - host: plik.example.com
      paths:
        - path: /
          pathType: ImplementationSpecific
  tls:
    - secretName: plik-tls
      hosts:
        - plik.example.com

Upgrading

When upgrading the chart, use:

helm repo update
helm upgrade plik plik/plik