Project:Elasticsearch-K8s

Elasticsearch Configuration

Overview

The Elasticsearch service is deployed in Kubernetes using a StatefulSet resource, which provides:

• Persistent storage for index data via PersistentVolumeClaims
• Stable network identities for pods
• Ordered deployment and scaling

Architecture

• Deployment Type: StatefulSet with headless service
• Storage: Persistent volumes using csi-rbd-sc storage class
• Configuration: Single-node discovery mode
• Security: Runs as user 1000 with fsGroup 1000

Service Endpoints

• HTTP API: Port 9200 (configurable via Values.port)
• Transport: Port 9300 (cluster communication)
• Service Type: Headless (ClusterIP: None)

Initial Setup

Index Configuration

After the service is first deployed, recreate the search index from within the Wikibase pod:

# Update search index configuration
php /var/www/html/w/extensions/CirrusSearch/maintenance/UpdateSearchIndexConfig.php --startOver

# Force reindex all content
php /var/www/html/w/extensions/CirrusSearch/maintenance/ForceSearchIndex.php

Configuration Values

Key Helm values that can be customized:

• replicas: Number of Elasticsearch replicas
• image.repository & image.tag: Elasticsearch Docker image
• port: HTTP API port (default: 9200)
• javaOpts: JVM options (e.g., heap size)
• resources: CPU and memory requests/limits
• storage.size: Persistent volume size