Internal PCS¶

DataRobot relies on persistent critical services: MongoDB, PostgreSQL, RabbitMQ, Redis, and Elasticsearch. To streamline the installation process, DataRobot provides a Helm chart that integrates the Chainguard charts for these services.

Important: Before proceeding, read the Persistent Storage requirements to ensure your cluster has the proper storage configured.

High-availability deployment¶

Although DataRobot recommends using external PCS for production workloads, you can configure internal PCS in a high-availability (HA) mode by supplying the following override values in your configuration:

elasticsearch:
  master:
    replicaCount: 1 

If HA mode isn't required, use the following override values in your configuration:

mongodb:
  replicaCount: 1
postgresql-ha:
  postgresql:
    replicaCount: 1
  pgpool:
    replicaCount: 1
redis:
  replica:
    replicaCount: 1
rabbitmq:
  replicaCount: 1
elasticsearch:
  coordinating:
    replicaCount: 0
  ingest:
    replicaCount: 0
  data:
    replicaCount: 0
  master:
    masterOnly: false
    replicaCount: 1