Getting started
The preferred way to install and manage okapi is via the okapictl CLI. Although each Okapi sub-system is released as a separate Docker image, its recommended to go through the okapictl route to avoid going through configuration hassles and / or dealing with Docker stuff yourself.
Here’s some recipes to get started quickly.
Install okapictl
The supported installation and demo workflows use okapictl. It is
distributed on PyPI and should be installed before choosing a workflow:
python -m pip install okapi-ctlThe package is available from the okapi-ctl PyPI project.
The controller release, Okapi images, and Helm charts should be kept aligned
to the same Okapi version. See the okapictl reference for
the available commands and versioning guidance.
Choose a path
| Goal | Recommended path | What it provides |
|---|---|---|
| Evaluate Okapi locally | Getting started locally | Okapi with local PostgreSQL and ClickHouse |
| Deploy Okapi in production | Production deployment | The four Okapi Helm releases in an existing Kubernetes cluster |
| Configure PostgreSQL | Configuring PostgreSQL | Users, permissions, and credentials for PostgreSQL |
| Configure ClickHouse | Configuring ClickHouse | Distributed topology and the shared ClusterIP endpoint |
The local workflows are intended for evaluation and development. Production deployments should use separately operated PostgreSQL and ClickHouse, durable storage, backups, TLS, and a secret manager.