Deploying Care
Care is self-hostable. You can run your own instance end to end, from the database and object store to the backend workers and the web frontend.
This section is written for operators and SREs standing up and running their own Care instance — people comfortable on the command line and with a cloud console. It is not an end-user or clinician guide.
Two ways to deploy
There are two complementary paths through this section. Pick the one that matches how you want to run Care.
| Path | Best when | What you get |
|---|---|---|
| Cloud provider guides | You are deploying onto one cloud and want a concrete, click-through recipe using that provider's managed services. | Step-by-step guides for AWS (ECS), Google Cloud (GKE, GCS), and DigitalOcean — managed Postgres, object storage, an image registry, and a CI/CD pipeline. |
| Kubernetes reference deployment | You want a provider-agnostic, self-managed production setup, or you run your own hardware. | The OHC reference architecture: a kubeadm cluster managed as infrastructure-as-code (OpenTofu + Helm) from the deployment-k8 repository. |
The two are not mutually exclusive — the provider guides show how Care maps onto a single cloud's managed services, while the Kubernetes reference is the portable, fuller-featured production setup that OHC runs itself.
What gets deployed
However you deploy it, a Care instance brings up the Care backend (web workers, Celery workers, and a scheduler), the Care frontend served as static assets, and the infrastructure they depend on: PostgreSQL, Redis, an S3-compatible object store, and — for a full production setup — OpenSearch/Elasticsearch, ingress with TLS termination, an image registry, and monitoring. The platform components Odoo and Metabase are commonly deployed alongside Care.
The cloud guides provision these from each provider's managed services; the Kubernetes reference runs them in-cluster from Helm charts.
What this section covers
Cloud provider guides
- Overview & infrastructure requirements — the components a production deployment needs, and how to choose a provider.
- AWS — deploy on ECS Fargate with a GitHub Actions pipeline.
- Google Cloud — deploy on GKE with Cloud Build, or host the frontend as a static site on GCS with Cloud CDN.
- DigitalOcean — deploy on the App Platform with managed Postgres and Spaces.
Kubernetes reference deployment
- Requirements — workload components, storage and backup expectations, and the tooling you need on your workstation.
- Reference architecture — the shape of the reference cluster: nodes, networking, ingress, storage, and supporting services.
- Provisioning the cluster — bringing up the Kubernetes cluster with kubeadm, the CNI, and storage.
- Deploying Care — applying the application
modules with the
deployer.pyworkflow. - Day-2 operations — backups, restores, monitoring, and routine maintenance.
Security & privacy
Applies to every deployment, however you host it:
- Data security — the standard security practices to enable at the edge, network, transport, data, and operational layers.
- Privacy & data protection — handling patient data responsibly and meeting your regulatory obligations.
:::tip Looking to contribute instead? If you want to run Care locally to develop against it rather than operate a production instance, see the Contributor's Guide. :::