Helm charts, Argo CD ApplicationSets and OpenTofu that stand up and test a DFE deployment on Kubernetes.
- One pin set for the whole stack in
versions.yaml: every chart, operator and image, with the image digests beside them. A release tag here is a certified stack version, and each release has its signed manifest, image list and upgrade graph. - Two layers on any Kubernetes, on-premises or in your cloud account.
bootstrap/bootstrap.shlays the base (cert-manager, External Secrets, Argo CD, adopting healthy ones the cluster has), then Argo CD reconciles the backing services and the DFE apps from the same charts everywhere. - The backing services deployed and wired: ClickHouse, Kafka on Strimzi or Redpanda, DocumentDB on CloudNativePG for the console, Envoy Gateway with OIDC, KEDA and External Secrets.
scripts/dfe-stackrenders the stack manifest, lists the DFE images to mirror for an air-gapped install, derives the upgrade graph from the stacksversions.yamldeclares and checks a customer's pins.scripts/dfe-opsdeploys, verifies, tears down and runs the acceptance tests against a cluster, or a local kind one.DFE_REGISTRYpoints a deployment at an image mirror of its own.
dfe-infra deploys the backing services and the apps. dfe-engine, the control plane, never does: it decides what the apps are configured to do and writes that to git. Everything here is generic, and a deployment's values live in its private deploy repo.
A Kubernetes cluster you run, at a version in the platform.kubernetes range versions.yaml gives, with kubectl pointed at it, plus helm, openssl and Python 3. The stack version is a release tag of this repo, and versions.yaml on main names the current one:
STACK=<stack version>
git clone --branch "$STACK" https://github.com/hyperi-io/dfe-infra && cd dfe-infra
(umask 077 && cp bootstrap/local.env.example bootstrap/local.env)Edit bootstrap/local.env before going on: DFE_DOMAIN, the domain every hostname sits under, DFE_REPO_URL, the git repository Argo CD reads the charts from, and DFE_STORAGE_CLASS, a StorageClass the cluster has (bootstrap installs local-path when the cluster has no StorageClass at all, and uses DFE_STORAGE_CLASS when classes exist and none is default). Then:
python3 scripts/dfe-ops preflight --mode single --env-file bootstrap/local.env
python3 scripts/dfe-ops stack-deploy --stack "$STACK" --mode single \
--target-revision "$STACK" --env-file bootstrap/local.envThe preflight is read-only and says whether the cluster is fit for the tier. stack-deploy installs or adopts the base, then hands over to Argo CD. The full walk-through, AWS and the one-host Docker install are on the published site, https://dfe.docs.hyperi.io/quickstart/.
The cluster secret's dfe.hyperi.io/profile picks the tier: slim, single or scale. Values merge from the chart defaults, then argocd/values/common.yaml, the per-cloud file, the profile file, and last the deploy repo's overlays, so you override single dials and never templates. docs/deployment/README.md has the cascade and the per-target guides.
- docs/deployment/: the guides per target (on-premises, RKE2, AWS), upgrades and rollback, storage, Kafka, managed backends, the portability matrix.
- docs/architecture.md: where this repo sits in the suite and the boundary with dfe-engine.
- docs/deployment/stack-versioning.md: what a stack release is and how
dfe-stackreads it. - docs/suite-graph.md: which repos are suite members and what moves when one releases (
suite.yamlis the source). - docs/testing-cycle.md and docs/threat-model.md: how a stack is tested, and what is exposed to whom.
- docs/README.md lists every page.
The suite's published docs are at https://dfe.docs.hyperi.io, and this repository's pages at https://dfe.docs.hyperi.io/dfe-infra/.
How to work in this repo, the commands that test a change and what tends to bite are in CONTRIBUTING.md. A question that is not a bug goes in a new issue.
Open bugs: this repo | whole DFE suite
BUSL-1.1, see LICENSE.