Files
infra/cloud/README.md

128 lines
6.4 KiB
Markdown
Raw Normal View History

# cloud
The single central hub. One deployment, internet-facing.
## What runs here
feat(sso): wire Keycloak SSO end-to-end across all apps New stack: - stacks/oauth2-proxy/ — per-app sidecars (mlflow, portainer, rabbitmq) that gate vhosts via nginx auth_request against Keycloak's wbd realm. Native OIDC wired into: - grafana (generic_oauth, role-attribute-path → Admin/Editor/Viewer) - jupyterhub (oauthenticator.GenericOAuthenticator) - node-red (passport-openidconnect; in-memory state store + users() resolver because adminAuth doesn't expose req.session) - jenkins (oic-auth plugin via JCasC; matrix-auth for authz; setup wizard suppressed; custom image with plugins.txt) Infra fixes uncovered while bringing the above online: - nginx-proxy: bump proxy_buffer_size to 16k so oauth2-proxy callbacks don't 502 on the JWT-bearing Set-Cookie header. - nginx-proxy: add `resolver 127.0.0.11 valid=30s` so service names re-resolve after sidecar recreates (was cross-wiring oauth2-proxy upstreams after restart). - jupyterhub: pass --allow-root to the singleuser spawner (hub runs as root inside its container; jupyter-server refused root without flag). - jupyterhub Dockerfile: install jupyterlab + notebook so SimpleLocalProcessSpawner has something to launch. - node-red Dockerfile: install passport-openidconnect into the image so settings.js can require() it. - portainer: pre-seed local admin via --admin-password=<bcrypt-hash> so the 5-minute "no admin → lockout" timer can never trigger. - deploy.sh: restore executable bit (was 644 in repo). Admin/viewer policy: - Created realm role `app-admin` in keycloak wbd realm. - Grafana maps app-admin → Admin (default Viewer). - Jenkins matrix-auth grants r.de.ren Overall/Administer, authenticated users get Overall/Read + Job/Read + View/Read. - Node-RED: NODERED_ADMIN_USERS env list → permissions "*", others ["read"]. (TODO: switch to app-admin realm role.) - JupyterHub: JUPYTERHUB_ADMIN_USERS env list. (Same TODO.) - Gitea: r.de.ren pre-created as local admin; OIDC auto-links via email. Docs: - README, cloud/README, stacks/oauth2-proxy/README, and per-stack READMEs updated to reflect the new state and remove resolved TODOs. - cloud/.env.example gains all the new OIDC client + cookie-secret keys. - cloud/README documents the full kcadm realm bootstrap, including the hardcoded-audience mapper and post-logout redirect URIs that are non-obvious gotchas. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-21 18:34:37 +00:00
nginx-proxy, wireguard-server, keycloak, oauth2-proxy, portainer, influxdb, grafana, node-red, rabbitmq, postfix, gitea, jenkins, mlflow, jupyterhub, frost, sql.
See [`../docs/architecture.md`](../docs/architecture.md) for the full network topology and ingress table.
feat(sso): wire Keycloak SSO end-to-end across all apps New stack: - stacks/oauth2-proxy/ — per-app sidecars (mlflow, portainer, rabbitmq) that gate vhosts via nginx auth_request against Keycloak's wbd realm. Native OIDC wired into: - grafana (generic_oauth, role-attribute-path → Admin/Editor/Viewer) - jupyterhub (oauthenticator.GenericOAuthenticator) - node-red (passport-openidconnect; in-memory state store + users() resolver because adminAuth doesn't expose req.session) - jenkins (oic-auth plugin via JCasC; matrix-auth for authz; setup wizard suppressed; custom image with plugins.txt) Infra fixes uncovered while bringing the above online: - nginx-proxy: bump proxy_buffer_size to 16k so oauth2-proxy callbacks don't 502 on the JWT-bearing Set-Cookie header. - nginx-proxy: add `resolver 127.0.0.11 valid=30s` so service names re-resolve after sidecar recreates (was cross-wiring oauth2-proxy upstreams after restart). - jupyterhub: pass --allow-root to the singleuser spawner (hub runs as root inside its container; jupyter-server refused root without flag). - jupyterhub Dockerfile: install jupyterlab + notebook so SimpleLocalProcessSpawner has something to launch. - node-red Dockerfile: install passport-openidconnect into the image so settings.js can require() it. - portainer: pre-seed local admin via --admin-password=<bcrypt-hash> so the 5-minute "no admin → lockout" timer can never trigger. - deploy.sh: restore executable bit (was 644 in repo). Admin/viewer policy: - Created realm role `app-admin` in keycloak wbd realm. - Grafana maps app-admin → Admin (default Viewer). - Jenkins matrix-auth grants r.de.ren Overall/Administer, authenticated users get Overall/Read + Job/Read + View/Read. - Node-RED: NODERED_ADMIN_USERS env list → permissions "*", others ["read"]. (TODO: switch to app-admin realm role.) - JupyterHub: JUPYTERHUB_ADMIN_USERS env list. (Same TODO.) - Gitea: r.de.ren pre-created as local admin; OIDC auto-links via email. Docs: - README, cloud/README, stacks/oauth2-proxy/README, and per-stack READMEs updated to reflect the new state and remove resolved TODOs. - cloud/.env.example gains all the new OIDC client + cookie-secret keys. - cloud/README documents the full kcadm realm bootstrap, including the hardcoded-audience mapper and post-logout redirect URIs that are non-obvious gotchas. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-21 18:34:37 +00:00
Every human-accessible UI is gated by **Keycloak SSO** (wbd realm). Apps with native OIDC (gitea, grafana, node-red, jenkins, jupyterhub) speak OIDC directly; apps without (mlflow, portainer, rabbitmq) are gated by an `oauth2-proxy` sidecar via nginx `auth_request`. See the [`Keycloak bootstrap`](#keycloak-realm-bootstrap-one-time) section below.
## Run
```bash
cp .env.example .env # fill in real secrets first
feat(cloud): single-shot deploy.sh + FROST stack + healthchecks Stage 5 — make the cloud composition spin up in one command and add the SensorThings (FROST) stack as a fully segregated tenant. cloud/deploy.sh — idempotent, 7-step bring-up: preflight → validate → up + wait → cert state → issue/renew → service status → endpoint smoke test. Reissues LE cert only when current issuer no longer matches ACME_CA_URI. Move-aside-then- restore-on-failure so the bootstrap cert survives a failed certbot. stacks/frost — new stack, segregated from shared sql/rabbitmq: - dedicated postgis container (frost-db) - dedicated internal mosquitto bus (frost-mosquitto) - frost-http + frost-mqtt on a private frost-internal network, joined to cloud-app only for nginx ingress at frost.wbd-rd.nl - shared mosquitto stack deleted; rabbitmq remains the only public MQTT broker (mqtt.wbd-rd.nl:8883 via stream proxy) stacks/sql — pg_isready healthcheck so keycloak/gitea/mlflow can gate on service_healthy via cloud-level depends_on overrides. stacks/nginx-proxy: - nginx-init service generates a self-signed bootstrap cert on fresh deploy so nginx starts before certbot has issued a real one - frost.wbd-rd.nl vhost (/FROST-Server → frost-http:8080, /mqtt → frost-mqtt:9876 WebSocket) stacks/mlflow — custom Dockerfile (upstream + psycopg2-binary) so the official image can speak to the shared sql backend. stacks/jupyterhub — DummyAuthenticator stub gated by JUPYTERHUB_ADMIN_PASSWORD; TODO comments point at OIDC + DockerSpawner. stacks/rabbitmq — config/{enabled_plugins,rabbitmq.conf} stubs (management + mqtt plugins, MQTT auth required). stacks/portainer — ports unpublished; nginx now the only ingress. stacks/node-red — pin to 4.1 (the floating "4" tag does not exist). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-21 16:37:58 +02:00
./deploy.sh # one-shot bring-up: containers + cert + smoke test
```
feat(cloud): single-shot deploy.sh + FROST stack + healthchecks Stage 5 — make the cloud composition spin up in one command and add the SensorThings (FROST) stack as a fully segregated tenant. cloud/deploy.sh — idempotent, 7-step bring-up: preflight → validate → up + wait → cert state → issue/renew → service status → endpoint smoke test. Reissues LE cert only when current issuer no longer matches ACME_CA_URI. Move-aside-then- restore-on-failure so the bootstrap cert survives a failed certbot. stacks/frost — new stack, segregated from shared sql/rabbitmq: - dedicated postgis container (frost-db) - dedicated internal mosquitto bus (frost-mosquitto) - frost-http + frost-mqtt on a private frost-internal network, joined to cloud-app only for nginx ingress at frost.wbd-rd.nl - shared mosquitto stack deleted; rabbitmq remains the only public MQTT broker (mqtt.wbd-rd.nl:8883 via stream proxy) stacks/sql — pg_isready healthcheck so keycloak/gitea/mlflow can gate on service_healthy via cloud-level depends_on overrides. stacks/nginx-proxy: - nginx-init service generates a self-signed bootstrap cert on fresh deploy so nginx starts before certbot has issued a real one - frost.wbd-rd.nl vhost (/FROST-Server → frost-http:8080, /mqtt → frost-mqtt:9876 WebSocket) stacks/mlflow — custom Dockerfile (upstream + psycopg2-binary) so the official image can speak to the shared sql backend. stacks/jupyterhub — DummyAuthenticator stub gated by JUPYTERHUB_ADMIN_PASSWORD; TODO comments point at OIDC + DockerSpawner. stacks/rabbitmq — config/{enabled_plugins,rabbitmq.conf} stubs (management + mqtt plugins, MQTT auth required). stacks/portainer — ports unpublished; nginx now the only ingress. stacks/node-red — pin to 4.1 (the floating "4" tag does not exist). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-21 16:37:58 +02:00
`deploy.sh` is idempotent — rerun any time. It will:
1. **Preflight** — check `.env` has all required vars
2. **Validate** `docker compose config`
3. **Bring up** containers, wait for `sql` healthcheck, wait for nginx :80
4. **Inspect cert** — figure out whether the current cert is self-signed, staging, or prod
5. **Issue / renew** the SAN cert via certbot only when needed (initial issuance, or when `ACME_CA_URI` no longer matches the current issuer); reload nginx
6. **Status** — show `docker compose ps`
7. **Smoke test** every `*.wbd-rd.nl` vhost over loopback
The script reissues the cert **only** when the CA in `.env` changes (e.g. staging → prod) or when only the bootstrap dummy is present — it does not waste Let's Encrypt rate limits on repeated runs.
### Staging → prod flip
1. Verify everything works with the staging cert (browser will warn — that's normal)
2. Edit `.env`: change `ACME_CA_URI` to `https://acme-v02.api.letsencrypt.org/directory`
3. `./deploy.sh` — script detects the CA change and force-renews against prod
## Ingress (host port bindings)
| Port | Container |
|---|---|
| tcp/80, 443 | nginx-proxy |
| tcp/8883 | nginx-proxy (MQTT-TLS via stream block) |
| udp/51820 | wireguard-server |
Everything else stays on the internal `app` / `data` / `mgmt` networks.
feat(sso): wire Keycloak SSO end-to-end across all apps New stack: - stacks/oauth2-proxy/ — per-app sidecars (mlflow, portainer, rabbitmq) that gate vhosts via nginx auth_request against Keycloak's wbd realm. Native OIDC wired into: - grafana (generic_oauth, role-attribute-path → Admin/Editor/Viewer) - jupyterhub (oauthenticator.GenericOAuthenticator) - node-red (passport-openidconnect; in-memory state store + users() resolver because adminAuth doesn't expose req.session) - jenkins (oic-auth plugin via JCasC; matrix-auth for authz; setup wizard suppressed; custom image with plugins.txt) Infra fixes uncovered while bringing the above online: - nginx-proxy: bump proxy_buffer_size to 16k so oauth2-proxy callbacks don't 502 on the JWT-bearing Set-Cookie header. - nginx-proxy: add `resolver 127.0.0.11 valid=30s` so service names re-resolve after sidecar recreates (was cross-wiring oauth2-proxy upstreams after restart). - jupyterhub: pass --allow-root to the singleuser spawner (hub runs as root inside its container; jupyter-server refused root without flag). - jupyterhub Dockerfile: install jupyterlab + notebook so SimpleLocalProcessSpawner has something to launch. - node-red Dockerfile: install passport-openidconnect into the image so settings.js can require() it. - portainer: pre-seed local admin via --admin-password=<bcrypt-hash> so the 5-minute "no admin → lockout" timer can never trigger. - deploy.sh: restore executable bit (was 644 in repo). Admin/viewer policy: - Created realm role `app-admin` in keycloak wbd realm. - Grafana maps app-admin → Admin (default Viewer). - Jenkins matrix-auth grants r.de.ren Overall/Administer, authenticated users get Overall/Read + Job/Read + View/Read. - Node-RED: NODERED_ADMIN_USERS env list → permissions "*", others ["read"]. (TODO: switch to app-admin realm role.) - JupyterHub: JUPYTERHUB_ADMIN_USERS env list. (Same TODO.) - Gitea: r.de.ren pre-created as local admin; OIDC auto-links via email. Docs: - README, cloud/README, stacks/oauth2-proxy/README, and per-stack READMEs updated to reflect the new state and remove resolved TODOs. - cloud/.env.example gains all the new OIDC client + cookie-secret keys. - cloud/README documents the full kcadm realm bootstrap, including the hardcoded-audience mapper and post-logout redirect URIs that are non-obvious gotchas. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-21 18:34:37 +00:00
## Keycloak realm bootstrap (one-time)
After `deploy.sh` succeeds, Keycloak is up at `https://auth.wbd-rd.nl/` with only the master realm. You need to create the `wbd` realm + per-app OIDC clients before SSO works. Driven entirely by `kcadm.sh` inside the keycloak container, so it's reproducible:
```bash
cd cloud
set -a && . ./.env && set +a
KC="docker compose exec -T keycloak /opt/keycloak/bin/kcadm.sh"
# 1. Authenticate against master realm
$KC config credentials --server http://localhost:8080 --realm master \
--user "$KEYCLOAK_ADMIN" --password "$KEYCLOAK_ADMIN_PASSWORD"
# 2. Create the realm
$KC create realms -r master \
-s realm=wbd -s enabled=true -s displayName="WBD R&D" \
-s registrationAllowed=false -s resetPasswordAllowed=true -s rememberMe=true
# 3. Create an OIDC client per app. Pattern:
# clientId = stable lowercase name (matches *_OAUTH_CLIENT_ID in .env)
# redirectUri = the app's documented OIDC callback URL
# Hardcoded audience mapper is critical for oauth2-proxy clients — without
# it the access token's aud will be [realm-management, account] and
# oauth2-proxy will 500 on the callback.
create_client() {
local CID=$1 REDIRECT=$2 SECRET=$3
local ID=$($KC create clients -r wbd \
-s clientId=$CID -s enabled=true -s protocol=openid-connect \
-s publicClient=false -s standardFlowEnabled=true \
-s "redirectUris=[\"$REDIRECT\"]" \
-s "attributes.\"post.logout.redirect.uris\"=\"https://${REDIRECT#https://*/}*\"" \
${SECRET:+-s secret="$SECRET"} -i)
$KC create clients/$ID/protocol-mappers/models -r wbd \
-s name=audience-self -s protocol=openid-connect \
-s protocolMapper=oidc-audience-mapper \
-s "config.\"included.client.audience\"=$CID" \
-s "config.\"access.token.claim\"=true" -s "config.\"id.token.claim\"=false"
echo "$CID -> $ID"
}
create_client gitea "https://git.wbd-rd.nl/user/oauth2/keycloak/callback" "$GITEA_OAUTH_CLIENT_SECRET"
create_client grafana "https://dash.wbd-rd.nl/login/generic_oauth" "$GRAFANA_OAUTH_CLIENT_SECRET"
create_client node-red "https://flow.wbd-rd.nl/auth/strategy/callback/" "$NODERED_OAUTH_CLIENT_SECRET"
create_client jenkins "https://ci.wbd-rd.nl/securityRealm/finishLogin" "$JENKINS_OAUTH_CLIENT_SECRET"
create_client jupyterhub "https://hub.wbd-rd.nl/hub/oauth_callback" "$JUPYTERHUB_OAUTH_CLIENT_SECRET"
create_client mlflow "https://ml.wbd-rd.nl/oauth2/callback" "$MLFLOW_OAUTH_CLIENT_SECRET"
create_client portainer-ce "https://ops.wbd-rd.nl/oauth2/callback" "$PORTAINER_OAUTH_CLIENT_SECRET"
create_client rabbitmq "https://mq.wbd-rd.nl/oauth2/callback" "$RABBITMQ_OAUTH_CLIENT_SECRET"
# 4. Create the realm role used for cross-app admin promotion
$KC create roles -r wbd -s name=app-admin \
-s "description=Grants admin perms across all wbd-realm apps that recognise this role"
# 5. Create the first operator user and grant realm-admin + app-admin
$KC create users -r wbd \
-s username=r.de.ren -s email=r.de.ren@brabantsedelta.nl -s emailVerified=true \
-s firstName=R -s lastName='de Ren' -s enabled=true \
-s 'requiredActions=["UPDATE_PASSWORD"]'
$KC set-password -r wbd --username r.de.ren --new-password '<initial-temp-password>' --temporary
$KC add-roles -r wbd --uusername r.de.ren --cclientid realm-management --rolename realm-admin
$KC add-roles -r wbd --uusername r.de.ren --rolename app-admin
# 6. Final per-app wiring
# - Gitea: run `gitea admin user create --admin --username r.de.ren ...` then
# `gitea admin auth add-oauth --name keycloak ...` (see stacks/gitea/README.md)
# - Everything else picks up its OIDC config from .env on next start.
docker compose restart grafana node-red jenkins jupyterhub
```
After that, every vhost in the smoke-test table redirects unauthenticated users to Keycloak. New teammates: add a user in `https://auth.wbd-rd.nl/admin/wbd/console/` → Users → Add user; default permissions are viewer/read-only across all apps until you also assign them the `app-admin` realm role.
## Adding a new stack
1. Create `stacks/<name>/` with `compose.yml`, `.env.example`, `README.md`.
2. Uncomment (or add) the `include:` entry in `compose.yml`.
3. Add the stack's env vars to `.env.example`.
4. `docker compose pull && docker compose up -d`.