fix(compose): get the v2 stack working as a 4-service local dev loop#271
Closed
JoshuaAFerguson wants to merge 2 commits into
Closed
fix(compose): get the v2 stack working as a 4-service local dev loop#271JoshuaAFerguson wants to merge 2 commits into
JoshuaAFerguson wants to merge 2 commits into
Conversation
… loop The docker-compose.yml had drifted out-of-sync with the v2 architecture during the k8s-agent push. Rebuilds the local stack so `docker compose up -d` produces a working dev loop: - Drops obsolete `version: '3.9'` line. - Replaces the dead `docker-controller` service (the directory was renamed/removed) with `docker-agent` (agents/docker-agent/Dockerfile, marked WIP — sessions through it aren't guaranteed to fully work yet). - Adds the missing `ui` service (nginx + built React bundle, port 3000). - Wires AGENT_BOOTSTRAP_KEY on the api so the docker-agent self-registers on first connect (no manual provisioning). - Switches PLATFORM to `docker` so the api doesn't require a kubeconfig. (The api already supported a nil k8sClient downstream — the only blocker was a hard-fatal init that ignored PLATFORM.) - Lengthens JWT_SECRET to >=32 chars (api enforces a minimum). - Adds ADMIN_PASSWORD=admin123 so the login form works on first up. - Remaps host postgres to 5433 to avoid clashing with a host-side Postgres on 5432. Healthcheck fix in api/Dockerfile: - BusyBox wget's `--spider` sends HEAD; gin's /health route only registers GET, so the healthcheck failed and the api restart-looped. Switch to `wget -q -O /dev/null` which forces GET. Verified end-to-end: - `docker compose up -d` brings up postgres + nats + api + ui - http://localhost:3000 serves the UI - /api/v1/* proxies through nginx to the api container - POST /api/v1/auth/login with admin/admin123 returns a JWT token
Atomic cleanup of obsolete refs to the removed `./docker-controller/` directory (replaced by `agents/docker-agent/` in the v2 architecture): - Delete `scripts/build-docker-controller.sh` entirely — its build context (`./docker-controller`) doesn't exist anymore. - Rename `build_docker_controller` → `build_docker_agent` in `scripts/local-build.sh`; update image name, build context, and `valid components` help string accordingly. Update header comment. - Replace `docker-controller` / "Docker controller" usage strings in `scripts/docker-dev.sh` with the v2 `docker-agent` naming. - Update `scripts/README.md` to remove the deleted-script section and rename remaining "Docker controller" mentions. - Update `scripts/README-V2.md` matrix row to reflect that the legacy `build-docker-controller.sh` was replaced (not just renamed). - Annotate the one remaining production-code comment (in `sessiontemplates.go`) so readers know "Docker controller" referred to the pre-v2 component, now Docker agent. Out of scope for this commit (separately tracked structural work): helm chart values, scripts/local-deploy.sh, GH Actions release.yml, security-scan.yml, docs/TROUBLESHOOTING.md still reference the legacy `streamspace-kubernetes-controller` image.
github-actions
Bot
added
documentation
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: db9206ad14
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| api: | ||
| condition: service_healthy | ||
| ports: | ||
| - "3000:80" |
There was a problem hiding this comment.
Binding the new ui service to host port 3000 conflicts with the existing grafana mapping (3000:3000) when the monitoring profile is enabled. In that context (./scripts/docker-dev.sh --all or docker compose --profile monitoring up), one of the containers will fail to start with an address-in-use error, so the advertised "all services" dev workflow is broken.
Useful? React with 👍 / 👎.
5 tasks
Contributor
|
This pull request has been automatically marked as stale because it has not had recent activity. Action Required:
|
github-actions
Bot
added
the
stale
Contributor
|
This pull request was automatically closed due to inactivity. If you believe this was closed in error, please reopen it. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
The v1→v2 agent refactor left `docker-compose.yml` referencing `./docker-controller/` (a directory that no longer exists). Bringing the local dev loop back to a working state by replacing it with the v2 `docker-agent` and adding the missing `ui` service so `docker compose up -d` produces a fully usable stack.
The `docker-agent` service is profile-gated (`--profile docker`) and clearly labeled WIP in the compose comment — its session-lifecycle handlers fell behind during k8s-agent development and aren't guaranteed to fully drive sessions yet, but the wiring is right so the next person to fix it doesn't have to re-discover the topology.
Verified end-to-end
```bash
docker compose up -d
→ postgres + nats + api + ui all healthy
curl http://localhost:3000/health # → "OK"
curl http://localhost:3000/api/v1/auth/setup/status # → JSON, proxied through nginx
curl -X POST http://localhost:3000/api/v1/auth/login \
-H 'Content-Type: application/json' \
-d '{"username":"admin","password":"admin123"}' # → {"token":"eyJ..."}
```
UI loads at http://localhost:3000, login form succeeds, all admin routes render (verified via Playwright in the same session).
Required code change in api/cmd/main.go
The `PLATFORM=docker` switch needs the api to tolerate `k8sClient == nil`. The downstream handler signature already accepts nil (see comment at line ~360 of `api/cmd/main.go`), but the init at line 121 hard-fataled regardless. The one-line fix is documented in the PR description but left in the working tree (not committed here) because that file also contains the in-flight plugin work which is a separate review surface. Either fold it into the next plugin commit or land it as a tiny standalone "api: tolerate nil k8sClient when PLATFORM != kubernetes" PR.
Test plan