c3: make fresh-clone install work + rewrite stale README + repo pointer
c3 was install-broken for a fresh checkout and its README described a months-old skeleton. Fixed so users can actually clone → install → run: - pyproject: wire the in-repo `club3090-tui-core` (tools/tui-core, not on PyPI) via `[tool.uv.sources]` so `uv pip install -e tools/serve-cockpit` resolves it from the checkout. Validated in a clean uv venv: single command → `c3` present, imports OK (single-step previously failed: "club3090-tui-core was not found in the package registry"). - README (serve-cockpit): rewritten to match the actual app — the two modes (Run & Operate · Bring & Validate), the real install (uv one-liner + plain-pip two-step), live keybindings, and the working/guarded status — replacing the "Phase 1 walking skeleton / 🧪 Experimental" text that no longer described anything. - Top-level README: a pointer to the cockpit so a cloner discovers it. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01EfF565T9eSLaqGzidyJ1Pm
This commit is contained in:
noonghunna
@@ -68,6 +68,8 @@ bash scripts/update.sh
|
||||
|
||||
`launch.sh` calls `switch.sh` (down old, up new) and then `verify-full.sh` so you know it's serving cleanly before you point a client at it. See [`scripts/`](scripts/) for all helpers.
|
||||
|
||||
> 🖥️ **Prefer a TUI?** The **serve cockpit (`c3`)** wraps the same flow — discover → serve → operate → validate — in one lazydocker-style screen (catalog, live GPU/scenes/containers, Doctor health, and the add-a-model pipeline). Install: `uv pip install -e tools/serve-cockpit` then run `c3`. See [`tools/serve-cockpit/`](tools/serve-cockpit/).
|
||||
|
||||
> ⚠️ **Single-card long-context note:** Cliff 2 (GDN prefill OOM at >~50K single-prompt) is **open** on 24 GB single-card vLLM. Genesis v7.72.2 PN59 was intended as the fix but doesn't engage on chunked-prefill. **Workarounds:** [`vllm/dual`](docs/DUAL_CARD.md) (TP=2 escapes it) or [`llamacpp/default`](docs/SINGLE_CARD.md#bulletproof-no-cliffs) (different engine, no cliff). Full diagnosis at [`docs/CLIFFS.md`](docs/CLIFFS.md).
|
||||
|
||||
---
|
||||
|
||||
@@ -1,60 +1,71 @@
|
||||
# club3090 serve cockpit (`c3`)
|
||||
# club-3090 serve cockpit (`c3`)
|
||||
|
||||
**Status: 🧪 Experimental**
|
||||
A **lazydocker-style terminal UI** for the club-3090 AI inference stack — the front door for
|
||||
*discover → serve → operate → validate* on a consumer NVIDIA rig. Wraps the same registry, launchers,
|
||||
and scripts you'd run by hand (`switch.sh`, `launch.sh`, `gpu-mode`, `health.sh`, the verify/bench
|
||||
suite) behind one keyboard-driven screen.
|
||||
|
||||
A lazydocker-style TUI for the club-3090 AI inference stack — the front door for
|
||||
discover → serve → manage on a consumer NVIDIA rig.
|
||||
> **Status:** working — read paths (catalog, estate, containers, health) are live; write paths
|
||||
> (serve / scene-switch / downloads) are guarded by a confirm + a single-writer reconcile lease.
|
||||
|
||||
## Phase 1 scope (this package)
|
||||
## What's inside
|
||||
|
||||
This is the **walking skeleton**: every navigation node is present and renders
|
||||
something so layout, density, and keybindings can be signed off before the
|
||||
plumbing work begins. The only real data source wired in Phase 1 is the
|
||||
Discover → Catalog tab, which is populated from the live registry via
|
||||
`registry_variant_rows` (read-only, no side effects). Everything else
|
||||
(Serve, Estate/Containers, Validate, and all row/scene/container actions)
|
||||
is a static placeholder stub — it will be wired in Phase 3.
|
||||
Two modes, shown as a tab bar; the producer mode is hidden in the lean view.
|
||||
|
||||
- **`1` Run & Operate** *(always shown)*
|
||||
- **Catalog** — the full registry of model variants; filter, inspect, and **serve** one (`⏎`).
|
||||
- **Orchestration** — live GPU cards, the `gpu-mode` scenes (incl. `ai-studio`), and supporting services; switch scene / stop.
|
||||
- **Containers** — running/stopped services with engine + port; drill into **Logs / Top / Config**, start a stopped one.
|
||||
- **Doctor** — "is it serving correctly?" — `health.sh` live + `verify` / `verify-full` reads, basic/full reports, and the power-cap sweep.
|
||||
- **`2` Bring & Validate** *(producer lane — hidden in lean view)*
|
||||
- The add-a-model pipeline: **① Bring** (fit-check an HF repo) → **② Serve** (generate a compose + serve untested) → … → **⑤ Promote**.
|
||||
|
||||
Launch shows both modes by default; **`c3 --lean`** (or **`[C]`** in-app) gives the consumer view
|
||||
(Run & Operate only).
|
||||
|
||||
## How to run
|
||||
|
||||
Install into the repo environment (editable):
|
||||
|
||||
```bash
|
||||
cd tools/serve-cockpit
|
||||
pip install -e .
|
||||
```
|
||||
|
||||
Then launch:
|
||||
The cockpit depends on an in-repo sibling package, **`club3090-tui-core`** (`tools/tui-core/`), which
|
||||
is **not on PyPI** — install both from the checkout.
|
||||
|
||||
**With [uv](https://docs.astral.sh/uv/) (recommended — one command; the local path is wired in `pyproject.toml`):**
|
||||
|
||||
```bash
|
||||
uv pip install -e tools/serve-cockpit
|
||||
c3
|
||||
# or equivalently:
|
||||
python -m club3090_cockpit
|
||||
```
|
||||
|
||||
The app resolves the repo root automatically from `__file__` location.
|
||||
Override with `C3_REPO_ROOT=/path/to/club-3090` if installed elsewhere.
|
||||
**With plain pip (install the core first, then the cockpit):**
|
||||
|
||||
```bash
|
||||
pip install -e tools/tui-core
|
||||
pip install -e tools/serve-cockpit
|
||||
c3
|
||||
```
|
||||
|
||||
Either way the launch is `c3` (or `python -m club3090_cockpit`). The app finds the repo root from its
|
||||
own location; override with **`C3_REPO_ROOT=/path/to/club-3090`** if you installed it elsewhere.
|
||||
|
||||
## Keybindings
|
||||
|
||||
| Key | Action |
|
||||
|-----|--------|
|
||||
| `1` | Discover mode |
|
||||
| `2` | Serve mode |
|
||||
| `3` | Estate mode |
|
||||
| `4` | Validate mode |
|
||||
| `r` | Refresh catalog (re-reads registry only) |
|
||||
| `⏎` | Primary action (no-op in Phase 1 — pops notice) |
|
||||
| `?` | Help |
|
||||
| `q` | Quit |
|
||||
| `1` / `2` | Run & Operate · Bring & Validate |
|
||||
| `↑ ↓ ← →` | move within / between the tab bar and content |
|
||||
| `⏎` | primary action for the focused row (serve / start / download / confirm) |
|
||||
| `k` | stop a service / cancel a download |
|
||||
| `f` | force-start (experimental — skips the fit gate) |
|
||||
| `r` | refresh the catalog (re-reads the registry) |
|
||||
| `Y` | copy the focused context to the clipboard |
|
||||
| `.` | toggle the left rail (full-width content) |
|
||||
| `C` | toggle lean view (hide / restore the Bring & Validate mode) |
|
||||
| `?` | help · `q` quit |
|
||||
|
||||
## Running tests
|
||||
|
||||
```bash
|
||||
cd tools/serve-cockpit
|
||||
pip install -e ".[dev]"
|
||||
# or: pip install pytest pytest-asyncio
|
||||
pytest
|
||||
```
|
||||
Fully headless — no TTY, GPU, Docker, or script calls (a conftest blocks any real spawn).
|
||||
|
||||
Tests are fully headless: no TTY, no GPU, no Docker, no script calls.
|
||||
```bash
|
||||
uv pip install -e "tools/serve-cockpit[dev]" # or: pip install pytest pytest-asyncio
|
||||
cd tools/serve-cockpit && pytest
|
||||
```
|
||||
|
||||
@@ -17,6 +17,13 @@ dependencies = [
|
||||
[project.scripts]
|
||||
c3 = "club3090_cockpit.__main__:main"
|
||||
|
||||
# club3090-tui-core is an in-repo sibling package (tools/tui-core), not on PyPI.
|
||||
# Wire it as a local editable path so `uv pip install -e .` resolves it from the
|
||||
# checkout. Plain `pip` doesn't read this — for pip, install tui-core first
|
||||
# (`pip install -e ../tui-core`) then this package. See README "How to run".
|
||||
[tool.uv.sources]
|
||||
club3090-tui-core = { path = "../tui-core", editable = true }
|
||||
|
||||
[tool.hatch.build.targets.wheel]
|
||||
packages = ["club3090_cockpit"]
|
||||
|
||||
|
||||
Reference in New Issue
Block a user