feat(blender): Blender 5.2.2 LTS on fv-ml1 GPU 3, on demand (Prime)

Uses linuxserver/blender (Selkies Wayland desktop, NVENC), digest-pinned. The
container sees only GPU 3, has restart "no", and runs only while in use,
because GPU 3 is the reserve card for a full-size vLLM seat. Web desktop on
:3001 with basic auth (vault fv-ml1/blender-web-password). /work is on /tank
and is not backed up; /config lives under /opt/docker (restic).

Acceptance: Cycles finds the card on OptiX and CUDA (sm_120 kernels ship in the
build). The heavy self-test renders in 3.54 s on OptiX vs 22.71 s on CPU, a
functional check with n=1. Web auth answers 401 without credentials and with a
wrong password, and 200 with the right one.
This commit is contained in:
vh
2026-09-27 13:57:35 -07:00
parent 0b8632a7ed
commit 7ae7193c21
4 changed files with 111 additions and 0 deletions
+4
View File
@@ -224,6 +224,10 @@ eats into a future big seat's profiling margin. Small seats go on GPU 0, which h
the most uncommitted headroom (its seats commit util 0.88; GPU 1 is at 0.975 and
GPU 2 at 0.96).
**GPU 3 on-demand tenant (Prime, 2026-09-27): `blender`** (`stacks/blender/`). It is up only
while in use (`restart: "no"`, about 270 MiB when idle with the desktop running, 0 when down).
The reserve still stands: whenever a full-size seat takes GPU 3, Blender stays down.
**Retired:**
- `llama-swap` (former GGUF multiplexer on :9292) — replaced by dedicated
per-model seats (e.g. `llama-charrp`); no longer running.
+6
View File
@@ -0,0 +1,6 @@
# blender — copy to /opt/docker/compose/blender/.env on fv-ml1 (mode 0600).
IMAGE=lscr.io/linuxserver/blender:5.2.2-ls241@sha256:9216c77ab2bf38a758390f802c5777554df605c53456dbcad828509a27a8f28b
HTTPS_PORT=3001
# Basic auth for the web desktop. PASSWORD source of truth: secret get fv-ml1/blender-web-password
CUSTOM_USER=blender
PASSWORD=
+51
View File
@@ -0,0 +1,51 @@
# blender
**Blender 5.2.2 LTS on fv-ml1 GPU 3, on demand.** Agents drive it; there is also a browser
desktop to watch it or take over. Prime, 2026-09-27: "go ahead with gpu 3, both". He does not
use Blender himself, so the agent side (MCP) is the primary interface.
| | |
|---|---|
| **Desktop** | `https://10.251.50.54:3001` (self-signed cert). Basic auth: user `blender`, password `secret get fv-ml1/blender-web-password`. |
| **Image** | `lscr.io/linuxserver/blender:5.2.2-ls241@sha256:9216c77a…` (Selkies 2.0 Wayland desktop, NVENC stream). |
| **GPU** | GPU 3 only (`NVIDIA_VISIBLE_DEVICES=3`). About 270 MiB is held while the desktop runs. |
| **Files** | `/work` → `/tank/blender` (projects, assets, renders; **not backed up**). `/config` → `/opt/docker/data/blender` (prefs, add-ons; restic). Files are owned by infra-ops (uid 1002), so agents can `scp` in and out. |
## ⚠ On demand: GPU 3 is borrowed
GPU 3 is the fleet's reserve card for a full-size vLLM seat (`servers/fv-ml1/README.md`).
Blender uses it only while in use:
```bash
ssh infra-ops@10.251.50.54 'cd /opt/docker/compose/blender && docker compose up -d' # start
ssh infra-ops@10.251.50.54 'cd /opt/docker/compose/blender && docker compose down' # stop, card back to 0
```
`restart: "no"`, so a reboot never brings it back. **When a big seat moves onto GPU 3, Blender
stays down.**
## Headless rendering (no desktop needed)
```bash
ssh infra-ops@10.251.50.54 'docker exec -u abc blender blender -b /work/<file>.blend -E CYCLES -o /work/out/frame_#### -a -- --cycles-device OPTIX'
```
Run as `-u abc`, the image's user, mapped to uid 1002, so outputs land owned by infra-ops.
## Acceptance (2026-09-27, 1356)
- Cycles sees the card on both OptiX and CUDA: "NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation
Edition". The build ships `kernel_sm_120.cubin` plus OptiX PTX.
- Self-test (`/tank/blender/render_test.py`): a subdivided glass monkey, 1920×1080, 1024 samples,
32 bounces, no denoise. **OptiX 3.54 s against CPU 22.71 s (96 threads).** That is n=1 per
device: a functional check that the GPU is really used, not a benchmark. A trivial default-cube
scene could not separate them (0.59 s against 0.65 s), which is why the self-test scene is heavy.
- Web auth: no credentials gives 401, a wrong password 401, the right one 200.
- Selkies: "Render node 1 encodes H264, AV1, H265 on nvenc"; the Wayland renderer runs GL on GPU 3.
## Agent control (MCP)
Being researched with dvalin-smithy-dev: which Blender MCP server, and how it reaches Blender
across hosts. This section fills in once the choice is made. ⚠ Most Blender MCP add-ons expose
"run arbitrary Python inside Blender" on a TCP port. Treat that port like a shell: bind it to
the container or the LAN only, never publish it wider.
+50
View File
@@ -0,0 +1,50 @@
# blender: Blender 5.2 LTS on fv-ml1 GPU 3, driven by agents (MCP) with a browser desktop for
# watching. Prime, 2026-09-27: "go ahead with gpu 3, both".
#
# ⚠ ON DEMAND ONLY. GPU 3 is the fleet's deliberately empty card, the reserve for a full-size
# vLLM seat (flash-next needs 93 of 96 GiB, and vLLM sizes KV against TOTAL VRAM). Blender
# borrows it: `restart: "no"`, so a reboot never brings it back, and it is stopped whenever a big
# seat needs the card. Start: `docker compose up -d`. Stop: `docker compose down`.
#
# Image: linuxserver/blender (Selkies web desktop, official Blender build with sm_120 CUDA +
# OptiX kernels). Its NVIDIA mode needs host driver >= 580 (fv-ml1: 580.65.06) and
# /dev/nvidia-modeset. HTTPS only (Selkies' WebCodecs need a secure context), self-signed cert.
# Basic auth from .env (CUSTOM_USER / PASSWORD; vault fv-ml1/blender-web). The desktop grants
# a shell inside the container, so never expose it beyond the LAN.
#
# Paths: /config (home: prefs, add-ons) → /opt/docker/data/blender (restic via /opt/docker).
# /work (projects, assets, renders) → /tank/blender (big, NOT backed up; renders are
# regenerable, and anything precious is copied out).
#
# .env (tunables): IMAGE, HTTPS_PORT, CUSTOM_USER, PASSWORD.
name: blender
services:
blender:
image: ${IMAGE:?set IMAGE}
container_name: blender
restart: "no"
runtime: nvidia
environment:
PUID: "1002" # infra-ops, so agents can read and write /work over ssh
PGID: "1003"
TZ: America/Los_Angeles
CUSTOM_USER: ${CUSTOM_USER:?set CUSTOM_USER}
PASSWORD: ${PASSWORD:?set PASSWORD}
NVIDIA_VISIBLE_DEVICES: "3" # GPU 3 only: never touch the serving cards
NVIDIA_DRIVER_CAPABILITIES: all # compute (Cycles), graphics (viewport), video (NVENC stream)
devices:
- /dev/nvidia-modeset:/dev/nvidia-modeset
volumes:
- /opt/docker/data/blender:/config
- /tank/blender:/work
ports:
- "${HTTPS_PORT:-3001}:3001"
shm_size: "1gb"
labels:
- homepage.group=AI - Studios
- homepage.name=Blender
- homepage.icon=si-blender
- homepage.description=Blender 5.2 on fv-ml1 GPU 3 (on demand)
- homepage.href=https://10.251.50.54:${HTTPS_PORT:-3001}