dsh-zcode-farm
Let a DeepSeek Harness agent see every ComfyUI instance in a farm and dispatch each generation job to the idlest one, instead of always hitting the busiest single endpoint. 适合拥有多个ComfyUI后端、需要均衡分配图像生成任务的团队。
インストール
$
npx -p @deepseek-ai/dsh dsh plugin --profile web add github:BOWLUNA/dsh-zcode-farmドキュメント
README 全文を読む ↗Configuration
- id: comfyui-farm
config:
primaryUrl: 'http://127.0.0.1:8188'
instances:
- { id: gpu-18301, baseUrl: 'http://127.0.0.1:18301', label: '5090 shard' }
- { id: gpu-18303, baseUrl: 'http://127.0.0.1:18303', label: 'Blackwell 102G' }
needVramGb: 8
maxQueueDepth: 3
queueWeight: 1000
probeTimeoutMs: 5000
probeRetries: 1
How an instance is chosen (two stages)
- Filter — unreachable, or free VRAM
maxQueueDepth→ out. - Rank —
score = freeVramGb − queueDepth × queueWeight
The default queueWeight = 1000 is deliberate: queue depth dominates the ranking and free
VRAM only breaks ties. That matches the intuition — an idle small GPU finishes sooner than a
busy large one.
Prefer "biggest VRAM wins"? Set
queueWeight: 0.