fix(strixhalo): bake ray-serve-apps into image, revert to --no-deps
Some checks failed
Build and Push Images / build (Dockerfile.ray-worker-intel, intel) (push) Has been cancelled
Build and Push Images / build (Dockerfile.ray-worker-nvidia, nvidia) (push) Has been cancelled
Build and Push Images / determine-version (push) Has been cancelled
Build and Push Images / build (Dockerfile.ray-worker-rdna2, rdna2) (push) Has been cancelled
Build and Push Images / build (Dockerfile.ray-worker-strixhalo, strixhalo) (push) Has been cancelled
Build and Push Images / Release (push) Has been cancelled
Build and Push Images / Notify (push) Has been cancelled

The Ray runtime_env pip virtualenv shadows /opt/venv, causing vllm to be
imported from a PyPI install instead of the source-built one. Baking
ray-serve-apps directly into the image avoids the virtualenv entirely.
This commit is contained in:
2026-02-09 20:49:58 -05:00
parent 72489b920e
commit 23f239ef69

View File

@@ -206,6 +206,15 @@ RUN --mount=type=cache,target=/root/.cache/uv \
'grpcio>=1.60.0' \ 'grpcio>=1.60.0' \
'protobuf>=4.25.0' 'protobuf>=4.25.0'
# ── Ray Serve application package ──────────────────────────────────────
# Baked into the image so the LLM serve app can use the source-built vllm
# from /opt/venv instead of a pip runtime_env virtualenv (which would
# shadow it with a PyPI vllm that lacks ROCm support).
RUN --mount=type=cache,target=/root/.cache/uv \
uv pip install --python /opt/venv/bin/python3 \
--extra-index-url https://git.daviestechlabs.io/api/packages/daviestechlabs/pypi/simple/ \
'ray-serve-apps==1.0.1+10'
# ── Verify vendor torch survived ─────────────────────────────────────── # ── Verify vendor torch survived ───────────────────────────────────────
# Fail early if any install step accidentally replaced the vendor torch. # Fail early if any install step accidentally replaced the vendor torch.
RUN python3 -c "\ RUN python3 -c "\