feat: add CI/GPU runner infrastructure migrated from FlowMesh_dev#15

Open

Qruixuan wants to merge 17 commits into

mainfrom

claude/migrate-flowmesh-work-E95UG

Commits on May 4, 2026

ci: migrate gpu-runner-setup-v2 changes from FlowMesh_dev

Qruixuan
authored and
claude
committed
chore: migrate ci/gpu-runner-setup-v2 docker + shared-copy changes from FlowMesh_dev

Qruixuan
authored and
claude
committed
revert: restore Dockerfiles and unit-tests.yml to FlowMesh main versions

Qruixuan
authored and
claude
committed
feat: add GPU requirements install + HF import error capture

Qruixuan
authored and
claude
committed
feat: add Docker CI compose infrastructure

Qruixuan
authored and
claude
committed
feat: add CI workflow and runner setup guide

Qruixuan
authored and
claude
committed
feat: add local CI runner script and fix template output destinations

Qruixuan
authored and
claude
committed
fix: redesign CI compose for FlowMesh's single-server architecture

Qruixuan
authored and
claude
committed
fix: update CI workflow and run_local.sh for single-server architecture

Qruixuan
authored and
claude
committed
fix: correct worker networking — host-mode workers need localhost URLs and exposed ports

Qruixuan
authored and
claude
committed
feat: add tests/integration/test_e2e.py — E2E smoke test for CI

Qruixuan
authored and
claude
committed
fix: use host bind-mount for worker results to avoid _VolumeInitializer

Qruixuan
authored and
claude
committed
fix: persist HF model cache across CI runs via host bind-mount

Qruixuan
authored and
claude
committed
fix: revert HF cache to named volume (same as FlowMesh_dev), bump vLLM timeout

Qruixuan
authored and
claude
committed
fix: replace gated llama model with open Qwen model in dag_inference.json
claude
committed
fix: harden ci.yml for zizmor pedantic audit
claude
committed
style: apply isort and black fixes
claude
committed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat: add CI/GPU runner infrastructure migrated from FlowMesh_dev#15

feat: add CI/GPU runner infrastructure migrated from FlowMesh_dev#15
Qruixuan wants to merge 17 commits into
mainfrom
claude/migrate-flowmesh-work-E95UG

Commits on May 4, 2026

ci: migrate gpu-runner-setup-v2 changes from FlowMesh_dev

chore: migrate ci/gpu-runner-setup-v2 docker + shared-copy changes from FlowMesh_dev

revert: restore Dockerfiles and unit-tests.yml to FlowMesh main versions

feat: add GPU requirements install + HF import error capture

feat: add Docker CI compose infrastructure

feat: add CI workflow and runner setup guide

feat: add local CI runner script and fix template output destinations

fix: redesign CI compose for FlowMesh's single-server architecture

fix: update CI workflow and run_local.sh for single-server architecture

fix: correct worker networking — host-mode workers need localhost URLs and exposed ports

feat: add tests/integration/test_e2e.py — E2E smoke test for CI

fix: use host bind-mount for worker results to avoid _VolumeInitializer

fix: persist HF model cache across CI runs via host bind-mount

fix: revert HF cache to named volume (same as FlowMesh_dev), bump vLLM timeout

fix: replace gated llama model with open Qwen model in dag_inference.json

fix: harden ci.yml for zizmor pedantic audit

style: apply isort and black fixes