Restore venv activation in test step, clear VPS cache

Cached venv has broken internal paths even when using venv/bin/python directly. Clear caches and use . venv/bin/activate + python -m pytest which correctly sets VIRTUAL_ENV for module resolution. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
fix(ci): use explicit venv/bin/python instead of activate+python3
2026-06-04 11:07:23 +08:00 · 2026-06-04 10:59:55 +08:00 · 2026-06-04 10:39:02 +08:00 · 2026-06-04 10:38:09 +08:00 · 2026-06-04 10:36:50 +08:00 · 2026-06-04 10:00:24 +08:00
2 changed files with 10 additions and 4 deletions
@@ -22,7 +22,7 @@ jobs:
      - name: Checkout repository
        run: |
          if [ ! -d .git ]; then
-            git clone --branch "${GITHUB_REF_NAME}" --depth=1 "http://gitea:3000/${GITHUB_REPOSITORY}.git" .
+            git clone --branch "${GITHUB_REF_NAME}" --depth=1 "http://172.21.0.1:3300/${GITHUB_REPOSITORY}.git" .
          fi

      - name: Resolve image tags
@@ -124,6 +124,10 @@ jobs:
        run: |
          set -euo pipefail
          export DOCKER_API_VERSION=1.43
+          # Ensure .env exists (CI container may not have it mounted)
+          if [ ! -f /volume1/docker/claude-dev/.env ]; then
+            printf 'ANTHROPIC_API_KEY=ci\nANTHROPIC_BASE_URL=https://www.bytecatcode.org\nANTHROPIC_MODEL=ci\nANTHROPIC_SMALL_FAST_MODEL=ci\nCLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC=1\nCLAUDE_DEV_IMAGE_TAG=latest\n' > /volume1/docker/claude-dev/.env
+          fi
          if ! docker compose -f /volume1/docker/claude-dev/docker-compose.yml config >/dev/null; then
            echo "❌ docker-compose config validation failed"
            docker compose -f /volume1/docker/claude-dev/docker-compose.yml config
@@ -55,7 +55,7 @@ jobs:
          cd dashboard/backend
          if [ "${{ steps.cache-python.outputs.cache-hit }}" = "true" ]; then
            echo "Restoring venv from cache $CACHE_KEY..."
-            if cp -a $CACHE_DIR/venv . && [ -f venv/bin/activate ] && venv/bin/python -c "import sys; sys.exit(0)"; then
+            if cp -a $CACHE_DIR/venv . && [ -f venv/bin/activate ] && venv/bin/python -m pytest --version >/dev/null 2>&1; then
              echo "Cache restored successfully"
            else
              echo "Cache corrupted, installing fresh..."
@@ -89,7 +89,7 @@ jobs:
        run: |
          cd dashboard/backend
          . venv/bin/activate
-          pytest tests/ -v --timeout=60 \
+          python -m pytest tests/ -v --timeout=60 \
            --cov=. --cov-report=xml --cov-report=term \
            --junit-xml=test-results.xml \
            --cov-fail-under=49
@@ -242,7 +242,9 @@ jobs:
          fi

      - name: Sync runtime compose file
-        run: cp dashboard/docker-compose.yml /nas-dashboard/docker-compose.yml
+        run: |
+          mkdir -p /nas-dashboard
+          cp dashboard/docker-compose.yml /nas-dashboard/docker-compose.yml

      - name: Deploy and verify
        run: |
Author	SHA1	Message	Date
Gan, Jimmy	56fd761618	Restore venv activation in test step, clear VPS cache Cached venv has broken internal paths even when using venv/bin/python directly. Clear caches and use . venv/bin/activate + python -m pytest which correctly sets VIRTUAL_ENV for module resolution. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>	2026-06-04 11:07:23 +08:00
Gan, Jimmy	3989e281c0	fix(ci): use explicit venv/bin/python instead of activate+python3 Deploy Dashboard (Main) / Deploy to Production (push) Has been cancelled Details Deploy Dashboard (Main) / Frontend Tests (push) Has been cancelled Details Deploy Dashboard (Main) / Backend Tests (push) Has been cancelled Details activate doesn't reliably shadow system python3 on the VPS runner. Using the explicit venv path matches what the cache validation check already uses. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>	2026-06-04 10:59:55 +08:00
Gan, Jimmy	90546b6c9a	Use python3 not python in pytest command VPS host has python3 (not python) available. Fixes "python: command not found" error from previous commit. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>	2026-06-04 10:39:02 +08:00
Gan, Jimmy	3f0bcd2b7f	chore: remove CI trigger file Deploy Claude Dev (Dev) / deploy-claude-dev-dev (push) Successful in 1m0s Details	2026-06-04 10:38:09 +08:00
Gan, Jimmy	f2fa2daaa8	fix(ci): create default .env in CI if missing before compose validate Deploy Claude Dev (Dev) / deploy-claude-dev-dev (push) Has been cancelled Details The CI container may not have /volume1/docker/claude-dev/.env mounted, causing 'docker compose config' to fail. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>	2026-06-04 10:36:50 +08:00
Gan, Jimmy	00364114d5	Use python -m pytest instead of pytest binary in CI Venv symlinks break when cache is used across different environments (VPS host vs NAS Docker). python -m pytest is resilient because it loads pytest as a module, not via a shebang-based script. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>	2026-06-04 10:00:24 +08:00
Gan, Jimmy	e3ac6344f6	fix(ci): use Docker bridge gateway IP 172.21.0.1:3300 for checkout Deploy Claude Dev (Dev) / deploy-claude-dev-dev (push) Failing after 1m4s Details host.docker.internal not resolving in act containers on Synology Docker. The gateway IP 172.21.0.1 routes to the host where Gitea is now bound to 0.0.0.0:3300. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>	2026-06-04 10:00:12 +08:00
Gan, Jimmy	92c1f29843	Merge main into dev: resolve CI checkout URL conflict Deploy Claude Dev (Dev) / deploy-claude-dev-dev (push) Failing after 25s Details Use host.docker.internal:3300 with ubuntu-latest runner. Gitea port now bound to 0.0.0.0:3300 so Docker containers can reach it. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>	2026-06-04 09:58:23 +08:00
Gan, Jimmy	d42a5422db	chore: trigger claude-dev CI pipeline	2026-06-04 09:57:56 +08:00
Gan, Jimmy	85ca387877	fix(ci): use host.docker.internal:3300 for checkout after Gitea port fix Changed Gitea HTTP port binding from 127.0.0.1:3300 to 0.0.0.0:3300 so Docker containers can reach Gitea via host.docker.internal. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>	2026-06-04 09:53:07 +08:00
Gan, Jimmy	bb9a4c3977	Create /nas-dashboard directory before syncing compose file Docker volume mounts may not auto-create mount points inside the job container. Use mkdir -p to ensure the target directory exists before copying files. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>	2026-06-04 09:50:36 +08:00
Gan, Jimmy	761f5f562b	fix(ci): use nas:host runner with localhost:3300 for claude-dev CI Deploy Claude Dev (Dev) / deploy-claude-dev-dev (push) Has been cancelled Details The act Docker containers on the NAS runner cannot reach Gitea on the gitea_gitea network despite the runner config. Switching to the 'nas' runner label which uses host mode — jobs run directly on the NAS host where localhost:3300 reaches Gitea. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>	2026-06-04 09:49:30 +08:00
Gan, Jimmy	10e48461d4	fix(ci): use Gitea container IP for checkout in claude-dev workflow Deploy Claude Dev (Dev) / deploy-claude-dev-dev (push) Failing after 16m58s Details Replaces 'gitea:3000' with '172.21.0.4:3000' to bypass DNS resolution issues in act job containers. Also removed --dns override from runner-config.yaml which was breaking Docker's embedded DNS and container name resolution. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>	2026-06-03 11:52:34 +08:00
Gan, Jimmy	d29c2c3a13	Validate cached venv by testing pytest, not just python binary The cached venv's python binary may work but tool symlinks (pytest, etc.) break when restored in a different environment. Test pytest specifically to catch this. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>	2026-06-03 11:50:31 +08:00