chore: update Docker configurations to streamline backend build and enhance E2E testing environment

docker/docker-compose.dev.yml

@@ -10,6 +10,11 @@
 
 name: surfsense-dev
 
+x-backend-build: &backend-build
+  context: ../surfsense_backend
+  args:
+    EMBEDDING_MODEL: ${EMBEDDING_MODEL:-sentence-transformers/all-MiniLM-L6-v2}
+
 services:
   db:
     image: pgvector/pgvector:pg17
@@ -69,7 +74,7 @@ services:
       retries: 5
 
   backend:
-    build: ../surfsense_backend
+    build: *backend-build
     ports:
       - "${BACKEND_PORT:-8000}:8000"
     volumes:
@@ -114,7 +119,7 @@ services:
       start_period: 200s
 
   celery_worker:
-    build: ../surfsense_backend
+    build: *backend-build
     volumes:
       - ../surfsense_backend/app:/app/app
       - shared_temp:/shared_tmp
@@ -140,7 +145,7 @@ services:
         condition: service_healthy
 
   celery_beat:
-    build: ../surfsense_backend
+    build: *backend-build
     env_file:
       - ../surfsense_backend/.env
     environment:

docker/docker-compose.e2e.yml

@@ -54,6 +54,8 @@ x-backend-env: &backend-env
   HTTPS_PROXY: http://127.0.0.1:1
   HTTP_PROXY: http://127.0.0.1:1
   NO_PROXY: localhost,127.0.0.1,0.0.0.0,db,redis,host.docker.internal
+  HF_HUB_OFFLINE: "1"
+  TRANSFORMERS_OFFLINE: "1"
 
 services:
   db:
@@ -95,6 +97,8 @@ services:
         # tests/ is excluded from the main context by .dockerignore;
         # the e2e stage's `COPY --from=tests-source` pulls it in here.
         tests-source: ../surfsense_backend/tests
+      args:
+        EMBEDDING_MODEL: sentence-transformers/all-MiniLM-L6-v2
       cache_from:
         - type=gha,scope=surfsense-e2e-backend
       cache_to:

surfsense_backend/Dockerfile

@@ -93,6 +93,9 @@ RUN printf '%s\n' \
     '    pass' \
     | python || true
 
+ARG EMBEDDING_MODEL=sentence-transformers/all-MiniLM-L6-v2
+RUN python -c "from chonkie import AutoEmbeddings; AutoEmbeddings.get_embeddings('${EMBEDDING_MODEL}')"
+
 # Install Playwright browsers (the playwright python package itself is in deps)
 RUN playwright install chromium --with-deps
 
@@ -104,6 +107,7 @@ RUN mkdir -p /shared_tmp
 ENV PYTHONPATH=/app
 ENV UVICORN_LOOP=asyncio
 ENV TMPDIR=/shared_tmp
+ENV PYTHONUNBUFFERED=1
 
 # Tune glibc malloc to return freed memory to the OS more aggressively.
 # Without these, Python's gc.collect() frees objects but the underlying

surfsense_backend/scripts/docker/entrypoint.e2e.sh

@@ -19,7 +19,8 @@ wait_for_db() {
     # depends_on/healthchecks already gate on db readiness, this is just
     # belt-and-suspenders so a slow first connection doesn't race migrations.
     for i in {1..60}; do
-        if python -c "from app.db import engine; import asyncio; asyncio.run(engine.dispose())" 2>/dev/null; then
+        echo "[e2e-entrypoint] db check attempt ${i}/60"
+        if python -c "from app.db import engine; import asyncio; asyncio.run(engine.dispose())"; then
             echo "[e2e-entrypoint] db reachable after ${i} attempts"
             return 0
         fi