Merge branch 'main' into postgres-vector-migration

Dockerfile

@@ -17,10 +17,14 @@ COPY static /app/static
 COPY src /app/src
 # Copy persisted data (e.g., chroma DB) so the container can access it if present
 COPY data /app/data
+# COPY the scripts directory so runtime init scripts (e.g. init_pgvector.py) exist
+COPY scripts /app/scripts
+
 COPY run.sh /app/run.sh
 
-# Make run.sh executable
+# Make run.sh (and optionally the init script) executable
 RUN chmod +x /app/run.sh
+RUN chmod +x /app/scripts/init_pgvector.py || true
 
 # Expose port 10000
 EXPOSE 10000

run.sh

@@ -6,9 +6,21 @@ WORKERS_VALUE="${WORKERS:-1}"
 TIMEOUT_VALUE="${TIMEOUT:-120}"
 PORT_VALUE="${PORT:-10000}"
 
-# Initialize the database
+# Initialize the database with retries (handle DB cold start)
 echo "Initializing database..."
-python scripts/init_pgvector.py
+MAX_RETRIES="${DB_INIT_RETRIES:-10}"
+SLEEP_BASE="${DB_INIT_SLEEP_BASE:-5}"
+
+i=1
+until python scripts/init_pgvector.py; do
+  if [ "$i" -ge "$MAX_RETRIES" ]; then
+    echo "Database initialization failed after $i attempts; exiting."
+    exit 1
+  fi
+  echo "init_pgvector.py failed (attempt $i/$MAX_RETRIES). Sleeping $((SLEEP_BASE * i))s before retry..."
+  sleep $((SLEEP_BASE * i))
+  i=$((i + 1))
+done
 
 echo "Starting gunicorn on port ${PORT_VALUE} with ${WORKERS_VALUE} workers and timeout ${TIMEOUT_VALUE}s"
 export PYTHONPATH="/app${PYTHONPATH:+:$PYTHONPATH}"