w

Dockerfile

@@ -22,9 +22,12 @@ RUN --mount=type=secret,id=HF_TOKEN \
 RUN echo "FROM /app/prem-1B-SQL.Q8_0.gguf" > /app/Modelfile
 
 # --- 5. Import the model into Ollama's registry ---
-# This makes the model available to serve
-RUN ollama create prem-sql-api -f /app/Modelfile
+# We start the server in the background (&), wait 5s for it to boot,
+# then run the 'create' command. The 'ollama serve' process
+# will automatically end when this RUN step completes.
+RUN sh -c 'ollama serve & \
+           sleep 5 && \
+           ollama create prem-sql-api -f /app/Modelfile'
 
-# The base image's default command is "ollama serve",
-# which will automatically start the API server on port 11434.
-# It will also serve our newly created "prem-sql-api" model.
+# The base image's default command ("ollama serve") will
+# now run and serve the "prem-sql-api" model we just created.