This task consists of:

• Properly setting depends_on and restart fields in Docker compose files (so that each service is restarted if unhealthy/crashed)
• Adding logic for restarting unhealthy containers
• Adding timeout/retry logic on backend for checking availability of other services backend depends on, namely Milvus, Ollama