AI microservice went down in production because of a simple port conflict caused by 2 Pair programmers. All containers were showing healthy, no restarts, no crashes, and docker ps looked perfectly normal, yet the service was completely unreachable.

Instead of relying on container status, I went straight to the logs.

While reading through them, I spotted a port mismatch. The application was trying to bind to a different port than what the environment expected.

In the .env file, AI_ASSISTANT_PORT was set to 5051 while the service was configured for 5050. A recent change had quietly introduced this configuration drift.

After correcting the .env, I rebuilt only the affected service using docker compose build --no-cache ai-assistant and docker compose up -d --no-deps ai-assistant. The service recovered immediately. A small configuration mistake caused a production outage, and the logs made it obvious once I looked in the right place.