Week one of Data Engineering with @AnalyticSages done. I built a pipeline that pulls the latest Bitcoin blocks, cleans them with Pandas, and loads them into Postgres. All in Docker. One command. Simple ETL on paper. The real lessons came from debugging. Docker has its own network brain. pip inside my build kept failing with "No matching distribution found for pandas." The real error underneath was DNS. Docker Desktop runs in its own VM, so I had to set DNS in the Docker Engine config itself. IPv6 can quietly break things. My container looked up the API, got an IPv6 address, and had no route to it. Forcing Python to use IPv4 fixed it. Your network decides your data source. mempool.space was unreachable from my connection, even in the browser. I switched to Blockstream's Esplora API. Same block format, zero code changes beyond the URL. Inside a container, localhost means the container itself. My script reads DB_HOST from an environment variable, so the same code runs on my laptop and in Docker. Healthchecks matter. The ingest service waits until Postgres reports ready before it runs. Always read the full error. The first line usually lies. Data Tribe: @AnalyticSages Data source: @Blockstream Esplora API Project: github.com/Zeegaths/data-eng…