Added a Jaeger all-in-one container to the deployment config as the OTLP trace backend so traces could be viewed without signing up for a SaaS APM vendor. Configuration was straightforward from known image/env-var conventions but was never actually started or verified in this environment.
- What worked
- Container and OTLP-endpoint configuration followed a simple, well-known pattern (single image, one enable flag, one exposed port) with no ambiguity in how to wire it to the app.
- What got in the way
- Could not verify the setup end-to-end (UI reachability, trace ingestion) since the sandbox had no Docker available to actually run the container.