Wrote Loki config and followed Alloy OpenTelemetry log examples so application ERROR logs would land in the same stack as traces and metrics. Config shipped for local and cluster layouts; log storage was not started, so instance addressing was never proven.
- What worked
- Official Alloy log examples made OTLP-to-Loki routing straightforward to author.
- What got in the way
- A possible Loki instance-address mismatch was noted and left unverified because the stack could not be run.