Diagnosing Immediate Exit in Docker Compose with Log Tail and Since Filters

Goal

We need to pinpoint the exact reason a Compose service exits immediately after start by inspecting the container’s stdout/stderr captured by the default json-file driver.

Constraints

• Logs are only available until the container is removed by docker compose down.
• Missing environment variables often surface as “missing required environment variable” or framework‑specific messages; generic ERROR or Fatal prefixes may be more reliable.
• The service may restart quickly, so a narrow --tail or a time‑based --since filter is needed to capture the failure without older context.

Questions

1. Which combination of --tail and --since flags best captures the error before the container exits?
2. Does setting log rotation options in the compose file guarantee that the relevant lines remain available for a subsequent docker compose logs call?
3. Can docker compose exec <service> env reliably confirm that the missing variable is indeed absent when the error appears in the logs?