If a Docker container fails with a non-zero exit code, container-executor attempts to run a docker logs -
tail=250 container_name to provide more details on why the container failed. While the idea is good, the current implementation will fail for most containers as they are leveraging a launch script whose output will be redirected to a file. The -tail option throws an error if no log output is available for the container, resulting in the docker logs command returning rc=1 in most cases.
I propose we remove this code from container-executor. Alternative approaches to handle logging can be explored as part of supporting an image's entrypoint.