Start with the ordinary workflow log. When a script runs but leaves an unclear error, enable step debug logging. If the job stalls during setup or fails before the first step, runner diagnostic logging is a better starting point.


Match the log to the failure


Step debug increases the detail in job step logs. Runner diagnostics adds runner and worker process logs to the downloadable archive. For example, when a setup command runs and fails, check the step log. When a runner never reaches the command, look at the process logs in the runner-diagnostic-logs folder.


Limit the investigation


If you have access to rerun the workflow, GitHub lets you enable debug logging for the rerun. For repository-level configuration, GitHub documents ACTIONS_STEP_DEBUG and ACTIONS_RUNNER_DEBUG. Prefer a narrow diagnostic run when possible. Review downloaded logs before forwarding them, especially if your scripts print environment or configuration details.


GitHub documentation: https://docs.github.com/en/actions/how-tos/monitor-workflows/enable-debug-logging