Persistent logging limits for Emacs message buffer in headless environments

When diagnosing failed configuration loads or deployments in headless environments, the volatile message buffer is often insufficient because it clears upon session termination. While the emacs-log-file variable provides a mechanism for persistent logging, there is uncertainty regarding the granularity of the data captured when a fatal Lisp exception occurs during startup.

Relying on debug-on-error to trigger backtrace buffers is ineffective in non-interactive sessions where no UI buffers can be inspected manually. It is unclear if the standard log file output includes the full call stack information typically provided by the backtrace command, or if it only records the final high-level error string.

Does setting emacs-log-file capture full backtrace details when debug-on-error is active, and is there a specific configuration to ensure the minibuffer-history is flushed to disk before a crash halts execution?