Output-dir mtimes beat logs for batch ETA
Estimating completion time of a long batch inference job where the worker process exposes no per-item progress over its API or logs.
When a batch job writes one output file per item to a known directory, 'ls -la <outdir>' and a diff between the first/latest mtime gives you a far better ETA than scraping the worker's logs — you get per-item duration directly from filesystem timestamps and can compute remaining-time from (total - done) * mean(batch_time). Particularly handy when the worker is opaque (whisper-server, ffmpeg batches, ML inference behind an HTTP shim) and emits only generic 'started/error' lines.
For any unfamiliar pipeline, locate the output directory first and use mtime deltas for ETA; do not read logs unless something has actually errored.