Retry Failed Indexing Jobs

A failed indexing job can now be retried from its job id, in Captain Studio or with the API. The retry is a new job that redoes only the files that failed and leaves the files that indexed alone.

• Retry in Studio: open a job on the Activity page and pick Retry N failed files when some of its files failed. A job that failed before it reached any file, or that timed out, offers Rerun job, which runs the job again, skips any file that already indexed, and bills the work it does. Jobs started by a sync offer Re-sync, which re-runs that sync. The first click asks for confirmation, and the new job opens from a link once it starts.
• Retry with the API: POST /v2/jobs/{job_id}/reindex starts the retry and returns the new job_id, plus reindex_of naming the original job. It reuses the original job's source, options and collection, and runs on the current pipeline, so fixes we've shipped since the first run apply.
• Safe to repeat: calling it again for the same job within 24 hours returns the same new job instead of starting another.
• When a retry is refused: a job can be retried for 90 days. A job that is still running, finished cleanly, replaced existing files, used a Dropbox token, or failed because its storage credentials were refused returns a 409 with a code saying why. A job whose collection or uploaded files are gone returns 410. The reference page lists every code.