BulkJobs
Loading and reprocessing data at volume, and following the work while it runs.
Submit a job
Queues work that is too large to do inside a request — a bulk load, a re-match, a quality rescore. The call returns immediately with a job id; progress is read from the job. When the queue is saturated the submission is refused with a retry hint rather than accepted and dropped. Requires `bulk-job.create`.
List jobs
Returns jobs with their status and live progress, most recent first. Progress is derived as the work happens, so a running job reports what it has actually completed. Requires `bulk-job.read`.
Load records in bulk
Submits a set of records for loading. Rows are validated and standardized on the way in, and a row that fails is reported individually rather than failing the batch. For a first load into an empty tenant, the initial-load option skips per-row history so the record's history begins at its first real change. Requires `bulk-job.create`.
Get worker capacity
Reports how much capacity exists for each kind of job and how much is in use, alongside what is queued and running. Read this to tell a slow queue apart from an undersized fleet. Requires `bulk-job.read`.
Get a job
Returns one job with its status, progress and summary. Requires `bulk-job.read`.
List a job's failed rows
Returns only the rows that failed, each with the reason. This is the list to fix and resubmit; the rest of the batch has already been applied. Requires `bulk-job.read`.
List a job's results
Returns the per-row outcome of a finished job. Requires `bulk-job.read`.
Cancel a job
Asks a queued or running job to stop. Work already committed is kept — a cancelled load is partial, not rolled back — and the job reports how far it got. Requires `bulk-job.cancel`.