Retry Failed Documents

Retry failed documents in a batch with intelligent filtering by error type and tier.

bucket_identifierstring required

The unique identifier of the bucket.

batch_idstring required

The unique identifier of the batch.

Request body

retry_mode'transient_only' | 'all'

Determines which types of failed documents to retry. 'transient_only': Only retry documents that failed with transient errors (network issues, timeouts, rate limits). This is the recommended default as it avoids retrying documents with permanent failures (invalid data, missing fields, incompatible formats). 'all': Retry all failed documents regardless of error type. Use this for force retries after fixing data issues or updating extraction logic.

tier_numsinteger[] nullable

List of specific tier numbers to retry failures from. OPTIONAL - If not provided or empty list, retries failures from all tiers. Use this to selectively retry a specific processing stage. Example: [2] would only retry failures from tier 2 (the third processing stage). Tier numbering starts at 0 (bucket → collection), tier 1+ are collection → collection.

max_retry_countinteger

Maximum number of times a document can be retried. Documents that have been retried this many times already are excluded. Prevents infinite retry loops for documents with persistent issues. Default is 3 retries per document. Set to higher value for more aggressive retries.

Example request

{
  "description": "Retry only transient errors (default, recommended)",
  "retry_mode": "transient_only"
}

Response

Successful Response

object required

v5

Retry Failed Documents

Path parameters

Request body

Example request

Response