Data Ingestion Queue: Job Detail Modal, Run Now and Retry Controls, Exact Solr Errors and Two API Fixes

The Data Ingestion Queue gets a full working surface: see exactly what a job did, run it now instead of waiting for cron, retry it after a fix, and read the precise Solr error behind each rejected document. Two API bugs are fixed alongside.

01 The Job Detail modal

Click any Job ID to open the full picture: status, progress, success and error counts, all timestamps, the complete error message, the processing result, and the entire payload that was submitted — with pretty-printed JSON.

02 Run Now, per job, per index, per user

Granular Run Now controls trigger immediate processing without waiting for the next cron cycle. You can run a single job, an entire index, or all pending jobs for a user. Each pending job shows its own Run button, and every index header carries a Run button whenever there are pending jobs beneath it.

03 One-click Retry

The Retry button resets a job back to pending and clears all progress, so you can re-run it after fixing whatever went wrong. It is available on completed, failed and stopped jobs.

04 Errors you can actually act on

• Exact Solr errors captured. When a document fails at the Solr level — unknown field, type mismatch, schema violation — the precise error returned by Solr is captured and shown in the Job Detail modal. No more guessing why a document was rejected.
• Short summaries in the table. Error columns now show a short summary such as "30 ok, 20 doc(s) failed — click for details" instead of the full raw error. Clicking the message opens the Job Detail modal with the complete per-document breakdown.

05 Two Data Ingestion API fixes

• content_type is now always set. The Data Ingestion API sets the field for every document. If you do not provide one, it defaults to text/html, so your documents appear in the web search tab instead of being classified as media. With rtf:true, the MIME type is auto-detected from the actual file content — PDFs get application/pdf, Word documents get their proper DOCX type, and so on.
• The rtf flag is reliably stripped. The internal rtf flag was not always removed before documents were sent to Solr, producing an "unknown field rtf" error that failed the entire batch. It is now removed regardless of how the value was encoded — boolean, string or integer.