Upgrading to Enterprise Search 8.6.0? See Upgrading and migrating.
Elastic Cloud deployments using Native connectors must include the Enterprise Search service, and the service must have at least 4 GB RAM per zone. As of 8.6.0, native connectors will not be available in Kibana for deployments running below this threshold.
To verify or change the RAM available to the Enterprise Search service on an Elastic Cloud deployment, see Infrastructure requirements.
- Use enhanced scheduling options for crawls with the Elastic web crawler. Crawls can be scheduled to avoid high-traffic time windows for specific websites.
Enhanced visibility into the status, duration and configured state of connectors and sync jobs is provided in the Content > Indices view, and the individual index Overview.
- See a new jobs detail flyout for every connector.
- Use our new Network Drive connector client (tech preview) to sync data from your file systems and network drives. Previously available as a connector package for Workplace Search, now available for search optimized Elasticsearch indices.
- Use our new Amazon S3 connector client (tech preview) to sync objects and more into search-optimized Elasticsearch indices.
Use new and improved tools to implement and manage NLP in search indices.
- Predicted value field processor – users can define a field for their ML model outputs to copy to, and a confidence threshold.
- Management tools – users can now see an index’s inference history, logs with any errors, and the JSON config for easy viewing/editing.
- Added diagnostic stats overview to the search indices screen.
- Added the ability to test ingest pipelines with a document from the same index.
- Added editable sync rules to specific connectors.
- Added the ability to detach Machine Learning inference pipelines for search indices without deleting the pipeline.
Updated the Add Inference Pipeline modal to allow choosing an existing
ml-inference-pipeline. Note that:
- You can only select pipelines that use a source field found on the current index.
- You cannot add the same pipeline twice.
- You currently cannot use a pipeline with a trained model that is unavailable to you. This will be addressed in follow-up work.
- Added the Ingestion logs panel to the pipelines tab for a search index. This will show up to 20 unique ingestion errors from inference pipelines stored on documents in the index.
- The Enterprise Search Node.js client is now generally available.
- Fixed a bug where a missing Github README could crash a Github Workplace Search content source sync job.
- Updated the displayed names of configurable fields for MySQL and MongoDB.
- Fixed App Search Search UI page render on Safari.