Elasticsearch Search and Search Explain API endpoints are now only available under
v0endpoints are removed.
You can now add filtering rules as part of your connector workflow, when ingesting data from MySQL databases.
Use new features for the Elastic web crawler, including:
- Full HTML extraction. You can now store the full HTML output of a URL in a single field. This enables additional processing of the content using ingestion pipelines.
- Extraction rules. Pull the exact content you want for a given field from any targetable HTML element. Makes it easier to extract web content without full access to the HTML source files.
- Programmatic scheduling. You can now register multiple schedules with different crawl configurations, overriding the default scheduling configuration for a crawler.
- Event logs. Users can now view crawl event logs in Discover directly from the UI.
- The Elastic connector framework, released as a technical preview in 8.4, has now reached beta. The framework has improved resiliency, with support for handling suspended jobs, unexpected stops, and transient errors.
4 existing connector clients are now in beta:
5 new connector clients are now in beta:
- Developers can now run end-to-end tests on connector clients, against real data sources, in an isolated environment. Docker Compose manages the setup of the dev environment, including both the mock Elastic instance and mock data source.
Integrate the new Behavioral Analytics (beta) feature to capture user behavior from your website or applications.
- Support for common behavioral events, like page views, search events, and click interactions.
- Deep integration with Search UI.
- Data lives in Elasticsearch, which enables you to easily build visualizations or analyze activity in Discover.
- Fixed out of memory errors for Docker container deployments for 2 GB instances, including Elastic Cloud, ECE, and self-managed Docker images.
- Fixed a bug where connector packages could not be connected if hosted at non-root URLs.
- Fixed an issue where Sharepoint Online and OneDrive content sources in Workplace Search needed to be connected by an authenticated Azure Global Administrator, and could not simply leverage permissions granted by an administrator.
- Fixed a bug where XML files could not be downloaded and processed by the Elastic web crawler.
- Fixed a bug where Elastic Cloud deployments would not downscale unused Enterprise Search infrastructure.
- Fixed a potential SSRF vulnerability in the Workplace Search Zendesk connector.
Enterprise Search connectors-related indices (
.elastic-connector-sync-jobs-v1) are no longer dynamic, to avoid potential data problems.
- Disabled running native connectors on images with less than 4GB of RAM, to avoid out of memory errors.
Fixed a bug where Native Connectors running in Elastic Cloud produced
debuglogs by default.
- Fixed an Analytics and API Logs issue in Elastic Cloud deployments, which prevented analytics and logs from being recorded.
- Added a UI warning about in-progress syncs, and in-progress syncs are now canceled when a connector is deleted.
- Added the ability to set labels on pipeline inference configuration for zero shot models.
- Password fields are now hidden by default in the connector configuration UI.
- Fixed a formatting bug which caused a Painless script for canceling syncs to break.
- Fixed a bug in the index details page where data for the index being viewed wouldn’t appear until the first poll.
- Fixed a bug to ensure the copy button for cURL requests appears on the connectors pipeline page.
- Removed synchronization errors from the connector configuration screen, to only show connector errors.
- Added localization for previously unlocalized text and numbers on the connector documents page.
- Disabled the ability to set a sync schedule for connectors, if synchronization is disabled.
- Specific connector filtering features are now hidden from the user, if not enabled.
- When using Elasticsearch index engines, fields missing subfields are now flagged as incomplete.