Release notes

Users can now add a buffer area filter around nodes in a graph while in map mode. This filter behaves like the Shape filter, affecting only the layers on the graph without filtering the nodes themselves.

Users are now prompted to overwrite or rename a graph when a single duplicate title is detected. If multiple duplicates exist, users can choose to create a new graph or change the title.

If an import partially fails, users are now prompted to either delete or keep the partially imported data before exiting. This helps to prevent duplicate record imports.

Fixed an issue where no notification appeared after an error occurred when using the Add to graph button in the Upload Center.

Fixed an issue where polygon shapes drawn as filters on the map remained visible even after the corresponding filter pill was disabled.

Fixed an issue where grouped edges could overlap during undo or redo operations in the graph.

Fixed an issue where Record-as-Relation (RaR) expansions did not respect the graph browser’s expansion limit.

Fixed an issue where inverse relations were ignored when adding a middle-entity document as a node in a RaR graph.

Fixed an issue where toolbar buttons became disabled when rapidly undoing or redoing actions.

Fixed an issue where the filter editor did not display phrase filters correctly.

Fixed an issue where valid error messages were not visible when adding fields from a spreadsheet.

Graph Browser can now be configured to point to an instance of the Enhanced Coordinate Map. When configured, it renders all pre-configured layers in map view mode. For more details, see configuring map view.

Records-as-relations now support relations to EIDs as incoming or outgoing relations. Upon adding a document from an entity type that is the middle entity of a record-as-relation, it will be added as an edge if all the relations contained in the record-as-relation are hidden relations.

Graph edges between nodes can now be grouped into a single-edge or per-direction group. This helps reduce visual clutter in the graph and makes it easier to analyze relationships between nodes. Grouping edges in the graph

Added an upload center to Siren Investigate that allows users with limited access privileges to add data. These uploads are stored in isolated data projects. For more information, see Upload center.

Dropped support for Elasticsearch versions earlier than 7.17.9. For more information, see version compatibility matrix.

The Elasticsearch client has been partly migrated to a more recent version.
The getClusters() API no longer accepts a parameter (it previously accepted admin or data).
Update any custom plugins to the following pattern:

const cluster = server.plugins.elasticsearch.getCluster();
const adminClient = cluster.getAdminClient();
const resp = await adminClient.get({ index, id})

const dataClient = cluster.getDataClient();
const resp = await dataClient.get(request, { index, id})

Removed legacy migrations for performance reasons. Direct upgrades to 15.0.0 from significantly earlier versions are no longer supported. Recommended path: first upgrade to the latest 14.9.x, then upgrade to 15.0.0.

Changed the default highlighter from unified to plain to improve performance. Users wishing to continue using the unified highlighter can set the default:highlight-type setting in Advanced Settings. For more information on the highlighter types, see the highlighting section of the Elasticsearch documentation.

In this release, we have completely removed support for the legacy tile_map visualization.
This visualization type was deprecated some time ago. If you still have tile_map visualizations in your system, please migrate them to the Enhanced Coordinate Map visualization before upgrading.

Graph Browser contextual scripts have been removed. The contextual menu of the Graph browser no longer supports scripts of type contextual. This change was made to simplify the Graph browser’s functionality and improve performance. If you were using these scripts, you will need to reimplement them using SirenAPI scripts. Existing contextual scripts will not be deleted, but they will no longer be executed.

Removed deprecated Graph Browser siren APIs:

The data import wizard will now read Excel date/time cells exactly as they appear in a spreadsheet (after their date format is applied). If you used saved transformers with Excel files, please use the new "Date formatting" option in the ellipsis button (...) menu to recognize the format and assign the time zone for your date columns, then save the transformers again.

Search Guard Classic is not supported on Elasticsearch 8. New installations or upgrades from Elasticsearch 7.x must use either Elastic Security or Search Guard FLX.
See Search Guard FLX or Elastic Stack security overview for more information.

Made the width of the icon picker more consistent across Investigate.

Reworked how formatted fields are handled in graphs to reduce the saved object size.

Performance: Removed an unnecessary fixed cost when updating the selection table.

Improved the layout of the records-as-relation modal.

The graph timebar now has a new dots visualization to show time data more granularly. For more details, see here Timebar

Optimized collections and dashboards fetching in the search app, resulting in faster initial load times.

Bumped node to 22.20.0. For more details, see https://github.com/nodejs/node/blob/main/doc/changelogs/CHANGELOG_V22.md#22.20.0.

Fixed an issue where creating a timelion-sheet object caused an error.

Fixed an issue where selecting an icon created an EID without form submission.

Fixed an issue in Graph Browser where cards could appear blank if the panel was opened after a selection change.

Fixed an issue where the Select all checkbox in the relations list on the Graph Browser was not correctly checked.

Fixed an issue where the cloning of a very large dataspace could fail.

Fixed an issue where hidden nodes were incorrectly expanded.

Fixed an issue where an internal server error was thrown when deleting the icon pack.

Fixed an issue in the graph selection table so columns now sort correctly based on their data type.

Fixed an issue where node counts were not updated correctly after undo/redo operations.