Added appropriate error messaging to the API when an organization admin attempts to approve a new user invite when the organization has already hit its user seat limit
Deprecated the app_name column from the role table, along with all related fields exposed through GraphQL
Fixed the deploy engine build process to allow it to trigger automatic package creation again
Creating a private user root folder now correctly creates a default ACE to ensure it’s properly provisioned under OLP conditions
Updated core-search to properly enforce OLP when called via API tokens or JWT tokens
Improved the cacheing performance of resource ACEs
UI Updates
Updated the viewers to show the cognition panel data loading skeleton immediately to prevent the page from visually “jumping” when the cognition data is finally loaded
Made three additional organization settings available during org creation to simplify initial org setup: Max TDO Assets, User Seat Count, and Enable Limit Enforced
Fixed cluster of bugs around transcription cognition output in the viewers, including missing line wrapping and accidental line breaks
Fixed bug preventing the transcription viewer from being able to switch between multiple transcription outputs on a single asset
Fixed a bug causing the form validation for the previously-removed admin seat limit to still trigger when submitting changes to the system settings for an organization
Updated the logic used when logging in through Okta to fix URL redirects not being honored
Engine Updates
Thorn CSAM Engine Released - This engine provides the capability to detect categories of explicit content including Child Sexual Assault Material (CSAM) and pornography. Access to this engine is controlled, so please ask your instance administrator about it if you’re interested.
FFMPEG engine supports custom output folder
Jobs for the Custom FFMPEG Engine can now specify an S3 location where the transcode output will be saved. This is an override to the default behavior of attaching the output as an asset on the source TDO.
Improve Processing Pod Startup
Processing pods could stall during start-up due to registration delays in edge-controller. The startup probe timing has been extended and made configurable.
Allow retry of Processing Task status updates
The status reporting for processing tasks was made more robust by adding retry logic for cases where the update task calls fail with 5XX errors.
