본문으로 건너뛰기
버전: docs v25.02

Edge App Release Notes

Edge App v3.6.0 (April 30, 2025)​

Bug fixes

  • Fixed stream deployment failure issue caused by the name of stream.
  • (GCP) Resolved the issue of Edge App versions v3.4.0 and v3.5.0 not functioning properly in the GCP environment.
  • (Windows) Fixed issues that occurred when multiple inference target files were triggered simultaneously.

Edge App v3.5.0 (Mar 25, 2025)​

Feature

  • Supports inference file trigger when re-uploading the same file

Improved

  • Improved to use less memory when transmitting inference result data to Edge Conductor.
  • Added storage manager layer to support and expand various platforms (aws, gcp, wsl) in a consistent manner.
  • (Cloud) The performance of the feature using the S3 library functions in the Edge App has been improved.

Bug fixes

  • Fixed a bug where Edge App judged ALO deployment as successful even though deployment was not completed during ALO deployment.
  • Added exception handling for cases where ALO fails to deliver messages to Edge App when an error occurs.

Edge App v3.4.0 (Jan 24, 2025)​

Feature

  • Applied Edge App Viewer: Developed an integrated version of the previously used tiny viewer to provide a more user-friendly UI.
  • Support for scheduled deployments at Edge App startup.
  • Installation of Edge App via Helm: By applying Helm packages, it enables efficient installation/deletion, resource management, and future updates of Edge App.
  • ALO Deploy Timeout Setting: Allows flexible adjustment of Timeout based on the size of the model docker image.
  • Inference Result Filename Delivery: Transmits the filename of the input file being inferred to the Edge Conductor UI along with the result, allowing users to verify the input file.

Improved

  • Improved web socket structure, leading to strengthened software architecture and enhanced performance.
  • Configured to allow parallel processing by creating threads when inferring/transmitting large files.
  • Improved the structure for managing environment variables.

Bug fixes

  • Fixed the issue of redeploying when the web socket is disconnected.
  • (Windows) Fixed the issue of not being able to find the input file in On-Premise WSL environments.
  • (Windows) Enhanced memory management efficiency in Edge App within On-Premise WSL environments.

Edge App v3.3.0 (Dec 4, 2024)​

Feature

  • Users can now configure the Edge Conductor to not upload inference artifacts files.
  • (GCP) Edge App is now executable within GCP.
  • (AWS) If EFS is present, Edge App logs can be stored through EFS.
  • (Windows) To optimize local storage, only the two most recent ALO docker images will be retained, with older images being deleted.
  • (Windows) You can start and close the Edge App from the Edge Viewer.
  • (Windows) You can check the ready state of the Edge App within the Edge Viewer.

Improved

  • The performance of the Edge App has been improved for WebSocket network connectivity.
  • (Windows) The Edge Viewer has transitioned from being based on Microsoft Edge to an Electron-based version.
  • (Windows) The UI design of the Edge Viewer has been improved.

Bug fixes

  • Fixed the issue where multiple ALO pods were being created for each Edge App instead of just one.
  • (Windows) Fixed the issue where deployments would fail if an Edge App was left running for more than a day.

Edge App v3.2.0 (Sep 23, 2024)​

Feature

  • An Edge App update feature has been added to Edge App.
  • When registering Edge App in edge conductor, a 'note' field has been added for users to identify the Edge App.
  • (Windows) By adding the Edge App icon to the start menu, Edge App can be launched upon reboot.

Improved

  • The serial-id used to identify Edge App in edgeconductor has been modified to be unaffected by environmental factors.
  • (Windows) The tact time for bolt and vision solutions has been improved to be within a maximum of 2 seconds.
  • (Windows) The icon images resolution for 'Edge Start', 'Edge Stop', and 'Tiny Viewer' on the desktop have been improved.

Bug fixes

  • (Cloud) Fixed an issue where inference results were not sent to edgeconductor when specific mode was set.
  • (Windows) Fixed an issue where Edge App would not function when detecting inference files.

Edge App v3.1.3 (July 7, 2024)​

Bug fixes

  • Modified to generate a unique Edge App serial number for multiple namespaces using the same conductor.
  • Among the three data output management modes 'overwrite', 'add_utc', and 'add_date', currently only 'add_utc' is supported for uploading data to the edge conductor. Modified to support all modes.

Edge App v3.1.1 (June 6, 2024)​

Feature

Added Redis communication method for sending various statuses to the Edge Conductor in all versions. In the cloud version, provided an installation script that allows users to install with minimal information input.

Improved

Modified all versions so that the inference status can be continuously viewed in the Edge Conductor while inferences are in progress. Changed cloud version to accept user input for the available memory capacity of the inference pod. Modified the cloud version so that the inference paths underneath the user-inputted S3 output path are mounted.

Edge App v3.0.1 (June 3. 2024)​

Feature

  • Add a feature to report errors to Edge Conductor when inference fails in ALO.
  • Add a feature to show in the UI that inference is in progress in ALO via Edge Conductor.
  • For the WSL version, add a feature to detect files in subfolders for data input paths.
  • Integrate the Tiny Viewer functionality into Edge Viewer and remove Tiny Viewer (when the browser size is reduced, Edge Viewer changes to Tiny Viewer).

Improved

  • For the WSL version, process data input triggers using a separate PowerShell file.
  • Update the WSL image version from v2.3.0 to v3.0.0 (in v3.0.0, the Edge App operates based on k3s instead of docker compose).
  • Complete the first phase of WSL packaging (change the icon, run the NSIS installer).

Bug fixes

  • Fix a bug in the WSL version where some files are missing when multiple files are imported simultaneously in PowerShell.

Edge App v2.3 (May 24. 2024)​

Feature

  • The Edge App can run on both Windows and Cloud platforms.
  • The Edge App version and in & out paths are displayed on the Edge Conductor.
  • The inferencing status is displayed on the Edge Conductor.
  • Inferencing fail information is shown in the notifications of the Edge Conductor.
  • The ALO running on the Edge App can also be executed on other types of nodes (servers) in the Cloud. The infrastructure can be selected based on the amount of data to be inferred.
  • In the Windows version, inference results can be monitored through a web browser.

Improved

  • Inferencing continues even if the connection with the Edge Conductor is lost.
  • Inference results are buffered and not lost even if the connection with the Edge Conductor is lost. The saved data is transmitted when the connection is restored.

Internal

  • The codes for SEH, Windows, and Cloud versions have been unified, and the entire structure has been made into containers. The structure has been designed to allow for automatic updates through the update center in the future.
  • Whenever an issue arises with each container, it broadcasts the issue to the Redis server, and the Controller detects it and sends it externally.

Edge App v2.3 (Mar 31. 2024)​

Feature

  • Restructure to execute AWS commands only in the cloud environment.
  • Change the log level from "error" to "warning" when there are no logs to send during log transmission.

Improved

  • Share Docker information between Docker Compose files to improve performance.

Internal

  • Apply edge conductor as a mock for writing test code.
  • Rename the outermost container of edgeapp to "frame" from "edge app".

Edge App v2.2.x (Feb 02, 2024)​

Features

  • Modify Edge App docker Architect to DinD
  • Add Edge App update from the Edge Conductor
  • Add Cloud Edge App driver
  • Add WSL distro image, install script

Bug fixes

  • Fix the issue about Deploy-Deploy Sequence