Release Notes: Time Series Analytics#

Version 2026.2#

Release Date: September 2026

This release introduces updated Intel GPU driver support, removed Model Registry integration, and hardened path handling, along with dependency upgrades and documentation improvements.

Improved

Intel GPU driver support: Updated the Intel® GPU driver/compute-runtime stack to version 26.27.39122, adding the corresponding install script logic for the new driver and IGC packages.
Deployment simplification: Removed leftover Model Registry integration references (MODEL_REGISTRY_URL environment variable and related config) from Docker Compose and Helm deployment artifacts, and updated unit tests to match.
Security: Bumped Kapacitor to 1.8.6, python-multipart to 0.0.32, and setuptools to 83.0.0 in requirements.txt.
Documentation: Updated release notes structure, GitHub branch references, and Helm deployment links across the user guide.

Fixed

Path traversal hardening: Added sanitize_dir_name() validation before deriving UDF working directories from configuration/environment input, addressing an SDLe-reported vulnerability in classifier_startup.py.

Version 2026.1#

Release Date: June 17, 2026

New:

Support for optional CPU core-pinning enabled with the CORE_PINNING environment variable, which allows the service to prefer specific core types (E-cores, P-cores, or low-power cores).
A /udfs/package API endpoint for uploading and extracting UDF deployment packages as tar archives via HTTP. The UDF updates do not require manual placement of files.

Improved:

The base Kapacitor Docker image is replaced by a Debian-based Python slim image with Kapacitor installed via .deb, reducing image size and improving flexibility.
Updated Intel® GPU drivers to support WCL (compute-runtime/IGC version 26.14.37833).
Updated the Kapacitor and Python library dependency versions.

Version 2026.0#

Release Date: March 27, 2026

This release improves deployment consistency, reliability, and documentation usability for Time Series Analytics.