v2.9.0 - 25/Mar/2026
New Features & Enhancements
AI Investigation Agent
The AI Investigation Agent can autonomously execute investigations into production incidents, identifying the most likely root causes and providing actionable remediation steps. It has access to all the tools offered by the MCP server, and uses an iterative approach to finding the root cause of an incident.
Liz, the AI Assistant
Liz serves as the intelligent, conversational front-end assistant for SUSE Observability. It acts as an orchestration layer, utilizing specialized tools from the MCP server to transform complex telemetry into guided dialogue.
-
Conversational Cluster Management: Allows users to find data using natural language, such as querying runtime states, crafting complex metric queries, or analyzing the impact of environment changes.
-
Automated Dashboarding: The agent assists in the creation of custom dashboards, streamlining the visualization of critical system health and performance metrics.
Model Context Protocol (MCP) Server
Description: The SUSE Observability Model Context Protocol (MCP) server provides AI Agents access to the information in SUSE Observability across all domains: topology, events, health, logs, traces and metrics.
-
Immediate Remediation: Coding agents can now combine CLI tool access with observability data to not only identify issues but also execute immediate fixes, such as performing a rollback using
kubectcommands. -
Closing the Developer Loop: By bridging the gap between infrastructure alerts and the codebase, the MCP Server enables agents to pinpoint specific bugs and implement fixes directly within the repository.
Infrastructure & Connectivity
-
Gateway API Support: The SUSE Observability Helm chart now supports the Kubernetes Gateway API as an alternative to a standard Ingress. For details, see the Ingress documentation.
-
Global flag: A single global flag is used to switch between a standard Ingress and the Gateway API.
-
Simplified configuration: Following Rancher-standard patterns, the configuration requires only the URL and TLS secret in most environments.
-
Bug Fixes
-
Health State Aggregation Logic: Improved the precedence logic for aggregated health states (e.g., at the Namespace or Cluster level). The Clear state now takes precedence over Unknown.
-
Impact: In scenarios where a component has many Unknown states and at least one Clear state (with no Critical or Deviating states), the overall health will now aggregate to Clear instead of Unknown.
-
-
Rancher Extension Snapshot API: Fixed an incompatibility issue within the Snapshot API request model to ensure stable data retrieval within the Rancher UI.