缺陷修复（Bugfix）

1. 修复 v1.21.0 版本全新安装过程中安装失败的问题。

本次版本更新聚焦于「LLM 可观测」模块，以下若无特殊说明，均为「LLM 可观测」相关功能。

新增特性（Feature）

1. 「数据集 - 实验」模块，新增多个实验对比功能：
   1. 对比多个实验，同一个评估维度的平均分，进行整体评估。
   2. 对比同一范例不同实验的 Output 内容、自动评估结果对比。
2. 「Prompt 管理」：同一个 Prompt 的不同版本，可以一键对比 Prompt 内容的差异。
3. 「应用分析」模块：新增「关联功能」页面，快速展示当前应用的数据都被哪些功能所使用。
4. 「人工评估」、「自动评估」、和「Prompt 管理」模块的列表，新增与之相关的“关联应用”或“关联数据集”。使得不同模块之间的数据关联更加清晰。

缺陷修复（Bugfix）

1. 修复主机部分容器没有展示出来的问题。
2. 修复内置应用 License 也统计的问题。
3. 修复前端页面展示英文的问题。
4. 修复 Dify 上报数据兼容性的问题。

重大更新（Breaking Changes）

缺陷修复（Bugfix）

1. 修复安装过程中检测端口号异常的问题。
2. 修复启动脚本指定服务重启命令失效的问题。

重大更新（Breaking Changes）

新增特性（Feature）

1. 管理：
   1. 新增「License 管理」模块：对不同模块核心资源量的使用进行管控和统计。
      

      注意：从1.17.0及之前的版本，直接升级到 1.18.0及以上版本，需要提前手动清空所有数据才可正常使用。如遇此情况，请联系技术支持。

   2. 新增「数据上报地址」模块：可以统一修改平台多个模块的数据配置信息，用于解决自动识别的数据上报地址与实际不符的情况。
2. 仪表板：
   1. 数字概览图表，可以配置映射规则，对查询值赋值、并配色。
   2. 折线图、柱形图图表，可以配置阈值标记、并配色。
   3. 仪表板列表支持批量导出 JSON。
3. LLM：
   1. 「会话分析」新增「Spans 列表」页面：展示每条 Span 的基本信息与性能数据，单击即可查看详情信息。
   2. 「会话分析」的 Traces 和 Spans 列表页，增加快捷筛选，可以根据类型、名称、使用模型、状态快速筛选出目标数据。

新增特性（Feature）

1. APM：
   1. 新增「自定义业务接口聚合规则」：可以将多个接口调用识别为同一个接口，有效地帮助客户从技术视角切换到业务功能视角，去观测接口的性能问题。
2. LLM：
   1. 新增「拓扑」模块： 为 LLM 应用提供了一个全面且直观的调用关系、性能状况全景图。通过展示所有 Span 节点之间的调用关系及聚合后的性能指标，能够快速洞察 LLM 应用的整体运行状况、聚焦问题节点，精准定位问题，从而实现高效优化与管理。

Features

• Dashboard:
  • Support querying LLM data and creating various custom charts.
  • Added view variables to enable switching variable values in dashboards, which can be applied in real-time to multiple charts. This helps users customize quick filtering and data analysis.
  • Support for PromQL query statements.
  • Added new chart type: Leaderboard.

Features

• Logs: Support full-text and fuzzy search for log messages. Searched content will be highlighted in blue within the log list or hover tooltips.
• Dashboard:
  • New chart types: Number Overview, Pie Chart, and Text.
  • When selecting metrics, field names and metric descriptions are displayed simultaneously to help users quickly select desired data metrics.
  • Automatically matches appropriate units after metric selection. When users manually modify units, appropriate data conversion is performed.
  • Support for custom fill colors.
• RUM: Quick filtering and search capabilities for performance event information in the View details page.
• Alerts: Alert policy details page now includes statistics and filtering capabilities for different severity levels of alert records.
• Added settings buttons to Dashboard, RUM, and LLM applications for quick access to settings pages.

Features

• Metrics Management: Added "Metrics Management" module to display all collected metric data. Currently supports metrics from the infrastructure monitoring module.
• ITIM: Added performance trend charts to host, container, and process detail pages for intuitive performance monitoring. Container and process detail pages also show performance metrics of associated hosts.
• Alert: Added alert policy detail page to display policy attributes, alert records, and latest configuration information.

New Features

• APM:

1. Added error information display to the topology graph in the "Resource Analysis" module.
2. Added multiple user experience improvements.

Bug Fixes

1. Fixed an issue where clicking on log alert record details in "Smart Alerts" resulted in an error.
2. Fixed incorrect unit display in the application list of "LLM Applications".
3. Fixed an issue where the application list in "LLM Applications" was not filtering by custom time range.
4. Fixed an issue where the Top 10 list in "LLM Applications" was not filtering by custom time range.
5. Fixed incorrect trace count display on the probability page in "LLM Applications".
6. Fixed abnormal trace filtering issues in "Service Details".
7. Fixed incorrect duration display for top-level spans in some "APM" flame graphs.
8. Fixed an issue where clicking on the honeycomb chart in "Infrastructure Monitoring" cleared the left-side filters.

Bug Fixes

1. Fixed an issue where CPU load statistics failed to calculate for hosts in "Infrastructure Monitoring".

Breaking Changes

Features

• APM: "Resource Analysis" module, added performance metrics display on hover for topology connection lines, showing metrics like call percentage, call count.
• LOG: One-click navigation from single log details to associated trace details page.

Bug Fixes

Fixed an issue where host statistics for infrastructure monitoring were failing.

Bug Fixes

1. Fixed the issue where users could not log in to the Lighthouse platform between 11 PM and 8 AM.

New Features

• APM:

1. Added "Service Dependencies" module, displaying collected resource performance data and call relationship topology.
2. Added associated log list display in trace detail page.

Bug Fixes

1. Fixed light-agent installation failure issue

Features

• APM: Added service call topology visualization for individual traces.

Features

• RUM:

1. Application creation page: Added two-way binding between input fields and default code.
2. Added support for collecting and displaying geographical location fields, including country, province, and city.