databricks-mlflow-tracing

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Databricks MLflow Tracing

Databricks MLflow 追踪

Use this skill for MLflow observability and tracing workflows on Databricks. Read
databricks-core
first for CLI authentication, profile selection, and workspace discovery. On Databricks, reuse the existing experiment ID or choose an absolute workspace path such as
/Users/<email>/my-experiment
or
/Shared/my-experiment
; relative names in the bundled onboarding examples are for non-Databricks tracking servers.
Read the matching guide directly:
RequestRead this guide
Choose a use case or get started with MLflowMLflow onboarding
Instrument an app or agent and verify tracingMLflow tracing
Retrieve traces by ID, time, status, or metadataRetrieve traces
Inspect one trace or explain an agent failureAnalyze a trace
Reconstruct or debug a multi-turn chat sessionAnalyze a chat session
Query token, latency, or quality metrics on a supported OSS MLflow serverQuery MLflow metrics
For agent evaluation, scorer creation, or implementing a fix after diagnosis, use
databricks-mlflow-evaluation
.
For work that spans multiple guides, use this order:
  1. Determine the use case with MLflow onboarding.
  2. Add or verify tracing with MLflow tracing.
  3. Retrieve traces before trace or chat-session analysis.
  4. Use Query MLflow metrics for aggregate trends.
For Databricks trace storage, read the tracing guide's Databricks reference before claiming that traces use Unity Catalog storage.
For a Unity Catalog trace location, verify the deployed identity rather than a local user. It needs experiment access,
CAN USE
on the SQL warehouse,
USE CATALOG
and
USE SCHEMA
, and explicit
MODIFY, SELECT
on each generated spans, logs, metrics, and annotations table.
ALL PRIVILEGES
alone does not satisfy those table grants. Confirm persistence by reading back a fresh trace written under the deployed identity.
将此 skill 用于 Databricks 上的 MLflow 可观测性和追踪工作流。 先阅读
databricks-core
,了解 CLI 身份验证、配置文件选择和工作区发现。 在 Databricks 上,复用现有的 experiment ID,或选择绝对工作区路径,例如
/Users/<email>/my-experiment
或
/Shared/my-experiment
;内置入门示例中的相对名称适用于非 Databricks 追踪服务器。
直接阅读对应指南:
请求阅读此指南
选择使用场景或开始使用 MLflowMLflow 入门
为应用或 Agent 添加追踪并验证追踪效果MLflow 追踪
按 ID、时间、状态或元数据检索追踪检索追踪
检查单个追踪或解释 Agent 故障分析追踪
重建或调试多轮聊天会话分析聊天会话
在受支持的 OSS MLflow 服务器上查询 token、延迟或质量指标查询 MLflow 指标
如需进行 Agent 评估、创建评分器,或在诊断后实施修复,请使用
databricks-mlflow-evaluation
。
对于涉及多个指南的工作,请按以下顺序操作:
  1. 通过 MLflow 入门 确定使用场景。
  2. 通过 MLflow 追踪 添加或验证追踪。
  3. 在进行 trace 或 chat-session 分析之前,先检索追踪。
  4. 使用 查询 MLflow 指标 分析聚合趋势。
关于 Databricks 追踪存储,请在断言追踪使用 Unity Catalog 存储之前,阅读追踪指南中的 Databricks 参考。
对于 Unity Catalog 追踪位置,应验证部署身份,而不是本地用户。该身份需要 experiment 访问权限、SQL warehouse 上的
CAN USE
权限、
USE CATALOG
和
USE SCHEMA
权限,以及每个生成的 spans、logs、metrics 和 annotations 表上的明确
MODIFY, SELECT
权限。仅授予
ALL PRIVILEGES
无法满足这些表权限要求。请在部署身份下读回新写入的追踪,以确认持久化。