2020 年 4 月April 2020

这些功能和 Azure Databricks 平台的改进已于 2020 年 4 月发布。These features and Azure Databricks platform improvements were released in April 2020.

备注

发布分阶段进行。Releases are staged. 在初始发布日期后,可能最长需要等待一周,你的 Azure Databricks 帐户才会更新。Your Azure Databricks account may not be updated until up to a week after the initial release date.

每个 Azure Databricks 工作区的唯一 URLUnique URLs for each Azure Databricks workspace

2020 年 4 月 28 日April 28, 2020

Azure Databricks 为每个工作区添加了新的唯一 URL。Azure Databricks has added a new unique URL for each workspace. 这一新的每工作区 URL 采用以下格式:adb-<workspaceId>.<random number>.databricks.azure.cnThis new per-workspace URL has the format adb-<workspaceId>.<random number>.databricks.azure.cn. 此 URL 是对目前为止用于访问工作区的现有区域 URL (<region>.databricks.azure.cn) 的补充。This URL is complementary to the existing regional URLs (<region>.databricks.azure.cn) that you have used up to now to access your workspaces. 这两个 URL 都继续受到支持。Both URLs continue to be supported. 但是,我们强烈建议你使用新的每工作区 URL,因为它与区域无关。We do, however, strongly recommend that you use the new per-workspace URL, because it is region independent.

如果你有使用旧区域 URL 的现有脚本或自动化工具,我们建议你更改脚本,使用新的每工作区 URL。If you have existing scripts or automation tools that use the old regional URLs, we recommend that you change your scripts to use the new per-workspace URLs. 我们还建议你在以后的任何脚本或自动化工具中使用每工作区 URL。We also recommend that you use the per-workspace URL in any future scripts or automation tools. 但是,由于 Azure Databricks 仍将支持现有区域 URL,因此,在新的 URL 推出后,现有代码仍将有效。However, because Azure Databricks continues to support the existing regional URLs, existing code will continue to work after the new URL rollout.

有关详细信息,请参阅从旧的区域 URL 迁移到每工作区 URLFor details, see Migrate from legacy regional to per-workspace URLs.

MLflow 跟踪 UI 增强功能MLflow tracking UI enhancement

2020 年 4 月 23 - 30 日:版本 3.18April 23-30, 2020: Version 3.18

MLflow UI 现在提供了一种方法,用于在删除根运行时删除子运行。The MLflow UI now offers an option to delete child runs when you delete a root run.

笔记本可用性提升Notebook usability improvements

2020 年 4 月 23 - 30 日:版本 3.18April 23-30, 2020: Version 3.18

此版本在使用笔记本时引入了几个可用性方面的改进:This release brings several usability improvements when working with notebooks:

  • 现在可以分别对上一个和下一个单元格使用 Shift + 向上箭头或向下箭头来选择相邻的笔记本单元格 。You can now select adjacent notebook cells using Shift + Up or Down for the previous and next cell respectively. 可以复制、剪切、删除和粘贴多选的单元格。Multi-selected cells can be copied, cut, deleted, and pasted.
  • 删除单元格时,默认情况下会显示一个用于确认删除操作的对话框。When you delete a cell, by default a delete confirmation dialog displays. 现在,可以在删除单元格时禁用确认对话框,方法是选择“不再显示此消息”复选框并单击“确认” 。You can now disable the confirmation dialog when you delete the cell by selecting the Do not show this again checkbox and clicking Confirm. 你还可以切换确认对话框,只需使用“帐户”图标 >“用户设置”>“笔记本设置”中的“启用命令删除确认”选项即可 。You can also toggle the confirmation dialog with the Turn on command delete confirmation option in Account Icon > User Settings > Notebook Settings.

使用 Azure Active Directory 令牌向 Databricks REST API 进行授权(公共预览版)Use Azure Active Directory tokens to authorize to the Databricks REST API (Public Preview)

2020 年 4 月 23 - 30 日:版本 3.18April 23-30, 2020: Version 3.18

现在,可以使用 Azure Active Directory 令牌向 Databricks REST API 2.0 授权。You can now use Azure Active Directory tokens to authorize to the Databricks REST API 2.0. 使用 Azure Active Directory 令牌,可以对新工作区的创建和设置进行自动化。Azure Active Directory tokens enable you to automate the creation and setup of new workspaces. 服务主体是 Azure Active Directory 中的应用程序对象。Service principals are application objects in Azure Active Directory. 还可以使用 Azure Databricks 工作区中的服务主体自动执行工作流。You can also use service principals within your Azure Databricks workspaces to automate workflows.

有关详细信息,请参阅使用 Azure Active Directory 令牌进行身份验证For details, see Authentication using Azure Active Directory tokens.

单个用户可以查看直通群集上的 Apache Spark 日志Single user can view the Apache Spark logs on a passthrough cluster

2020 年 4 月 23 - 30 日:版本 3.18April 23-30, 2020: Version 3.18

以前,如果用户不是管理员,标准模式群集上的凭据传递就不允许该用户查看 Apache Spark 日志。Previously, credential passthrough on a Standard mode cluster did not allow a single user to view the Apache Spark logs if they were not an admin user. 现在,对于具有附加权限的用户,即使该用户不是管理员,也可以查看 Spark 日志。Now a single user with attach permission can view Spark logs even if they are not an admin.

Databricks Connect 现在支持 Databricks Runtime 6.5Databricks Connect now supports Databricks Runtime 6.5

2020 年 4 月 20 日April 20, 2020

Databricks Connect 现在支持 Databricks Runtime 6.5。Databricks Connect now supports Databricks Runtime 6.5.

Databricks Runtime 6.1 和 6.1 ML 支持结束Databricks Runtime 6.1 and 6.1 ML support ends

2020 年 4 月 16 日April 16, 2020

对 Databricks Runtime 6.1 和用于机器学习的 Databricks Runtime 6.1 的支持已于 4 月 16 日结束。Support for Databricks Runtime 6.1 and Databricks Runtime 6.1 for Machine Learning ended on April 16. 请参阅 Databricks 运行时支持生命周期See Databricks runtime support lifecycle.

Databricks Runtime 6.5 正式版Databricks Runtime 6.5 GA

2020 年 4 月 14 日April 14, 2020

Databricks Runtime 6.5 引入了许多库升级和新功能,其中包括:Databricks Runtime 6.5 brings many library upgrades and new features, including:

  • Delta 表中所有写入、更新和删除操作的操作指标现在显示在表历史记录中Operation metrics for all writes, updates, and deletes on a Delta table now appear in table history
  • 可以对在 Delta Lake 流式处理微批次中处理的数据进行速率限制You can rate-limit the data processed in Delta Lake streaming micro-batches
  • Snowflake 连接器已更新为 2.5.9Snowflake connector is updated to 2.5.9

有关详细信息,请参阅完整的 Databricks Runtime 6.5 发行说明。For more information, see the complete Databricks Runtime 6.5 release notes.

用于机器学习的 Databricks Runtime 6.5 正式版Databricks Runtime 6.5 for Machine Learning GA

2020 年 4 月 14 日April 14, 2020

Databricks Runtime 6.5 ML 引入了以下库升级:Databricks Runtime 6.5 ML brings the following library upgrade:

  • MLflow 已从 1.5.0 升级到 1.7.0MLflow upgraded from 1.5.0 to 1.7.0

有关详细信息,请参阅完整的 Databricks Runtime 6.5 ML 发行说明。For more information, see the complete Databricks Runtime 6.5 ML release notes.

用于基因组学的 Databricks Runtime 6.5 正式版Databricks Runtime 6.5 for Genomics GA

2020 年 4 月 14 日April 14, 2020

用于基因组学的 Databricks Runtime 6.5 是基于 Databricks Runtime 6.5 构建的。Databricks Runtime 6.5 for Genomics is built on top of Databricks Runtime 6.5. 有关此版本中包含的改进和库的详细信息,请参阅完整的用于基因组学的 Databricks Runtime 6.5 发行说明。For details about the improvements and libraries included in this release, see the complete Databricks Runtime 6.5 for Genomics release notes.

Databricks Runtime 6.0 和 6.0 ML 支持结束Databricks Runtime 6.0 and 6.0 ML support ends

2020 年 4 月 1 日April 1, 2020

对 Databricks Runtime 6.0 和用于机器学习的 Databricks Runtime 6.0 的支持已于 4 月 1 日结束。Support for Databricks Runtime 6.0 and Databricks Runtime 6.0 for Machine Learning ended on April 1. 请参阅 Databricks 运行时支持生命周期See Databricks runtime support lifecycle.