4.0.4 (November 2019)¶
- Fixed an issue where some execution events tracked by Domino could be logged or presented out of order.
- Fixed an issue where Domino executions with Spark integration could create Kubernetes resources in the wrong namespace.
- The console output panel for a Domino run will now surface and display more types of errors.
4.0.3 (October 2019)¶
With the release of Domino v4.0.3, Datasets functionality has been added to the new platform infrastructure. A feature first introduced in V3.3 is now accessible with the updated architecture.
- Fixed issue where the “modified” column in the Environments and Models table of the UI wouldn’t sort chronologically.
- Restores support for connecting to VPNs from Run containers
- Various minor bug fixes and stability improvements
4.0.2 (October 2019)¶
- Fixed issue where Model API’s timeout override was not taking effect
- Fixed issue where Control Center could become inaccessible when a job’s queue end time and run end time are the same time stamp.
- Various additional minor bug fixes and stability improvements
4.0.1 (October 2019)¶
- Multiple minor bug fixes and adjustments to the default configuration settings for new deployments
4.0.0 (September 2019)¶
Welcome to Domino 4!
In addition to helpful new features for data scientists and project leaders, Domino 4 introduces a new architecture with all components running on Kubernetes. This change makes Domino easier to install, configure, monitor, and administer, and allows Domino to run in more environments than ever before. Visit admin.dominodatalab.com to learn about the technical design of Domino 4 and read guides for configuration and administration.
Domino 4.0 fully sunsets support for V1 environments. Previously, V1 environments had been demarcated with an asterisk when listed in your project settings environments list. Typically, these should not be present for Domino deployments which originated after the release v3.0.
Domino 4.0 fully sunsets support for legacy API endpoints. Only Model APIs are supported. Typically, legacy API Endpoints should not be present for Domino deployments which originated after the release v3.0.
Many previous interfaces and options for managing Domino executors (e.g. the legacy “Dispatcher” interface) have been replaced with the introduction of the new Kubernetes compute grid. There are new dashboards for viewing Kubernetes infrastructure and active execution pods, and new options for configuring Hardware Tiers.
Click to read more about Managing the compute grid in Domino 4.
Domino 4.0 removes support for SSH access to a Run container.
Domino 4.0 removes support for arbitrary Docker arguments for things like custom volume mounts.
Domino 4.0 temporarily removes support for connecting to VPNs from Run containers. Support returns in 4.0.3.
In Domino 4.0, user logins must use the new Keycloak authentication service. Any existing legacy LDAP integrations will need to have their configurations migrated to Keycloak.
Domino 4.0 ships with a new collection of Domino 4.0 standard environments. Users who want to use NVIDIA GPUs in Domino 4.0 will need to switch their compute environments to the latest version as Domino now utilizes NVIDIA Docker. Note that these new standard environments do not support working with GPUs in Python 2.
Domino now runs fully kubernetes native. Both front ends, central services and executors now run on the Domino kubernetes platform. Read more about the new infrastructure.
Domino 4.0 adds a new Assets Portfolio that allows users to quickly discover and see key information about the data products they have access to in Domino, including Model APIs, Apps, Launchers, and Scheduled Jobs.
A new Project Manager admin role is available. This role grants a user contributor access to projects owned by other users who are members of the same organization as the project manager. This allows the project manager to view those projects in the Projects Portfolio, discover their published assets in the Assets Portfolio, and view the projects’ contents as a contributor.
Domino 4.0 introduces Project Goals. Goals represent outcomes or subtasks within projects. Project contributors can link files, Workspace sessions, Jobs, Apps, and Model APIs to goals, which show up on the goal card in the project overview. This provides a way to track all work related to a specific goal in the project, and can make navigating large and busy projects easier.
New options are available in the Notifications and Workspace Settings sections of user Account Settings that allow for opt-in to email notifications or auto-termination for long-running Workspace sessions with a configurable duration.
Admins also now have additional options for defining which Workspace sessions to treat as long-running, enforcing notification requirements for users, and sending additional global notifications about long-running sessions to admins.
- Visual styling and design for tables, buttons, links, accordion headers, breadcrumbs, and tab navigation have all been improved and made consistent across the Domino application.
- Run usage functionality is impaired and will be addressed in an upcoming Domino version.