Follow

Release Notes

2.1 (early August 2017)
 

Compute Environments V2 - General Availability

The new Compute Environments are now available for all customers that have the Kubernetes infrastructure installed. 

  • Environments are versioned
  • Environment versions are built once and then cached. This leads to more reliable performance leading to less time spent at the “Building” run state.
  • Past runs can be re-executed with the same environment as originally leading to better reproducibility
  • Environments can be used with Models (see below) to create immutable, reproducible, versioned models. 
  • Documentation: V2 Compute Environments

 

Configurable interactive session tools - General Availability

Many of our users have asked for custom interactive session tools beyond the current Domino offering of Jupyter, RStudio, PySpark, Zeppelin and H2O notebooks.  This new functionality gives Domino the flexibility to support any web based tool like sqlpad or JupyterLab.  Users can configure the tools available to a project via its Environment (v2 only), making it easy to:

  • Upgrade to a newer version of currently supported Domino notebooks like Jupyter or RStudio
  • Add new web based tools like sqlpad and JupyterLab
  • Manage the standard default tools for a team or organization across all projects 

See our support article for more information or contact your CSM.

 

SAS Studio support

Through the new functionality around configuring interactive session tools, Domino users can configure environments to make SAS Studio available in projects. This means SAS Studio will launch and sync back to Domino along with the other Domino-supported tools, such as Jupyter Notebooks and RStudio. Customers will need a SAS for Containers license in order to use this feature, and will need to work with their Domino CSM.

 

Direct run support for SAS scripts in Domino

Users can now execute SAS scripts in Domino via web UI and CLI by simply choosing a file to execute, just as they are already able to do with R and Python.

 

Limit hardware tiers by organization

Domino administrators are now able to limit access to specific hardware tiers by organization. This functionality can come in very handy granting access to a specialized set of resources for large projects or teams that need more horsepower.

 

No longer ignoring .rproj files in default .dominoignore rules

We have heard feedback from our customers that Domino should not be ignoring .rproj files - this is no longer the default behavior for new projects (for more information about how .dominoignore works, please see the "How to prevent syncing certain files in your project" help article).

 

 

Improvements to Model Manager (Beta)

We have the following major updates:

  • Resource management improvements to reduce chances of models not getting enough memory to startup 
  • Models from projects that have imported projects will now also import those projects by default. 
  • Models can now support cross origin requests and sample Javascript code was added for this. 
  • Models can now take request bodies of upto 10MBs.
  • Documentation: Model Manager Overview

*Please contact your customer success manager for help enabling this and beta-testing feature.

2.0 (early July 2017)

 

Model Manager (beta)

We have done a complete re-architecture of our “API Endpoints” functionality, which allows teams to deploy Python and R models in seconds as horizontally-scalable, high-availability API endpoints.

  • The new architecture supports web-scale use cases at low latencies
  • Sophisticated promote-to-production workflow
  • Provides improved security and access controls with “Models” now treated as a first-class entity, distinct from the Projects that developed them

*Please contact your customer success manager for help enabling this feature.

Shared Storage Support 

Domino now support shared storage for imported projects, significantly speeding up the use of data projects imported into experimentation projects. Users can manage the space being used in their shared storage through the Domino Admin interface.

Faster Finishing Times 

We've made some changes under the hood to significantly improve run finishing times.

Security Improvements 

We've made some improvements to user account security, including increasing password complexity requirements.

 

1.59 (June 2017)

Redesigned Interactive Sessions screen

We have redesigned and enhanced our Interactive Sessions screen to make it easier to use and sync with Domino. Some of the new functionality includes:

  • Improved status messaging, including surfacing when file changes in the session have yet to be committed to Domino
  • The ability to submit a commit message when stopping and committing changes from within an interactive session
  • An improved filter on the "files" section of your project to make it easier to find the revision you're looking for

Screen_Shot_2017-06-20_at_5.44.03_PM.png

 

 

1.56 (late April 2017)

Technology preview*: Configure new Interactive Session tools

Many of our users have asked for custom interactive session tools beyond the current Domino offering of Jupyter, RStudio, PySpark, Zeppelin and H2O notebooks.  This new functionality gives Domino the flexibility to support any web based tool like sqlpad or JupyterLab.  Contact your customer success manager if you would like use a new tool in Domino.

 

Technology preview*: New Model Management

We have done a complete re-architecture of our “API Endpoints” functionality, which allows teams to deploy Python and R models in seconds as horizontally-scalable, high-availability API endpoints.

  • The new architecture supports web-scale use cases at low latencies
  • Sophisticated promote-to-production workflow
  • Provides improved security and access controls with “Models” now treated as a first-class entity, distinct from the Projects that developed them
  • Read about it in our Domino 2.0 blog post

 

The following capabilities are also in technology preview:*

  • Environment Management V2

 

We have introduced Data as a new area of the product.

The addition of "Data" as first-class entity is a new concept we have introduced to simplify the data workflow for our data scientists and data science teams.

  • Easily create data sets from scratch or from within an existing project
  • Keep an inventory of all data sets in one location and easily find data sets available to you
  • Data Sets are automatically configured to "export" files so that projects can easily import them

 

User Experience Improvements

  • You can now see repository branch information without having to click to edit:

Screen_Shot_2017-05-03_at_6.35.42_PM.png

  • Made the save button more intuitive when editing a repository:

Screen_Shot_2017-05-03_at_6.40.06_PM.png

  • Results Consumers can now leave comments to provide feedback on files in a project
  • We fixed the drop target for uploading files so that it is now much easier to drag & drop files into a project

screenshot-2__1_.png

 

Infrastructure Improvements

  • Fixed a bug with pip-installs of private Github repos where git domain credentials were not working with no repository enabled
  • Fixed a bug where users could not batch run a Jupyter notebook that was nested in a folder
  • Fixed a bug where emails sent to invite collaborators to a project were not rendering custom messages with special characters properly

 *Please contact your customer success manager for help enabling these features


1.55 (mid April 2017)

The following capabilities are in technology preview*:

  • Environment Management v2

 

User Experience Improvements

  • Continued improvements to our technology preview of Environment Management v2
  • We have added Notebook Settings that give you the option to automatically shut down a notebook session after a set amount of time:

Screen_Shot_2017-05-03_at_4.19.43_PM.png

  • We fixed a UI issue with .zip and .gz files so that we no longer attempt to render these file types
  • Fixed a bug where navigating to a run URL was not directing you to the specified run

Infrastructure Improvements

  • Fixed a bug where Kerberos authentication was not working for collaborators
  • Fixed a bug where getting the status of a repository took a long time and blocked the ability to stop the run

 *Please contact your customer success manager for help enabling these features


1.54 (late March 2017)

The following capabilities are in technology preview*:

  • Environment Management v2

Two-factor Authentication is now available for all users and deployments.

This release introduced the ability for Administrators to require 2FA for specific users or entire deployments.  Contact your customer success manager if you would like this enabled in your account.

User Experience Improvements

  • Continued improvements to our technology preview of Environment Management v2
  • We removed the requirement for email when creating a new organization:

Screen_Shot_2017-03-30_at_1.26.12_PM.png

Infrastructure Improvements

  • Fixed a bug in our Git support where the repository did not finish cloning before the run started

 *Please contact your customer success manager for help enabling these features


1.53 (mid March 2017)

The following capabilities are in technology preview*:

  • Environment Management v2
  • Two-factor authentication is now available

User Experience Improvements

  • Continued improvements to our technology preview of two-factor authentication
  • Results Consumers can now see the API Endpoint Publish page with the status and version of any published endpoints as well as instructions for how to execute them:
    Screen_Shot_2017-05-03_at_3.17.43_PM.png

 

  • Moved the "Collapse Setup Output" option to the top of the page on the console output tab of the runs dashboard to make it easier to navigate to the logs and view results button at the bottom of the page:
    Screen_Shot_2017-05-03_at_3.24.21_PM.png

 

  • Fixed a bug where the link to an imported dataset in the runs details page was navigating to the wrong location
  • Fixed a bug so that now when you click "Discard" from the run pane, there is a confirmation modal just in case you were not intending to discard your results:
    Screen_Shot_2017-03-16_at_9.54.13_AM.png

 

 *Currently limited to customers on our cloud offering. Please contact your customer success manager for help enabling these features


1.52 (Late February 2017)

The following capabilities are in technology preview*:

  • Environment Management v2
  • two-factor authentication (2FA)

User Experience Improvements

For projects with an attached Git repository

  • Users now have the ability to specify the repository reference (commit id, branch, tag) directly from the advanced batch run dialog
  • From the runs dashboard, when the user selects "Re-run with Original Version" we use the originally specified repository reference
  • From the runs dashboard, when the user selects "Re-run with Latest Version" we use the most up-to-date repository reference

 *Currently limited to customers on our cloud offering. Please contact your customer success manager for help enabling these features


1.51 (mid February 2017)

All Domino-managed cloud environments now have our new first-class Git integration enabled. This release introduced the ability the to specify a branch, tag, commit, or custom git reference along with stability improvements. Thank you again to all of our beta testers!

The following capabilities are in technology preview*Environment Management v2

User Experience Improvements

  • Outputs in the interactive run tab now wrap so users do not have to scroll horizontally to see full output logs.
  • Continued improvements to our technology preview of V2 environments including the ability to change the visibility of an environment after it has been created. (Environments can be owned by a user, organization, or be globally accessible to all users.)

Infrastructure Improvements

  • Domino now logs the timestamps related to run states to enable administrators to perform meta-analysis on the health of their Domino cluster.

 * Please contact your customer success manager for help enabling these features


1.50 (early February 2017)

The following capabilities are in technology preview*:

  • Environment Management v2
  • Improved Git Integration and Support

User Experience Improvements

  • Continued improvements to our technology preview of new Git support including better error messages and styling updates of the run’s details pane
  • Continued improvements to our technology preview of V2 environments including user interface tweaks around navigation (including adding “Environments” to the top blue navigation bar to be more discoverable) and around how we treat archived environments. We also fixed a bug where environments with VPN were not able to be edited by non-administrators
  • We have moved our notification banners to the top right of the screen so the navigation bar is always accessible
  • Fixed a bug where a scheduled run could not be edited

Infrastructure Improvements

  • Domino administrators can now save time by overriding hardware tier hard drive disk sizes without having to create a new AMI. We’ve documented this process in our Cluster Controls knowledge base article
  • Added additional garbage collection functionality to help ensure executor disk drives are clean and have enough room for new runs

 * Please contact your customer success manager for help enabling these features


1.49 (late January 2017)

The following capabilities are in technology preview*:

  • Environment Management v2
  • Improved Git Integration and Support

User Experience Improvements

  • Improvements to the new environment management system including the ability to set environment variables and various user experience improvements
  • Improvements to git integration including the reconciliation of legacy and new Git Credentials on users’ account settings page
  • Added logic to ensure preparation and build logs always stream to a user
  • Jupyter startup logs no longer show up in red as if they were an error
  • You can now remove run tags

Infrastructure Improvements

  • Added configuration option to set the minimum amount of free space on an executor
  • Fixed an edge case where a successful run would be marked as “Failed”
  • Fixed an issue in the Domino Client (command line client) for Mac users
  • Fixed an issue where executor termination timeouts were not being acted on

 * Please contact your customer success manager for help enabling these features


1.48 (early January 2017)

Technology preview*: New environment management

  • We rebuilt Environments from the ground-up to support future expansion of Domino. Supported by new infrastructure for increased reliability. 
  • Improved linking between environments, runs, and projects.
  • Environment construction is more efficiently cached for faster and more reliable performance leading to less time spent at the “Building” run state
  • Documentation: V2 Compute Environments
  • Related blog post: Enabling Data Science Agility with Docker

UX Improvements

  • Re-organized the Projects Overview screen to include tabs for easier navigation for users with many projects
  • Related to the First-class Git integration* introduced in 1.47, we added the ability to authenticate using SSH keys and improved the documentation:
  • Fixed a bug where run output was not always displayed in the run details pane

Infrastructure Improvements

  • Added the ability for admins and support users to reorder the run queue from the Dispatcher view
  • Improved Domino Client (command line client) stability and made errors more user-friendly

* Please contact your customer success manager for help enabling this feature


1.47 (mid December 2016)

Technology preview*: First-class Git integration and support

For teams that rely on Git management systems, we wanted to support their experimental workflows

  • Public/Private GitHub (Enterprise) repository support from the Files page in a project
  • If using private repos, users can specify a Personal Access Token from the Accounts settings page (Git Domain Credentials)
  • Runs will alert and can be cancelled if Git repos have uncommitted changes when run is stopped

UX Improvements

  • New feature flag system to help Domino administrators enable features both globally and for specific users

Infrastructure Improvements

* Please contact your customer success manager for help enabling this feature


1.46 (early December 2016)

UX Improvements

  • Excel files are now rendered in the browser
  • Project import settings have been moved to the "Files" tab
  • You can now SSH into batch runs
  • Added note to Access & Sharing page clarifying that organizations can be added as collaborators to projects

Infrastructure Improvements

  • Improved executor and log retrieval reliability and performance
  • Improved reliability of connections for downloading and uploading project files
  • Fixed edge case where review requests were unable to be merged

1.45 (early November 2016)

Improved experience for results consumers featuring full-page file sharing and an improved versioning UI

UX Improvements

  • Added progress indicator to console output for runs during the “Preparing” stage; this is useful for projects with many and/or large files
  • We now prompt users to confirm that they want to archive a comment
  • Moved the ability to rename a project to the Settings page
  • Added an R (httr) example to the API endpoints examples

Infrastructure Improvements

  • API endpoint stability improvements
  • Improved executor stability and ability to recover from errors
  • Domino admins can now specify an AWS region when using SES for email
  • Domino admins can now specify user data for hardware tiers

1.44 (late October 2016)

Structured Statistics Improvements

Domino users can use structured statistics (e.g. R^2, p-value) to evaluate the performance of experiments. Two enhancements to the dominostats.json facilitate easier consumption of results:  

  • dominostats.json rendered as a table in results pages and the results dashboard
  • dominostats.json visually diff’ed in runs comparison

Easier Access to Published Applications

Domino allows users to create web applications for easy consumption of data science results by other users. We now directly link to published apps from the Project Overview page so it is easier to discover and consume models these information products.

Enhancements & Bug Fixes

  • Improved Markdown file rendering format
  • Warn project owners and collaborators when they attempt to stop a run they didn't start
  • Added a way for admins to create password reset links
  • Bug fixes for SSH for on-premise deployments
  • Fixed “Invite Pending” collaborator status to not be case-sensitive
  • Fixed an issue where HTML special characters would be rendered incorrectly in the file comparison view
  • Fixed an issue where the Domino CLI would fail to upload in projects over a certain number of files

1.43 (early October 2016)

Organization Ownership Transfer
Organization owners can now transfer ownership of the organization to another member.

Enhanced Kerberos Support
New user-specific keytab support, allowing users to authenticate as themselves when connecting to Kerberos-secured systems, like YARN.

Enhancements & Bug Fixes

  • SSH access tweaks: fixed an issue where users couldn’t connect in certain network configurations, and SSH sessions now automatically start in the Domino working directory and have basic environment variables set
  • Tweaked project entitlements for collaborators: non-owners can now no longer see certain aspects of the project configuration.
  • Users can no longer archive preparing runs (which could create a situation in which is was impossible to stop those runs)
  • Added a new Domino API endpoint to upload a file
  • Fixed an error when forking a project using a pre-existing name.
  • You can now access the user who started the run as an environment variable
  • Fixed an issue where runs started from the previous commits would use the wrong starting commit 
  • Modal windows can now be dismissed using the escape key
  • Fixed a bug where HTML files hosted in Domino could not source other files within the project
  • Added a new read-only "support staff" role who has all the read privileges of the original "support staff" role but cannot do any of the write actions
  • Fixed an issue where, in a large review request, the “too many files” warning message was misleading
  • Fixed a bug where the Domino CLI failed when encountering git submodules

 


 

1.42 (late September 2016)

LDAP Integration Improvements
With LDAP/Microsoft Active Directory support, Domino allows users to log into their private Domino deployments using their Windows credentials. This allows Domino administrators to centrally administer user accounts.

Domino has integrated LDAP / Active Directory for some time, but in this release, we have rewritten our implementation from scratch to make this connector more robust and to fix some specific issues, including:

  • Making the LDAP login failure message configurable
  • Ensuring quick-start projects are created for LDAP users by default

Kerberos Support for Hadoop / Spark Clusters
Domino now supports Kerberized authentication for YARN clusters, making Hadoop and Spark workloads function much better through Domino.

With this functionality, users can now attach a Kerberos keytab to Projects. Runs executed in this Project will automatically authenticate as that specified Kerberos principal.

Enhancements and Bug Fixes

  • Tweak project entitlements for result consumers: they can no longer see certain aspects of the project configuration
  • Expose the hardware tier that a run is executing on both as an environment variable and also via the API
  • Fix an issue where the CLI download link was misconfigured
  • Overhaul how public projects handle runs started by anonymous users
  • Small interface enhancements: fixed an issue when clicking organization name from the account settings page would scroll to the incorrect location and fixed the formatting of notebook control buttons when in embedded mode 
  • Fix incorrect documentation link on App Publishing screen 

 


1.41 (early September, 2016)

Enhancements

  • (AWS Mode) Added the ability to tag launched instances with a dynamically computed expiration date
  • Prevent user from creating an environment with an empty name
  • Clarify file size limits when uploading files through the browser
  • Make email configuration simpler by automatically cleaning up certain settings
  • Improve how cached files are handled on Executors, making out of space errors much less likely to affect user Runs

Bug fixes

  • Fix an issue that would cause project creation to fail nondeterministically

1.40 (late August 2016)

Enhancements

  • Prevent user from inputting invalid Git host in the Git authentication integration panel
  • Managing the space used by the working directory of all Executors is now centralized in the Admin section. (Previously it was necessary to log into each Executor individually.)

Bug fixes

  • Fixed an issue with noisy but harmless alerts when restarting the Dispatcher node
  • Fixed an issue in EC2 mode where Domino executors could not boot when no public hostname was set
  • Fixed an issue where Domino runs would fail if many runs completed at nearly all the same time (DOM-1659)

1.39 (mid-August 2016)

New features

  • Domino now supports connecting to Spark clusters using a YARN backend.
  • We've extended full Domino support for Spark through Zeppelin notebooks. Users can now interact with their Spark clusters using Jupyter, RStudio, and Zeppelin through Domino.  

Enhancements

  • Users can create folders in projects through the Files tab

Bug fixes

  • Under certain circumstances, anonymous users could not execute runs in projects that had "allow anonymous execution" enabled.  This is now fixed.
  • Loading the settings page of projects using a deleted custom environment now works
  • Fixed an issue where image overlay in run comparisons did not work in Firefox

1.38 (early August 2016)

New features

  • Domino now allows for direct SSH access. Users can SSH into the executor environment where their code is running. This feature provides users all the power of UNIX / command line tools to augment the work they were doing inside a batch script or an interactive session (for example, to act as a secondary interface to access data, code, and git repos). Note that this feature must be enabled by an admin, and is currently available for on-premises deployments only.

Enhancements

  • When Spark integration is enabled, a run will fail if spark-defaults.conf is not writable. In the past, runs would continue in these circumstances.
  • When a user requests to stop a run and discard the results, Domino requests confirmation from the user before continuing.
  • Users with read-only access to a project can now submit review requests. Previously, these users could fork projects but not merge them, and only users with contributor access to the original project could submit review requests.
  • Admins can now set the total number of files rendered during merges. Previously, this was capped at 250. Additionally, forking and merging is now faster, even with a large number of rendered files.

Bug fixes

  • Domino now properly ignores casing when ordering files on the File Browse page.
  • The Users page has been redesigned and optimized to improve performance.

 


1.37 (mid-July 2016)

New features

  • Domino users can now set their notebook working directory from the Domino web interface. This option can be configured in the Environment variables section of the Settings screen.
  • Users can also select environments from an automatically-generated drop-down list of all their organization’s available environments. This list includes all environments to which the user has access, regardless of which organization controls each environment, and can be found in the Compute environment section of the Settings screen.
  • Administrators can now migrate logs between Elasticsearch and Mongo. This can be accomplished within the Admin section by selecting Migrate Run Logs from the Advanced menu. From there, search for the appropriate logs using the run ID.

Enhancements

  • Domino can now execute a scheduled job before the previous scheduled job ends. This enables users to run scheduled jobs in quick succession, when necessary. Previously, Domino would wait to launch the second scheduled job until the first one was complete.
  • It is now possible to configure a range of ports for Domino to use with Spark. This addresses issues seen with certain types of firewalls inadvertently blocking access to Domino.

Bug fixes

  • It is now possible to download files with names that include commas or nonstandard characters directly from the Domino UI.

 


1.36 (early July 2016)

New features

  • Users can now easily access and use SparkR from within RStudio. For details on this new feature, visit the Domino blog: https://blog.dominodatalab.com/enhanced-apache-spark-integration/
  • Docker runs are now subject to resource constraints. This helps keep memory usage under control during Docker runs. These constraints can be configured by your organization’s Domino administrator. For more details on configuring Docker resource constraints, contact Domino’s Customer Success team.

Enhancements

  • Admins will receive a warning from the Dispatcher dashboard if an executor instance is unreachable. Previously, no notice was provided.
  • An explanation of Spark configuration options has been added to the Spark configuration page, along with a link to more detailed descriptions of those options.

 


1.35 (late June 2016)

Enhancements

  • The user interface of the File Detail page has been enhanced to improve clarity and provide for a more intuitive user experience with regard to notebooks. The buttons for launching notebooks and running files have been moved and redesigned to improve visibility.

 

Additionally, the inline file actions have also been reordered to help deliver a more consistent user experience across the browsing and detail views.

  • Users can now easily add multiple email addresses at once to an organization. Simply separate each address with a comma, space, or tab. Previously, users had to re-click on the Members tab after adding each new email.

  • Users can also add multiple collaborators to a project at once. Previously, after adding each collaborator, users were taken back to the Hardware and Environment tab, instead of remaining on the Access and Sharing tab.

Bug fixes

  • For RStudio interactive sessions, the Open Session button has been moved to the bottom of the console log to improve visibility. Previously, this button was at the top of the log, and it was often pushed out of view as lines were added to the log.

 


1.34 (early June, 2016)

New features

  • Users can now stop runs without saving the changes generated by those runs. This means Domino users can engage in more experimentation and engagement of “what if” scenarios without worrying about overwriting anything in the process.

Click the “Stop & Discard Changes” button to stop the run without saving anything.

  • Users now have the ability to archive comments. This will allow users to “clean up” comment threads as discussions move in different directions while still retaining a level of auditability, resulting in cleaner, less cluttered threads.
  • Domino now provides PySpark support to Spark standalone clusters. Users can now launch PySpark notebooks and run commands in these environments.
  • When comparing runs, users can now directly compare the Rmd code used for each run.

Enhancements

  • Domino’s “Add or edit a scheduled run” tab now includes a link to the accepted format for custom cron strings. This is important for users who are using custom expressions to schedule runs.



Bug fixes

  • The “File Details” page now includes a link that enables users to quickly view the run that generated a given output file. The link had not been available in the most recent previous release.

 


1.33 (late May, 2016)

New features

  • It is now easier for users to change project names after they’ve been created. This allows users to update a project’s name as it evolves, in order to better reflect its nature. This can be done by clicking on the Rename link in the Project Overview page.
  • Domino admins can now set a default environment for users based on membership in organizations. This approach will make it easier for those users to collaborate seamlessly.
  • Users can now preview comments in real-time, enabling them to take advantage of the more advanced features in Domino’s commenting system without having to worry about posting comments that include formatting mistakes.
  • When building a launcher, users can now create a checkbox toggle so that end-users can pass a TRUE/FALSE value to scripts. Checkboxes can be created via the “Type” drop-down list box in the “Parameter” section of the New Launcher screen.

Enhancements

  • Domino now provides users with enhanced abilities to layer custom docker files over base images.
  • The Domino CLI is now bundled with installs, in order to ensure compatibility. It can also be downloaded via the support site, if necessary.
  • When logging in via the CLI, cloud prod users will have to specify the host they are logging into. This is done by typing domino login $HOST instead of simply domino login.

 


1.32 (mid-May, 2016)

New features

  • A new user interface in Domino’s Admin section gives users the ability to configure SMTP for on-premises deployment themselves. Previously, only Domino’s engineers could do this.
  • Users can now create new environments by cloning existing ones. Previously, new Domino environments could only be built from scratch. Both admin- and non-admin-level users can clone their own environments; however, only admin-level users can clone global environments.

Enhancements

  • Domino now provides users with more detailed error messages, giving them an easy way to report the full stack trace of the exception back to Domino’s engineers.
  • Domino’s CLI version management has been improved and streamlined to ensure the correct version of the CLI is always downloaded.

Bug fixes

  • When a user makes a file content change through the UI, there is no longer a delay before those changes are visible. Once a change is committed, it appears instantly.
  • Domino’s scheduling policy has been adjusted to better balance loads across all eligible executors. Previously, it was possible in some situations for a single executor to become overloaded, while others sat idle.

1.31 (early May, 2016)

New features

  • Users can now publish apps from their projects, in the same way they can publish API endpoints. This enables them to use Domino as a hosting cloud for apps like Shiny or Flask. Users can take advantage of this feature to more easily share the results of an analysis with a non-technical audience. This feature is currently available as Alpha for our VPC and on-premises customers. Domino SaaS customers should contact us directly for access.
  • Users can now compare different versions of files across time. This feature is built into the Files tab and provides a convenient way for users to see which files have changed, on a file-by-file basis. Previously, file comparisons in Domino could only be done from the Runs tab.
  • Versioning in Domino is now easier. From inside the web UI, users can revert individual files from older runs to the current project without having to change each file manually. This can be done for previous runs as well.
  • Users will now see the projects they own at the top of the Project Overview tab, with the other projects they are associated with listed below. This provides users with a cleaner, less overwhelming display of their current projects.
  • Runs can now be scheduled for specific recurring days and times. For example, users can schedule runs to occur on the first Saturday of every month, at a time of day when database demand is low. This can be accomplished by including a custom trigger expression, using the syntax described here.
  • Pre- or post-setup scripts can now access environment variables. This gives the user the power to customize things as root that require those variables. For example, users could customize their environments to automatically login to github when they launch Rstudio sessions. If you’d like to learn more about how this feature will help you accomplish more with Domino, contact us for more information.

 


1.30 (late April, 2016)

New features

  • Admins can now easily track resource usage on a per-user basis by running a report. This tool is designed to help admins with capacity planning and chargeback. The Admin Usage Report is accessible in the Admin dashboard, and can be quickly configured to deliver specific information over specified time periods. Results are delivered in a .CSV file.
  • Users can now publish apps from within their projects, in the same way they can publish an API endpoint. This enables users to share analyses with broader groups of non-technical users.
  • Project owners will now receive email notifications whenever a collaborator is added to a project.

Enhancements

  • Users can now keep forked projects synchronized with the original project. This makes it safer to work on complex, time-consuming features and reduces the need to worry about drift.

Bug fixes

  • Stopped RStudio sessions are no longer reported as failed runs in the card view.
  • Under certain circumstances, requests to stop runs were either not processed by Domino, or were processed but ineffective. With this release, these situations were addressed, and the process of stopping runs is now more robust and stable.

1.29 (late April, 2016)

New features

  • Users can drag and drop files from their computers directly into a Domino project, by clicking "upload" on the Files page. The upload process no longer requires the use of the command line interface.
  • Users are now able to delete individual files from within the UI itself.
  • By clicking the “Download Project” button, users are now able to export a Domino project as a single .zip file:
Using ‘domino sync’ is no longer required for this.
  • Users can quickly and easily test out new ideas by creating new files directly within the UI.
  • Users can save and retain subsets of projects by selecting and downloading specific files directly from the file browser.
  • Domino now provides a much greater level of control for users who wish to start runs from within the browser:
    • Naming the run
    • Setting run parameters
    • Specifying a hardware tier
    • Scheduling the run to repeat
    • Re-publishing API endpoints with updated files from successful runs
Previously, these tasks could not be accomplished in one easy and centralized fashion.
  • Users now have the option to download and remove folders via an inline menu.

Enhancements

  • When a user follows a link that identifies a specific tab, that tab immediately becomes visible. The user is no longer required to manually select the appropriate tab to view the desired content.
This applies only to the sections of Domino where tabs are used, specifically:
    • Runs
    • Project settings
    • Organization settings
    • Review requests

Bug fixes

  • Search results now prevent the inclusion of archived projects, leading to a cleaner, more streamlined set of results.
  • Runs cannot start until working directories have been prepared for all dependency projects.

1.28 (early April, 2016)

Enhancements

  • Domino’s top navigation bar now scales and responds better on narrow screens.

Bug fixes

  • Users can now stop runs immediately after Scheduling but before Preparing without generating an error message.
  • Anonymous users may now visit public projects without generating an error message.

1.27 (early-March, 2016)

Features

  • Users who have an interactive session open for 12 hours will now receive an automated email, reminding them of the open session.
  • Users can now easily launch a Jupyter notebook when viewing it as a file.
  • Admins can use Domino’s new search indexing capabilities to deliver a better search experience to users, and to prevent downtime.

Enhancements

  • When users are added to a project, the notification email they receive now specifies who added them.
  • Domino now renders more gracefully on narrow screens.
  • Domino’s search tool now indexes both README files and the names of users involved with a project, so they can be included in search results.
  • There are several enhancements to Domino’s tag functionality for both on-premises and VPC deployments:
    • All tags are now searchable.
    • Approved tags now appear first in the auto-completion list whenever the user applies tags. This is intended to help prevent the generation and usage of duplicate or unnecessary tags.
    • Users can explicitly search for approved tags.
    • Users now enjoy more powerful options for searching for, discovering and sorting tags.
  • Search results now highlight searched text, enabling more efficient reviews.
  • The Project Overview page now includes a timestamp, for quick sorting of recent updates.
  • Users can now use tags to filter searches conducted within the comments search tab.

Bug Fixes

  • .Rdata files now render properly on the Comparison page.
  • Failed Jupyter notebook launches no longer appear as if they are still in progress.

1.26 (late-February, 2016)

Features, Enhancements and Bug Fixes

  • Significantly enhanced capabilities related to knowledge management and discoverability of projects, adding an array of improvements to:
  • Designed and deployed an improved projects overview page that highlights both “popular” and “suggested” projects --in addition to your own projects-- to help users discover work they might be interested in but didn’t already know about
  • Improved scalability for API Endpoints: there is now significantly less performance degradation in high concurrency situations
  • Fixed a bug where the runs dashboard wouldn’t properly refresh after starting the first run in a project 

1.24 & 1.25 (early-February, 2016)

Features

  • An array of significant new additions to Domino Search functionality:
    • Significantly improved search performance
    • Users can now search for filenames and project tags, in addition to project names, descriptions, and discussions

  • “File upload” parameters in Launchers can now accept multiple files at a time
  • Users can run .rmd and .rhtml files directly — Domino will automatically process them as knitr templates
  • Enhanced the “re-run” functionality to allow re-running against the exact past revision in addition to re-running the same command against the latest revision
  • (On-Premises and VPC) Added support for a new “Librarian” role 
    • Librarians are meant to be curators of project metadata (e.g., tags) on behalf of the organization, ensuring that projects are tagged and categorized consistently.
    • A Librarian has full ‘list equivalent’ access to all projects, even if a project is marked as private
  • Add new tag management dashboard for Administrators and Librarians
    • Administrators and Librarians can promote a standard tag to a new ‘approved’ tag type
  • Add the ability for user to initiate direct searches from tag counts

Enhancements

  • Significant enhancements throughout the Domino user interface
    • Improve the user interface for editing and managing compute environments
    • (On-Premises and VPC) Refine the user interface for managing central server configuration variables
    • Enhanced the layout for the search results page to improve usability
    • (On-Premises and VPC) Improved the general layout of the Dispatcher dashboard
  • Made a number of existing features more discoverable through the UI, including Run Diagnostics Statistics (dominostats.json), ‘dominoignore’ (to exclude files from sync’ing) and ‘dominoresults’ (to control what shows up on the results page)

  • Add efficiencies to page rendering for faster, more efficient interactions with the Domino Web application

Bug fixes

  • Corrected the display of the number of active runs when the executor is not available
  • Corrected the formatting of the endpoint code snippet
  • Corrected the File Browse view so it renders properly when a project name is very long
  • Fixed syntax highlighting for .m file rendering
  • Corrected issues with margins on the search and projects overview pages
  • Corrected issues preventing some runs of Jupyter notebooks
  • Corrected stuttering issues using Jupyter on Chrome & Safari

1.23 (mid-January, 2016)

Features

  • (On-Premises) Added basic support for project tagging as a way to create more organization around projects
  • (Beta) Added support for “Reviews”
  • (Beta) Added support for tagging Runs to better organize experiments on the Runs dashboard
  • (Beta) R/Python package exports: a project can export its contents are a Python or R package that multiple other projects can import.

Enhancements

  • Some moderate latency improvements to API Endpoints (more coming soon)
  • Discussion page now shows full thread associated with each comment
  • UI improvements to commenting (posting a comment no longer reloads the page)
  • Rda files now render with their contents in the files/results view
  • CSV files now render as interactive tables in the file and results view
  • Improved UX on the project settings page
  • Run diffing now provides an option to view image diffs in an "overlay" view for easier visual comparison

Bug fixes

  • File upload window in web site didn’t work in Firefox
  • Project overview page would report API Endpoints as “running” runs for a project
  • RStudio sessions showed a scary but innocuous warning upon startup
  • "?" help command didn't work in RStudio sessions

1.22 (mid-November, 2015)

Features

  • "Data Imports" let you create shared, reusable data sets that other projects can import. More info.
  • Allowing users to permanently install dependencies in Docker containers, to avoid long install times when running their scripts. More info.
  • Improved organization support and management
    • UI for creating and managing organizations directly
    • Organization project page shows all of organization's projects
  • Users can now transfer project ownership through the UI. More info.
  • (On-Premises) we are now able to upgrade executors without stopping users' jobs, in most cases

Enhancements

  • Improved UI for interactive notebook sessions. "Slide-out" pane lets you control syncing without taking up screen space
  • Users now stay logged in for longer, as long as they are regularly using the site
  • Jupyter notebook sessions now run on standard ports, making them more easily accessible behind corporate firewalls
  • (Cloud) Allowing runs to save files that are more than 5GB
  • Added support for @mentions in comments
  • (On-Premises) Automatically checking and creating database indexes to improve performance
  • (On-Premises) Support for using Amazon SES to send emails when deployed in a VPC environment
  • (On-Premises) Support for adding custom tags to EC2 instances that Domino creates
  • (On-Premises) Blobs are now compressed when they transfer, improving performance of file sync operations

Bug fixes

  • Organization members get notifications as though they were project members
  • Fixed a bug that would make it look like a notebook session never started, when it had simply taken a long time to start

1.21 (end-August, 2015)

New Features

  • Project search
  • Added support for spinning up Rodeo sessions
  • Added support for Beaker notebook sessions
  • ipynb file rendering in the browser (1.20 added this for on-premises deployments. 1.21 adds this for the cloud as well)
  • User-specific environment variables
  • Added a project-level view that shows all discussion on a project
  • Added ability to export environment variables from one project and import them into another

Enhancements

  • (On-Premises) Improved users dashboard provides a better sense of which users are active
  • (On-Premises) centralized logs (executors, front-end, etc all feed to one place)
  • Notification emails will include attachments even when using email.html template
  • On the overview page, projects sorts by last updated
  • UI improvements to make the "notebook" experience better
  • Now able to run/execute Jupyter notebooks directly from the "view file" page
  • Able to create a new documentation (README) file directly from the Documentation section of your project
  • Rendering results files asynchronously for faster response
  • Changed "re-run" behavior to use the latest version of the project (as opposed to the version at the time of the original run)
  • Scheduled runs can now be edited only by the user who created them (or the project owner)
  • Added a "report view" for the results dashboard that strips away all Domino UI and just shows the results themselves
  • Improvements to the commenting UI
  • Added ability to purge logs for past API endpoint releases

Bug fixes

  • Now able to launch scripts from the files view for files in subdirectories
  • Fixed several bugs related to runs search
  • "Expanded" file view won't try to render binary files anymore
  • Fixed a bug that would cause excessive memory use when viewing a file with lots of revisions
  • Fixed bugs related to sanitizing inputs on several text input boxes

 


 

1.20 (mid-July, 2015)

New Features

  • IPython Notebook improvements.
    • Upgraded to Jupyter, including Python, R and Julia kernels, as well as Terminal access
    • (On-Premises) Rendering notebooks directly in the file view
    • Better support for batch running of notebooks
    • Added a “download only” sync option for a running notebook (to pull down new changes into your notebook session without uploading your local changes)
    • Better UI for working with Notebooks 

  • Added support for GPU instances
  • Added Spark in our base container, so you can more easily run local clusters or connect to your own cluster

Enhancements

  • (On-Premises) RStudio starts up with correct working directory
  • (On-Premises) Docker image updates now automatically deploy across your cluster
  • (On-Premises Admins can set the target executor UI
  • (On-Premises) Dispatcher dashboard now updates without a page refresh
  • Support for mathjax formatting in comments

  • The Run Number can now be accessed via the DOMINO_RUN_NUMBER environment variable from your code.
  • Allow personal plan users to have collaborators
  • Comparison view makes better use of horizontal space
  • Markdown files render in the file view
  • YAML files now render properly in the comparison view
  • Added a way to specify a title for a scheduled run
  • Added a “failure only” notification level
  • Now recording timestamps in the CLI log file
  • Added diagnostic stats to the Run objects returned by runs_list() in the REST API
  • Increased the max notification attachment size
  • File view now shows commit list

Bug fixes

  • Documentation page for public projects can now be accessed by anonymous users
  • Last several lines in the file editor would be truncated
  • dominostats.json won’t be attached to results notification emails anymore
  • Now able to comment on a running run
  • Comparing a run that wasn’t finished would cause an error — not anymore
  • Will now show stdout for Notebook sessions if any setup (e.g., installing requirements) fails
  • Run button wasn’t showing up for files with spaces in their names
  • Race condition with run numbering


1.19 (early-June, 2015)

New Features

  • Commenting and discussion! You can now comment on runs, results and comparison.
  • You can now sync file changes from a running Notebook session, so you can get new files into your running notebook environment
  • (On-Premises) blobs are now compressed in the blob store, reducing storage requirements

Enhancements

  • Added API Endpoint examples in other languages
  • You can now create a scheduled run from an existing Run (on the Run's "details" tab)
  • Project owners can change collaborators' notification settings
  • Added an ability to run more file types from the file browse view
  • Improvements to comparison views, better handling of wider content types
  • Comparison view highlights differences between the runs' commands
  • Commands in the R package no longer display a warning when first run

Bug fixes

  • Windows installer now works even if Java isn't already installed
  • Will now prevent you from publishing an API Endpoint if the function name is blank
  • Handle extra whitespace at the end of API Endpoint file/function names
  • Fixed a broken contact link
  • Setting MTU to 1500 on all EC2 instances types, addressing a problem that happens on c3 instance types where the MTU is too high and causes some database operations to hang

  


1.17.1 (mid-May, 2015)

  • Added a new "launcher only" collaborator access level, letting you enable users to access launchers without viewing any underlying files in a project.
  • Improved performance in the web UI when browsing projects with many files

 

1.17 (mid-May, 2015)

New Features

We added two powerful new features to make it easier to develop your models. Check out our guide on using these together for Model Tuning & Experimentation.

  • "Run Comparison" shows you side-by-side differences between inputs and results of two runs
  • "Diagnostic Statistics" let you track structured data about your runs, e.g., AUC

Enhancements

  • Project collaborators can manage environment variables
  • Project owners can change project environment
  • (On-Premises) Admin pages have a more consistent look and feel
  • "Archive project" description is more accurate
  • Moved "Manage Organization" link to Account page
  • Separate tabs for run output, details, and performance graphs to make the UI more humane
  • Able to "pause" scheduled runs
  • (On-Premises) date format in log files is more humane
  • The next occurrence of a scheduled run won't happen until the previous one has succeeded

Bug fixes

  • Now able to use a URL to access a run even if it isn't on the first page of the Runs Dashboard
  • Notebook sessions will now be named sensibly even if the starting user doesn't have a "name" on their account

1.16.1 (late-April, 2015)

Enhancements

  • Able to view directory sizes through the Files view on the web interface
  • Able to delete files through the files view on the web interface
  • (On-Premises) Admin UI now allows viewing and clearing out cached run directories
  • (On-Premises) improved log-rolling behavior to reduce log file sizes
  • Small enhancements to the Launchers UI

Bug fixes

  • Python API Endpoint scripts running in a subdirectory can now load load modules normally
  • Couldn't start a batch run of an ipynb file that was in a subdirectory

1.16 (mid-April, 2015)

Enhancements

  • Added a new Documentation section for each project, that will render any README (or README.md) file in the project
  • Able to create a Run record from work done locally, using domino run --local "<command>"
  • Automatically title Notebook sessions with the user who started them

Added domino.sync to our R package. Get the updated package by running

  • install.packages("domino", repos='http://cran.r-project.org', type="source")
  • Collaborators can't stop other collaborators' runs.
  • (On-Premises) Added a better UI to allow admins to create and manage Organizations
  • (On-Premises) Dispatcher dashboard shows CPU/Memory usage for in-flight runs
  • (On-Premises) Can stop a run directly from the Dispatcher dashboard
  • (On-Premises) Added functionality for Domino Admins to execute queries directly against the meta-data database

Bug Fixes

  • Fixed an issue where reported CPU usage would sometimes be negative
  • Can now run files with parentheses in the file name
  • CPU/Memory profiling now works properly
  • On certain browsers, the file revision list wasn't visible on the "view file" page

1.15 (early April, 2015)

Features

  • The Runs Dashboard now shows CPU and memory usage over time for each run (rather than just an instantaneous snapshot)
  • (On-Premises) Admins can now define and manage their cluster through the UI

Enhancements

  • Fine-grained control over which results to display on the web. You can now control which files will be displayed as “results” by explicitly listing them in a “.dominoresults” file in your project. Learn more.
  • Run numbering: Runs now have an auto-incrementing human-readable number, so you can more easily refer to them.

Bug Fixes

  • Domino client would create a lot of files in the system’s temp directory

Version 1.14 (mid Feb, 2015)

  • Fixed issue where files that weren't actually changed were appearing on the Results Dashboard
  • In on-premise mode, allow alert recipients to be configured
  • For public projects, anonymous users can now use run launchers
  • R Notebooks now support plotly interactive charts
  • Added (very) alpha support for Scala Notebooks
  • Improved performance of rendering results dashboards
  • Assorted API endpoints improvements:
    • Keep active release serving if a newer but broken one is published
    • Mitigated issue with R endpoints where concurrent requests would occasionally fail
    • Add ability to unpublish an endpoint
    • Dramatically improved our ability to upgrade endpoint functionality, so upgrades should be significatly easier, faster, and safer

Version 1.13 (mid Jan, 2015)

  • Domino-specific environment setup logging is now visually separated on the runs dashboard.
  • Fixed issue where downloading files with certain extensions via the browser would not get unzipped properly
  • Make it easier to republish endpoint servers
  • Fixed issue in R notebooks where a single cell failure would bring down the entire notebook
  • Separate setup-specific logs from logging from main script
  • Various improvements to our internal infrastructure

Version 1.12 (late Dec, 2014)

  • Added new project mode where results of runs are not merged back to the main line. This should improve an experimentation workflow in the following ways:

    • Clarifies which run generated which result files
    • Intermediate files won't pollute the project
    • Result files don't need to be downloaded to the client as often which saves on download time and also makes resetting much less likely

    This setting is available on the project's setting page.

  • For users on our Business plans, we've added the ability for you to modify the environment your runs are executed in. For instance, you can install packages and start services your code requires. Please contact us if you'd like more details.
  • Existing scheduled runs can be edited
  • From the "Edit file" screen, you can now directly save and run the file with one click
  • Values of project environment variables can now be shown
  • Added API access (and Python bindings) to start batch runs
  • Fixed issue where installing packages in an API Endpoint server file would fail if a CRAN repo wasn't specified
  • (Finally) fixed extremely annoying and confusing bug where clicking in the title of the run didn't also select the run

Version 1.11 (early Dec, 2014)

  • Scheduled runs can override the project hardware tier
  • Stability and performance improvements
  • Improvements to API Endpoints functionality:
    • User-visible log of endpoint startup, to make debugging easier
    • Improved JSON to R/Python object conversions
    • Return elapsed time and commit ID in response automatically

Version 1.7 - 1.10 (Aug - Nov, 2014)

We've been so busy adding new features that we forgot to update the release notes. Highlights from the last couple months include:

  • "Secure environment variables" let users securely store sensitive passwords and keys and inject them safely into their code at runtime
  • Significantly improved performance of the CLI when working with projects that contain many files
  • Fixes for two major system-level security issues: Shellshock and POODLE
  • Improved CLI help functionality
  • Made "search" on the runs dashboard work better
  • Added support to run IPython Notebooks in “batch mode” — so they can be scheduled, deliver alerts, etc
  • Improved email notification formatting
  • Launchers remember most recently used values
  • Launchers can be defined to use a specific hardware tier
  • Layout improvements to the runs dashboard
  • Added a weekday-only option for scheduled runs
  • Improvements to IPython Notebook support to make it more usable
  • Added beta support for interactive "R Notebooks"
  • Improved auto-scaling of our cluster so runs should be "queued" less often
  • Lots of minor bug fixes

Version 1.6 (Late June, 2014)

New features and enhancements

  • Add human-readable descriptions to parameter values in dropdown in Run Launchers, so launchers are better documented for end-users
  • Console error output is now correctly interleaved, so users won’t be confused as to when an error occurred
  • All uploads and downloads from the client happen multi-threaded, so transfers should happen much faster
  • Rearchitected the run scheduler so now runs should spend much less time “queued”
  • You can now edit (small) files directly in the browser
  • Added new sync CLI command that is much less annoying than download/upload
  • Reorganized notification settings to make more sense
  • Made writing the code-part of the Run Launchers easier by more automatically handling parameter values for the programmer
  • Scheduled runs have a notification list
  • Download an entire folder as a zip from the web site (Enterprise deployments only)
Was this article helpful?
0 out of 0 found this helpful

Comments