RELEASE NOTES
2020.10
-
DataTask :
-
DataTask becomes multi-tenant (multi-organization): a single DataTask instance can be shared between several organizations while ensuring access to resources.
-
Datatask’s embedded git is also aware of these multiple organizations.
-
-
Admin UI :
-
Namespace management : namespace creation is available in the admin interface. Used to access resources, a group can be scoped to one or several namespaces for which they can deploy, delete, list and view resources.
-
Organization management : a user can be assigned to an organization from the administration interface.
-
-
Deploy :
-
An application can be hidden on the portal by setting the value
false
to the propertypublished.visible
. -
Services and deployments can be updated by clicking the update button from the task list. No longer need to delete and re-deploy them, the update is done without downtime.
-
Once deploying a new service, your accesses are updated automatically. No need for re-authentication.
-
The value "All namespaces" has been added to the task and pipeline lists. It allows to display the task or pipeline list of all namespaces you have access to.
-
2020.03
New features
-
Portal :
-
New UI design.
-
-
Deploy & Pipeline :
-
New UI design.
-
The deployment of a task is now done in three steps :
-
Loading the manifest.
-
Editing the manifest.
-
Deployment of the task.
-
-
The launch of a pipeline is now done in 3 steps. :
-
Loading the pipeline.
-
Editing the pipeline.
-
Launch.
-
-
You can now deploy tasks and pipelines from templates.
-
You can add custom icons or images to your services. To do this you must edit the field
published.icon
.
-
2020.01
New features
-
Deploy & Pipeline :
-
API :
-
Added
tolerations
andnode-selector
options in themanifest.json
configuration. You can now manage node affinity and assign tasks to specific nodes. -
The
published
field of themanifest.json
for a service becomes optional. You can now deploy a service without exposing it externally by leaving thepublished
field blank or omitting it.
-
-
2019.09
New features
-
Admin UI :
-
Users :
-
Added a button to duplicate a user.
-
Added a button to delete a user directly from the list.
-
-
Groups :
-
Added a button to duplicate a group.
-
Added a button to delete a group directly from the list.
-
-
-
Deploy & Pipeline :
-
UI :
-
Pipelines are now sorted by date from the most recent to the oldest.
-
The pipeline tracking interface now includes the current step of each task.
-
The pipeline tracking interface now gives you the ability to view logs for a specific task.
-
The interface that displays tasks is now limited to tasks of one week by default. With the possibility to display the oldest ones.
-
You can find a task that has just been started in the list of all tasks using a
find in list
button. -
Added different links to Kubernetes resources on the drawers of each task.
-
-
API :
-
Added
rewrite
andtimeout_ms
options in themanifest.json
configuration for service deployment : You can now deploy more complex services such as a Jupyter Lab. -
Added
sticky_sessions
option in themanifest.json
configuration for service deployment : If your service has several replicas, users will always fall back on the same pods. This feature is useful for horizontally scaling of statefull services. -
Added
bypass_auth
option in themanifest.json
configuration for service deployment : If you set this option to true, your service will be deployed with public access. -
Injection of an environment variable
TASK_LOGGING_ID
to each task to allow the user to send the application logs of a task to the same filter as the deployment logs.
-
-
2019.06
New features
-
New administration interface (Admin) to manage your users and groups :
-
Users :
-
Show all users in the database.
-
Create / Update / Delete a user.
-
Add or remove a user’s shortcuts.
-
-
Groups :
-
Show all groups in the database
-
Create / Update / Delete a group.
-
Add or remove a secret of the services accounts of a group.
-
Add or remove a user from a group.
-
View the list of applications to which the group gives access.
-
-
-
Deploy & Pipeline :
-
Added the
volumes
option in themanifest.json
configuration : You can now deploy a task with one or more mounted volumes. -
When using the API Deploy from within the cluster, added the possibility to provide the user’s identity by an "Identity" header (without the "authtoken" cookie).
-
Added
deployment
type in themanifest.json
: You can now deploy adeployment
type task.
-
-
Addons :
-
Added a template for automatic job generation through a python notebook.
-
Added a template for launching a job or cronJob on a compute engine instance with GPU, CPU, RAM, … configuration option.
-
2019.04
New features
-
Deploy & Pipeline :
-
Added the
image-pull-policy
option in themanifest.json
configuration. -
Added the
run-if-error
option in thepipeline.json
configuration : You can specify whether a task should be started even if its dependency fails. -
Added an option to create a cronJob on the interface directly (without using the
manifest.json
configuration). -
Added the
svc-account-secret
option in themanifest.json
configuration : You can specify a secret containing a service account to which you have the rights, your task will then be launched with an environment variable "GOOGLE_APPLICATION_CREDENTIALS" pointing to the service account.
-
Improvements & Bugs fixes
-
Adding a "Home URL" to users: The user will be redirected to his "Home URL" after connecting to the platform.
-
Deploy & Pipeline :
-
Optimizations.
-
Improved visualization of pipeline status.
-
Added the possibility to connect an external git and other than gitbox to Deploy.
-
Added the output of the step of building the docker image in the logs.
-
Added the identity of the user launching a task or pipeline to deployment logs.
-
2019.02
New features
-
New reverse proxy architecture (no downtime when adding a new service) : Powered by the Ambassador.
-
Management and configuration of new users and groups through the Group-Manager Script.
-
Deploy & Pipeline, add management of :
-
Repositories : You can specify the repository of your code.
-
Branches : You can specify the branch that should be used.
-
Namespaces : You can specify in which namespace the task should be deployed.
-