This page details all aspects of accessing and using the ML Cloud. If you are newcomer to the AI Center and the ML Cloud, we recommend reading over this entire document to familiarize yourself with the features and capabilities available within the portal. If you are looking for information about a specific topic, you can use the navigation links on the left to quickly jump to that section of the page. If you are unable to find the information you need in the Help Pages available, please submit a ticket with your query.
There are several steps that need to be completed in order to fully utilize the ML Cloud Service and Portal.
To obtain an account with the ML Cloud, ask your PI or Group Manager to request a new user account through our User Management System where they will see the button to request an account. From there they need to fill out information such as Name, email, start and end date for the requested account:
Then they should fill out the form:
Once the request has been submitted, we will proceed with granting you access to the ML Cloud Resources.
Read the Good Conduct section: The Ml Cloud is a shared resource and your actions can impact other users (Last Update: 17/02/2023).
All ML-Cloud services share the same identity backend. The account/credentials can therefore be used to log into all all of them. These accounts are exclusive to the ML-Cloud.
To obtain access to the ML Cloud a username will be provided to you. You will notice it will contain few letters followed by several numbers, such as "pbb111". The reason behind this naming convention is based in GDPR so we do not process personal identifiable information (such as First and Last Name) once the user is offboarded.
Once a user is logged in, a drop-down menu topped with the user's name will be displayed on the right side of the navigation menu. When toggled, this menu exposes links to the user's dashboard, to manage the user's account, to submit a ticket for support, for portal administrators to manage user onboarding status and to log the user out.
Under this section of the portal users can manage their profile, update their personal information, change their password, request activation of licensed applications, manage tickets and integrate third-party applications.
To obtain Research Group Access to the ML Cloud resources please request access through the ticketing system: support@mlcloud.uni-tuebingen.de stating from which organization you are requesting access to (Tübingen AI Center, Machine Learning Cluster of Excellence). We will contact the representatives to confirm your group is eligible for access and then proceed with creating it.
There are few steps that are initiated when a user leaves the AI Center and which will be triggered once the user (or a group representative) indicates leaving period. ML Cloud User offboarding should be taken seriously to ensure that the organization's data and resources are protected.
The process of offboarding is triggered by the Group PI or Group Admin by specifying a date of user leaving or contacting the ML Cloud Team through the ticketing system before the user leaves. The ML Cloud team triggers the offboarding procedure by sending a message with offboarding explanation to the user with the following key points:
1. Account Access: effective the offboarding date the user's ML Cloud Account and associated credentials are deactivated. Type of resources the user will have no access to:
1.1 Slurm - the user needs to make sure that no jobs are running past the effective date of offboarding.
1.2 Openstack
1.3 Gitlab - If the user has any projects they would like to retain from the ML Cloud gitlab, please fork them from the ML Cloud gitlab to a preferred repository outside of the ML Cloud before the date of departure.
1.4 ML Cloud Mailing List - the user's email to the ML Cloud mailing list will be removed.
2. Data and Files: Prior to offboarding, we kindly request that the user reviews their data and files stored in:
It is important to securely store (outside the ML Cloud) any necessary documents, data, or files one may require for future reference or handover purposes.
What happens to a user's data (in each storage point) after the offboarding date?
As a general policy, the user's access to data stored in these points will be removed. Twelve (12) months after the effective date of departure (or by a direct instruction at any point from the PI), the data will be permanently deleted from each point and all data lost. Here are some exceptions:
Pointers for the remaining data storage points:
