Getting Started

Welcome! This guide walks you through the first actions you need to start using Sciserver.

Invite recieved

Sciserver is a private, invitation‑only platform. You should have received an email that contains a link to the Sciserver portal and the address of the administrator you can contact for help (see Support & Contact).

Before you reach out to the admin, please complete the following onboarding actions. They are required for the system to recognise your account and to enable the rest of the workflow.

#

Action

Details

1

Create your Sciserver account

Open https://sciserver.mpe.mpg.de and click “Create a new account.” Use exactly the same email address that the invitation was sent to.

2

Log in twice

Return to the portal, sign in again with the credentials you just created, and stay logged in for a few seconds. This login registers your account with the internal services and is required for any subsequent actions.

3

Subscribe to the mailing list

When updates occur that may affect your workflow , or major issue happen, users are informed via the mailing list. Register with the same email on the official mailing list.

Important: Skipping step 2 will prevent the platform from completing the provisioning process, and you will encounter errors when trying to use Sciserver later.

Once these three steps are finished, you are ready to explore the platform. The rest of this page will walk you through the main UI elements, how to access your files, and how to launch a Compute container.

If you encounter any issues during the steps above, please double‑check that you used the correct email address and that the confirmation link was followed. Should problems persist, contact the administrator via the Support & Contact page.

Welcome to Sciserver – we’re excited to have you on board!

Accessing the portal

  1. Open your web browser and go to the Sciserver URL: https://sciserver.mpe.mpg.de

  2. Log in with your credentials (pre-approved by relevent project lead). (If you do not have an account, get in touch with your supervisor or project-related lead.)

Tip – You must login at least once for your user to be enabled.

Uploading a file

This step is optional but useful to verify that the Files service works.

  1. Click Files in the apps navigation.

  2. In the User Volumes section click on persistent, and then Upload button, and select a small text.

  3. Verify that the file appears in the list and can be downloaded again.

Tip – All files uploaded this way will be owned by root user inside your containers.

Launching an interactive Compute session

  1. From the main dashboard click Compute → Create Container.

  2. Set a name you can identify later.

  3. Select a compute domain.

  4. Choose an image – for a first‑time run select Sciserver Base.

  5. Select user volumes (optional). By default, scratch and persistent are preselected. Other shared folders should appear here as well.

  6. Select a data volume (optional). If you are part of a special project, you should see the different available data volumes for you here. If you feel one is missing, reach out to your contact person to request or clarify data access.

  7. Click Create.

Sciserver will create and list the new container on the same page. You can click on the name to open that container. And manage it (stop/start) via the UI.

Tip – If you ever need to contact IT support, make sure to include the URL of the container in your request for help, as it carries information useful for IT team to identify it.

Submitting your first Compute Job

Use a job when the workload is long‑running or needs exclusive hardware.

  1. Return to the Sciserver portal and click Compute Jobs → Run Command.

  2. Job Alias (Name): demo‑job

  3. Compute Domain: Hardware specs. Usually only one choice.

  4. Compute Image: Usually the same as your regular compute.

  5. Data Volumes (optional): Any Data Volumes required to get your work done.

  6. User Volumes: Wherever target you need to place output files on.

  7. Command (shell): Here you can instruct the job to run a script you may have created or run directly your shell command. Consider that Jobs needs at least one location to place the output of the task. Either allowing it to create a new one, or choosing an existing User Volume.

    python -c "import time; print('Job start'); time.sleep(60); print('Job end')"

  8. Click Create Job.

  9. Monitor the job from the Jobs list; when the status changes to COMPLETED you can download the stdout/stderr logs.

Next steps

Congratulations! You have logged in, uploaded a file, launched a container, and submitted a compute job. From here you can explore the rest of the platform or dive deeper into the sections listed above.

Topic

Where to go

Managing Python environments with mamba

Compute Container

Using pre‑populated data volumes

Files And Volumes

Advanced job submission (GPU, multi‑node)

Compute Jobs

More help

Support and Contact

For complete, official instructions and advanced topics, please refer to the upstream Sciserver documentation at https://www.sciserver.org/support/.

If anything went wrong, open a ticket in the internal help‑desk or consult the “Troubleshooting” page.