Frequently Asked Questions
How to request for access to a Compute Cluster?
In order to use resources on the HPC infrastructure it is required to associate your Linux user account with corresponding department or working group Slurm account in the workload management system (Slurm):
- Contact an account coordinator for your department or working group and fill out a user access request. Please specify the Slurm account name provided by your account coordinator.
- If no account coordinator exists for your department or working group then request access by filling a user access request specifying
default
as your requested account. (Note that users can be associated with multiple Slurm accounts. Follow instruction under default-account to configure your account settings.
Instructions on requesting a Linux user account are described in How to get an account? on the public GSI web-site. The processing of requests for new Linux accounts may take multiple days. Therefore make sure to prepare in advance if an account is required at a specific date for new co-workers or a workshop.
How to subscribe to the HPC-INFO mailing list?
Operational issues of the HPC infrastructure are communicated using the HPC-INFO mailing list. This list announces schedule maintenance as well as operation issues. Use following links to subscribe or unsubscribe from the HPC-INFO mailing list:
An archive of all mails distributed over the HPC-INFO mailing list is available on the Listserv web-interface.
How to Request the Installation of Software?
Request the installation of additional software with a Software Installation Request (SIR) in a GitLab issue 1. Additional software is not installed on the cluster nodes, but will be included into a virtual application-environment (VAE).
Software with dependencies on X11 libraries like web-browsers, graphical text editors, integrated development environments (IDEs) or image viewers will not be install in VAEs. However basic support for X11 is available on the cluster. Consider to build a custom container image for your X11 based application.
Where to Find Information about Cluster Utilization?
Information about the utilization of the Virgo cluster and the associated storage systems are made available to users with multiple monitoring dashboards 2. The monitoring interface is build with Grafana 3 and is reachable from outside the GSI networks. Visit the Virgo User Monitoring at: https://cluster.hpc.gsi.de
The Virgo User’s Dashboard 4 includes information about the pending and running jobs for all Slurm accounts, as well as the number of utilized CPUs. The Lustre User’s Dashboard 5 displays write and read IO throughput to the shared storage in /lustre
accumulated for each account.
How to request a new Slurm Account?
Collaborations and experiments associated to GSI/FAIR can apply for a new Slurm account using the account request form. During creation of a new account a responsible account coordinator will be established. Account coordinators are more privileged account users with permission to:
- Add users to an account.
- Create sub-accounts.
- Modify fair-share and limits for account users.
- Control all jobs associated to an account.
- Grant coordinator privileges to other users.
How to Request a Resource Reservation?
Account coordinators can request a resource reservation for nodes, CPU cores and GPUs using the resource reservation request in the GSI issue tracking on GitLab. Users need to contact their account coordinator in order to reserve resources on their behalf.
Note that resources need to be freed before a reservation. Depending on the resource this may take more then a weeks time. Therefore plan accordingly and request a reservation ahead of time.
Read the cluster reservation section in the user guide to find more details bout using a reservation during job submission. A list of all reservations in the systems is printed with the command: sinfo -T
.
How to Request a CVMFS Repository?
If you are a software coordinator for an experiment or working group affiliated to GSI or FAIR, you can request a dedicated CernVM-FS repositories for software distribution to the compute clusters. Details are available in the “CernFM-FS Server documentation” 6. Please contact cvmfs-service@gsi.de for further instructions and planning of the corresponding resources.
How to Reference Virgo in Publications?
Researchers using the cluster resources are kindly asked for acknowledgment.
Consider to a reference to your publications by adding a line like:
“This research was supported in part by the cluster computing resource provided by the IT Department at the GSI Helmholtzzentrum für Schwerionenforschung, Darmstadt, Germany.”
How to change my Linux user account configuration?
Verify your Linux user configuration on the login nodes:
# ...print your default shell
echo $SHELL
# ...list the Linux groups for your user
groups
Request a change of your Linux user account buy sending a mail to it-service@gsi.de. This included changes to the default shell or requests to join additional Linux groups to access data on shared storage. Note the newgrp
7 command is used to change the default group ID for a shell session. This is useful in order to ensure that output data is written with the correct group onto shared storage.
Footnotes
GitLab Issues, GitLab Project
https://docs.gitlab.com/ee/user/project/issues/↩︎Grafana Dashboards for Users
https://cluster.hpc.gsi.de/grafana/dashboards↩︎Open Source Grafana
https://grafana.com/oss/grafana/
https://grafana.com/docs/↩︎Virgo User’s Dashboard
https://cluster.hpc.gsi.de/grafana/d/Rh4dw-cMz/virgo-users-dashboard↩︎Lustre User’s Dashboard
https://cluster.hpc.gsi.de/grafana/d/A51zalpGk/lustre-users-dashboard↩︎CernVM-FS Server Documentation, GSI
https://git.gsi.de/dc/cvmfs-server↩︎newgrp
Manual Page
https://manpages.org/newgrp↩︎