Skip to content

Frequently asked questions

Login/2FA

The connection to HAICORE@KIT cannot be established.

Please note that you have to be connected to one of the trusted networks to be able to access HAICORE@KIT. If you are not on campus, you have to establish a VPN connection to your institution first.

The "Your OTP:" prompt never appears and the connection hangs/times out instead.

Please note that you have to be connected to one of the trusted networks to be able to access HAICORE@KIT. If you are not on campus, you have to establish a VPN connection to your institution first.

The One-Time Password is not being accepted.

There are multiple possible reasons for this:

  • The Token was not fully initialized. Please note that that after the QR code has been scanned, an OTP has to be generated and entered back into the web interface to confirm that the token works correctly.

  • The One-Time Passwords are time-dependent. Please make sure that the local clock on your device is set correctly.

  • Make sure you are using the correct Token to generate the OTP.

The service password is not being accepted.

Please keep in mind that the dedicated service password is different from the account password you are using to log into FeLS.

I have lost my token. Can I register a new one?

Yes, of course. Please open a support ticket using the account you use to log into HAICORE@KIT, use the subject "Reset token" and add a short description of what has happened.

There is an error message by the pam_ses_open.sh script.

Your account is most likely in the "LOST_ACCESS" state because the entitlement is no longer valid or there was a problem during the communication between your home institution and the central bwIDM system. Please try the following steps:

Log into FeLS, look for the FH2 entry and click on Registry info. Your "Status:" should be "ACTIVE". If it is not, please wait for ten minutes since logging into the bwIDM causes a refresh and the problem might fix itself. If the status does not change to ACTIVE after a longer amount of time, please contact the support channels.

"Received disconnect: Too many authentication failures" during login.

The SSH client is trying too many SSH keys, which causes the server to close the connection after a limited amount of tries. To force a password login use SSH with the following options:

ssh -o PreferredAuthentications=keyboard-interactive -o PubkeyAuthentication=no
HAICORE@KIT is not shown as a Service in FeLS

If you are not a KIT user make sure you selected Helmholtz AAI as your Home Organization. Sometimes it can help to login to FeLS again in a new private Tab.

Software

Why can't I install software using apt, apt-get, yum etc.?

HAICORE@KIT is a multi-user system that you are sharing with hundreds of other users. To keep the system safe, reliable and secure for everybody we cannot allow users to install software on a system-wide level using administrator tools like apt, apt-get, yum, zypper etc.

You can install necessary software components in your HOME directory.

Batch system

My jobs are not starting and the "Reason" field is set to "(AssocGrpCPUMinutesLimit)"

This means that at least one of your computing time quotas, either the one for CPU hours or the one for GPU hours, has been used up. Please use the commands listed in the Project accounting section to check your quota and resource consumption.

Full projects can be extended if the granted resources have been consumed before the effective end of a project. Please refer to the project proposals section for more information.

Message Passing (MPI)

My MPI jobs are failing with the error message "ORTE has lost communication with a remote daemon."

There are multiple possible explanations you should check:

  • Your job might be running into a memory limitation. Please not that if you do not specify a memory requirement for your job, then the default of --mem-per-cpu=1600mb or --mem-per-cpu=3300mb (depending on the node type, see here) is used. This might not be what you want. It is always better to specify your memory requirements. If in doubt, use the maximum limit per node, e.g. --mem=243200mb or --mem=501600mb.

Last update: September 27, 2023