Access to the HPC system HYDRA

Remote access

For security reasons, direct login to the IBM iDataPlex HPC cluster HYDRA is allowed only from within the MPG networks. Users from other locations have to login to one of our gateway systems first. Use ssh to connect to HYDRA:

ssh hydra.rzg.mpg.de

You always have to provide your (Kerberos) password on the HYDRA login nodes, SSH keys are not allowed.

Secure copy (scp) can be used to transfer data to or from hydra.rzg.mpg.de.

The SSH key fingerprints for hydra.rzg.mpg.de are:

1024 8d:3c:4b:a2:9e:15:fa:15:3e:ae:b2:9b:dc:99:f7:91 (RSA)
1024 6d:c4:11:9d:b5:8c:53:79:8f:a4:dd:3b:7a:9a:ba:02 (DSA)

 

Using compute resources on HYDRA

The pool of login nodes hydra.rzg.mpg.de is mainly intended for editing, compiling and submitting your parallel programs. Interactive usage of the Parallel Operating Environment (POE) on the login nodes is not enabled. Test or production jobs have to be submitted to the LoadLeveler batch system which reserves and allocates the resources (e.g. compute nodes) required for your job. Short test jobs (runtime < 15 min) which are requiring 1, 2, 4 or 8 cores will run on a dedicated node with short turn around times. Further information on the batch system is provided here.

Interactive debug runs

If you need to debug your program code you may login to hydra-i.rzg.mpg.de and run your code interactively (2 hours at most).
But please, take care that the machine does not become overloaded. Don't occupy all the 20 cores and please do not request more than 30 GB of main memory. Neglecting these recommendations may cause a system crash or hangup!

Document Actions