HPC2020: How to connect

From outside ECMWF, you may use Teleport through our gateway in Bologna, jump.ecmwf.int. Direct access through ECACCESS service is not available.

$> tsh login --proxy=jump.ecmwf.int
$> ssh -J user@jump.ecmwf.int user@hpc-login
# or for users with no formal access to HPC service:
$> ssh -J user@jump.ecmwf.int user@ecs-login

For all the details of this connection method please see the Teleport documentation, where you will find how to best configure your SSH settings.

You may also connect via SSH from another ECMWF platform such as the Linux VDI.

$> ssh hpc-login
# or for users with no formal access to HPC service:
$> ssh ecs-login

$> ssh-key-setup

See also HPC2020: Persistent interactive job with ecinteractive if you wish to customise the resource limits of your interactive session.

You can also open a JupyterLab session through the ECMWF JupyterHub service. See HPC2020: Jupyter Sessions with JupyterHub for more details.

List of main entry points

You can use the hpc-* or hpc2020-* names in the table below if you just need to use the default HPCF complex, or the specific names for each of the complexes. For ECS, you may use the ecs-* names. Your usual interface to connect would be the *-login names, and for remote submission of jobs we would recommend using the *-batch names as they are dedicated login nodes for job submissions. If you need to set up a cronjob, then the *-cron names are to be used.