Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

Axon is a Beowulf style cluster where a scheduler, in this case Slurm, is used to submit compute tasks, called jobs, to one or more nodes in the cluster. Jobs are submitted to the cluster via the login node (axon.rc.zi.columbia.edu) as show in the example below.

...

The login node is for preparation of compute jobs, please don't run any cpu intensive tasks on the login node. Compute intensive preparation can be done via Slurm interactive jobs on the compute nodes.

Connecting via SSH

SSH (Secure Shell), a secure protocol to used primarily to connect to Linux servers, is the way to connect Axon to submit jobs. Connecting via SSH is typically done from the command line interface such as Command Prompt in Windows or Terminal on the Mac and running the command ssh as shown below.

...

titleSSH from other Columbia hosts

...

 Windows users: SSH was only added to Windows 10 recently, so if you're somehow using an older version a third party program such as Putty or MobaXterm is required to connect.

Code Block
languagebash
themeFadeToGrey
titleTo login to the cluster ssh into the login node (replace uni123 with your uni)
➤ ssh aa3301@axon.rc.zi.columbia.edu
aa3301@axon.rc.zi.columbia.edu's password:
Last login: Fri Dec 27 15:35:33 2019 from adm.rc.zi.columbia.edu
[aa3301@axon ~]$


Info
Code Block
languagebash
themeFadeToGrey
titleTo login to the cluster ssh into the login node (replace uni123 with your uni)
ssh uni123@axon

The login node is only accessible from wired connections inside the Columbia Morningside campus, secure wifi or the CUIT VPN at this time. While the login node is not currently accessible from the Internet, it is accessible from most parts of the Columbia so you can ssh from another host which is available on the Internet such as other HPC login nodes or Cunix.

Once connected via ssh to the cluster login node, axon.rc.zi.columbia.edu, compute jobs can be launched using the Slurm job scheduler. 

.

rc.zi.columbia.edu