Noether Compute Cluster: Basic Usage

What is Noether?

Noether is a Linux compute cluster being made available to members of the HEP Group for exploratory high-throughput computation.

The cluster presently consists of a login node, which can be accessed from the internet, job-scheduling nodes, which are restricted to the sysadmins, and a number of work-nodes, which are available to end-users via the login node only under the aegis the HTCondor batch schduler.

Getting an Account and First Login

As expained in the FAQ, to request an account on Noether, please ask your supervisor/line manager to send an email to the BLACKETT-SUPPORT mailing list. When your request is approved, you will be supplied out-of-band with an initial set of login credentials.

Then to connect to Noether for the first time please ssh with your supplied username to noether.hep.manchester.ac.uk. Please note that if you are connecting from off-Campus you will need to install an initialise the GlobalProtect VPN tool provided by IT Services. On your initial connection attempt the supplied password will need to be changed, as in the following sample ssh session:

    $ ssh mrtest@noether.hep.manchester.ac.uk
    Password: *******                                            # <---- enter the issued password
    WARNING: Your password has expired.
    You must change your password now and login again!
    Changing password for user mrtest.
    Current Password: *******                                    # <---- enter the issued password AGAIN
    New password: ************                                   # <---- now choose a NEW password
    Retype new password: ************                            # <---- re-enter your NEW password
    passwd: all authentication tokens updated successfu lly.
    Connection to noether.hep.manchester.ac.uk closed.

The next time you ssh to Noether, entering the new password will give you a shell session on the Noether:

    $ ssh mrtest@noether.hep.manchester.ac.uk
    mrtest@noether.hep.manchester.ac.uk's password: ************
    [mrtest@vm119 ~]$                                            # <---- On Noether's login node!

Interactive Sessions

Use of the login node for heavy computational work is prohibited: users are expected to conduct intensive interactive shell sessions the clusters work-nodes, some of which have been set aside for this purpose. Now directly ssh-ing to a work-node is not permitted: users should issue the command condor_submit -i or the alias qrsh as in the following example:

    $ condor_submit -i  getenv=True
    1 job(s) submitted to cluster 2724.
    Welcome to slot1_1@wn3801320.hep.manchester.ac.uk!
    You will be logged out after 7200 seconds of inactivity.
    [mrtest@wn3801320 dir_394806]$ pwd
    /scratch/condor_pool/condor/dir_394806

Here the shell session of user mrtest was teleported to work-node wn3801320 under the auspices of the HTCondor scheduler (hence the time limit). Note that the working directory getenv=True is not mrtest’s home directory on Noether! – it is a scratch directory which is local to the node. Heavy IO work should be confined to these local scratch directories. However, mrtest can easily access his cluster-wide home directory simply by issuing cdas follows:

    [mrtest@wn3801320 dir_394806]$ cd
    [mrtest@wn3801320 ~]$ pwd       
    /gluster/home/mrtest                            # Now in cluster-wide $HOME directory
    [mrtest@wn3801320 ~]$ cd $_CONDOR_SCRATCH_DIR
    [mrtest@wn3801320 dir_394806]$ pwd 
    /scratch/condor_pool/condor/dir_394806          # Now back in job-specific node-local scratch directory

It is thus necessary to copy critical code and data back into one’s home or Lab directory before the interactive session terminates.