Castor/docs/Using SHARCNET.ipynb

Step 1: Register

1. Create an account on the Compute Canada website
2. You will be asked for Compute Canada Role Identifier (CCRI) of your sponsor or supervisor. A CCRI is an identifier of the form abc-123-01.
3. You and your supervisor will receive an email after creating the account; please confirm your application.

Step 2: SSH Setup

1. Update your SSH configuration

cat >> ~/.ssh/config

2. Copy the following content after adding your username

Host sharc
User <username>
Hostname graham.computecanada.ca
ServerAliveInterval 30
TCPKeepAlive yes
Protocol 2,1
Identityfile ~/.ssh/id_rsa

3. Copy your SSH key to the server as an authorized key:

ssh-copy-id sharc

4. Login. You may have to wait for sometime for your account to be updated on the server.

ssh sharc

Step 3: Environment Setup

1. Download and setup Anaconda

cd projects
wget https://repo.continuum.io/archive/Anaconda3-5.0.1-Linux-x86_64.sh
chmod +x Anaconda3-5.0.1-Linux-x86_64.sh
./Anaconda3-5.0.1-Linux-x86_64.sh

2. Update ~/.bashrc by adding the following:

export PATH=/home/<username>/anaconda3/bin:$PATH

3. Source

source ~/.bashrc

Step 4: Setup SM model

1. Follow the detailed instructions from here

Step 5: SHARCNET basics

Following are the list of common commands on SLURM. For more details, refer this

Basic commands:

1.To submit a job:

sbatch <job_name.sh>

Your job should have atleast the following as a part of the preamble:

• CPU job:

#!/bin/bash
#SBATCH --time=00:01:00
#SBATCH --account=def-<name-of-the-sponsor>

• GPU job:

#!/bin/bash
#SBATCH --nodes=1
#SBATCH --gres=gpu:2
#SBATCH --ntasks-per-node=32
#SBATCH --mem=12288
#SBATCH --time=10:00
#SBATCH --account=def-<name-of-the-sponsor>

The output of the job will be stored in the slurm-<jobid>.out file

2.To monitor jobs in the queue:

squeue -u <username>

3.To see the completed jobs:

sacct

4.To cancel a job:

scancel <jobid>

In [ ]: