Skip to content

Latest commit

 

History

History
98 lines (48 loc) · 3.18 KB

Guidelines For Workstation Use.md

File metadata and controls

98 lines (48 loc) · 3.18 KB

Guidelines For Using The Workstation

How to use the workstation remotely via ssh

For logging in to the machine

1) Run in the terminal  ssh -p 433 associate@10.4.81.122 

2) Accept the key 

3) Enter the Password

For using jupyter notebooks

1) Run in terminal    ssh -p 443 -L 8880:127.0.0.1:8880 associate@10.4.81.122

2) Accept the key 

3) Enter the Password

Useful commands for Linux:

1) lsblk - to check the disk drives info. 

2) htop  - to check the CPU Utilization.

3) nvidia-smi --query-gpu=timestamp,temperature.gpu,utilization.gpu,utilization.memory,memory.total,memory.free,memory.used,power.draw,clocks.sm,clocks.mem,clocks.gr  --format=csv -l 1 

    To check realtime GPU stats.

4) w - to see current users.

5) mkdir - to make your own directories.

6) Conda related commands can be found in the CONDA CHEAT SHEET.pdf

General Guidelines for remote use via ssh

- Sharing of ssh credentials will lead to strict action against members or associates.

- Avoid turning it off and on again 

    - In case a reboot is required follow this checklist:

        1) Check if any other user is logged on by using the command w in terminal.

        2) Check for system usage stats to verify that important processes are not running in the background htop / nvidia-smi / i7z.

        3) Check tmux sessions by using the command tmux attach.

- DO NOT Log-off or put the machine in sleep mode just lock it when required (win+l).

- DO NOT interfere with directories which do not belong to your work.

- DO NOT install packages on your own, ask the Administrators to set up the environment for you. 

- Loading personal data on the machine is strictly prohibited.

- DO NOT attempt to train on GPU when GPU is already busy with other training tasks. 

- Avoid working while CPU utilization is high might lead to system instability

Things that are ready to use

- Kaggle API is installed to download datasets easily over ssh.

- Conda is up and running.

- Basic python packages are already installed.

- Tensorflow and Pytorch have been installed along with Keras and fastai.

General Guidelines for woking on Samantha locally

- Keep the desk clean at all times. 

- The machine desk is strictly reserved for work purposes only don't sit near the machine for chitchat.

- Chrome-Driver may open up during your use don't worry about it is to log in to the network automatically.

Troubleshooting

-  Error "Connection failed" or "unavailable" This means that the machine is probably turned off.

-  Error "No route to host is available" 

    This means that the  machine is disconnected from the Wi-Fi Network 

    The only way to deal with this problem currently is by reconnecting to the 'SRM CAMPUS' wifi on the machine manually.

-  The Internet is not working on the machine, cd to home directory and run python3 login.py in the terminal. A background process is already 
 
   running to login to the wifi automatically if it's causing trouble contact administrators.

-  If any other issue arises contact the administrators immediately.