vpn

• networkmanager-fortisslvpn-git from aur
  • networkmanager-fortisslvpn had problems setting the DNS correctly - aur package magically fixes it so did not investigate further
• similar package exists for ubuntu/fedora

email

• iitk email - https://linux.cc.iitk.ac.in/lininfo/IITK_email_introduction.html
• cse email - https://www.cse.iitk.ac.in/pages/CSEDeptInfo.html
• linux/mac - thunderbird, fdroid - fairemail

turing server

• copy ssh keys: ssh-copy-id $MYUSERNAME@turing.cse.iitk.ac.in
• symlink to webpage home: ln -s /homepages/global/$MYUSERNAME ~/www

software

• office 365 - http://web.iitk.ac.in/ccnew/Office365/Office_365_subscription_at_IITK.htm (internal link)
• github student pack - https://education.github.com/pack
  • intellij ultimate etc.

python and gpu server usage

• login to gpu1/gpu2: ssh $MYUSERNAME@gpu1.cse.iitk.ac.in or ssh $MYUSERNAME@gpu2.cse.iitk.ac.in
  • use image.cse.iitk.ac.in or image1.cse.iitk.ac.in or image2.cse.iitk.ac.in if GPU is not needed
• create directory under /data: mkdir /data/$MYUSERNAME && chmod 700 /data/$MYUSERNAME
• download and install miniconda: cd /data/$MYUSERNAME && curl -O https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh && bash Miniconda3-latest-Linux-x86_64.sh
  • accept the license by pressing q and then typing yes
  • give install path as /data/$MYUSERNAME/miniconda3
  • type yes for executing conda init
• download and install mambaforge:
  • cd /data/$USER
  • wget -O Mambaforge.sh "https://github.com/conda-forge/miniforge/releases/latest/download/Mambaforge-$(uname)-$(uname -m).sh"
  • bash Mambaforge.sh -p /data/$USER/mambaforge
    • <ENTER> -> yes -> <ENTER> -> yes -> <ENTER>
  • . ~/.bashrc
  • echo '[[ -f ~/.bashrc ]] && source ~/.bashrc' >> ~/.bash_profile
    • to ensure ~/.bashrc is sourced on ssh
• create new environment for a project: conda create --name dlproject
• activate environment: conda activate dlproject
  • need to do this each time you start working on a project
• install any required packages
  • tensorflow, pandas, jupyter: conda install jupyterlab pandas tensorflow-gpu
  • librosa: conda install librosa
• start notebook without opening browser: jupyter lab --ip=0.0.0.0 --no-browser
• miscellaneous:

  • check GPU usage: watch -d -n 0.5 nvidia-smi

  • check and limit GPUs to be used in python:

    gpus = tf.config.list_physical_devices("GPU")
    gpus   # list of GPUs available - 3 as on 2022.04.06
    
    # only use the 3rd GPU
    tf.config.set_visible_devices(gpus[2], "GPU")

  • disk usage: df -h /data (overall) or du -sh /data/$MYUSERNAME (per user)

hostel ips

• https://www.iitk.ac.in/cc/IP_Details/Hostel_IP_Address.html
• dns: 172.31.1.130,172.31.1.1,172.20.160.23

wpa_supplicant.conf file for raspbian

country=IN

network={
  ssid="iitk(Highspeed-5GHz)"
  scan_ssid=1
  key_mgmt=WPA-EAP
  eap=TTLS
  identity="youruserid"
  password="yourpassword"
  phase2="auth=PAP"
}

dhcpd static ip file for raspbian

interface eth0
# dummy values below
static ip_address=198.51.100.235/21
static routers=198.51.100.1
static domain_name_servers=172.31.1.1 172.31.1.130
static domain_name=h5.iitk.ac.in
static domain_search=cc.iitk.ac.in iitk.ac.in

# make wlan0 higher priority interface
interface wlan0
metric 100