Linux cookbook


This cookbook is built for quick and easy search for commands and deployments of Linux, Pip, Conda, Jupyter, Remote pycharm, Apache



  • open HTTP server with specific port
python -m SimpleHTTPServer $PORT &
python -m http.server $PORT
  • Recursively count the number of file under current dirctory, including subdirctory

ls -lR | grep "^-"| wc -l

  • get the size of the file

du -sh where s refers to SUM, and h refers to be friendly to human to view

  • display port usage

netstat -tunlp | grep $PORT

  • print CUDA version

cat /usr/local/cuda/version.txt nvcc -V

  • Trash bin path


  • Look up self public ip


  • look up public ip of the domain address
  • count the number of line of files
wc -l $FILE_NAME
  • change access permission
chmod [Option] [ugoa][+-=][rwx] $FILE_NAME

Option: -R recursive

User letter
The user who owns it u
Other users in the file’s Group g
Other users not in the file’s group o
All users a
Permission letter
Read r
Write w
Execute (or access for directories) x
  • convert py2 to py3
2to3 [-w] [-n] <file_name>

-w will write to the origin file, -n will save a backup

  • change default gpu to run xorg and gnome process

modify /usr/X11/xorg.conf and replace BusID:

Section "Device"
    Identifier     "Device0"
    Driver         "nvidia"
    VendorName     "NVIDIA Corporation"
    BoardName      "GeForce TitanX"
    BusID          "PCI:2:0:0" # change the BusID to what lspci | grep VGA shown

.bashrc file

  • : is to segregate the PATH
echo $PATH
>> /home/jiahl/code-server/:/home/jiahl/.linuxbrew/bin
  • append local executable file to environment variable

~/.bashrc is a file to store alias, path etc.

For example, after appending the following to the .bashrc file, everything executable can be executed in any directory.


Note that using source ~/.bashrc to make bash resource file effect.

  • -x make .sh file executable

Generally, .sh file is not executable execept its mode has been changed:

chmod a+x <FILE_NAME>.sh

less mode

  • k move upward a line

  • j move downward a line

  • u move upward a half page

  • d move downward a half page

  • b move upward an whole page

  • f move downward an whole page

  • = display the information of the current line

  • \<string> search forward in the file

  • ?<string> search backward in the file

  • n search the next occurrence of the string

  • N search the previous occurrence of the string


  • sudo dpkg -l | grep nvidia list all drivers relative to nvidia

  • sudo apt remove --purge '*nvidia*' remove and clear specified package or drivers

  • sudo dpkg --force-all -P <package> forcely remove the library difficult to deleted in previous step

  • sudo add-apt-repository ppa:graphics-drivers/ppa add ppa for the installation

  • ubuntu-drivers devices show available drivers online

  • sudo apt install nvidia-driver-<version> nvidia-settings install nvidia driver

  • cat /usr/local/cuda/version.txt or nvidia-smi show the version of cuda

  • cat /proc/driver/nvidia/version or nvidia-smi show the version of nvidia-driver


  • Automatic log in to remote server

add following to ~/.ssh/config

Host <host_var>
    HostName    <host_name>
    Port        22
    User        <user_name>

if public key is appended to the server:/~/.ssh/authorized_keys, password won’t be needed



To install the ruby without privilege of root, the third-party package RVM is strongly suggested. reference link

  • using RVM

RVM installer script uses GPG to check signatures; because of that, the first thing to do is to download the gpg key

gpg --keyserver hkp:// --recv-keys 409B6B1796C275462A1703113804BB82D39DC0E3

then download and execute the script from

curl -sSL | bash -s stable

source the file

source ~/.rvm/scripts/rvm

disable RVM from trying to install necessary software via apt-get

rvm autolibs disable

get the listt of known Ruby versions

rvm list known

install the latest, eg:

rvm install 2.6.3


  • leave no space
STR = "foo" # wrong
STR="foo" 	# correct
  • use ```` or $() to assign a command
DIRS=$(find . * -type d | grep -v "\.")
echo ${DIRS}
  • calculate and loop

assign must not leave space expr must leave space

a=1 # no space
for i in $(seq 1 2 20)
	a=`expr ${a} + 1`
    echo "Welcome $i times"
    echo "a = $a"


Welcome 1 times
a = 2
Welcome 3 times
a = 3
Welcome 5 times
a = 4
Welcome 7 times
a = 5
Welcome 9 times
a = 6


  • pip accelerate
pip install -i $PACKAGE
  • pip ungrade package
pip install $PACKAGE --upgrade


  • check all envs
conda info -e
  • create the env
conda create -n xxxx 
  • create and clone the env
conda create -n xxxx1 --clone xxxx2
  • remove the env
conda remove -n xxxxx --all
  • activate the env
source activate xxxx
  • deactivate the env
source deactivate xxxx



  • convert *.ipynb to .py

jupyter nbconvert --to script *.ipynb

  • start the jupyter with specific port

nohup jupyter notebook --port=7141 --ip= --allow-root --notebook-dir='~/' &

  • create a passowrd to avoid using TOKEN to log in

check if the config file exists or not

jupyter notebook --generate-config

generate password

jupyter notebook password

Warning: Starting the notebook server before setting up the password

Info: After using this way to creat a port, jupyter notebook stop command is usually invalid, therefore kill is commended

  • stop the jupyter with specific port

jupyter notebook stop $PORT where PORT is the port got to stop

  • insert true tabs in jupyter editor/notebook

jupyter --config-dir to create config file for user

then create ~/.jupyter/nbconfig/edit.json:

    "Editor": {
        "codemirror_options": {
            "indentWithTabs": true

And create ~/.jupyter/nbconfig/notebook.json with:

    "CodeCell": {
        "cm_config": {
            "indentWithTabs": 2
  • add conda env
python -m ipykernel install --user --name $ENV_NAME --display-name "$DISPLAY_NAME"
  • remove conda env
jupyter kernelspec remove $ENV_NAME
  • import technique

working directory:

                 |            |
                 |            |
                 |            |


import module1
import module2

In notebook.ipynb,

import sys, os

import MyPackage.modeul1

Note that there are several reason for No module named 'XXXX', one is that packages of .py must be in one utility directory, and init file of the package must be built in previous. Another is that current working directory(cwd) must be appended to system path


  • create virtual environment
virtualenv <env_name>
  • activate the virtualenv
source <env_name>/bin/activate
  • deactivate the virtualenv

Remote pycharm

  • import tf error
print(os.environ.get('LD_LIBRARY_PATH', None))

add path of output to env variable in configuration



  • install on Ubuntu
apt-get install python-pip python-setuptools
pip install git+
  • run in the background
ssserver -p $PORT -k $PASSWORD -m aes-256-cfb --user nobody -d start

where PORT and PASSWORD are refered to the ones of the server.

  • stop the ss
ssserver -d stop
  • check the log
less /var/log/shadowsocks.log


For linux operating system:

  • create SOCKS5 proxy
nohup sslocal -s $IP -p $SERVER_PORT -k "$PASSWORD" -l $CLIENT_PORT [-d] -t 600 -m aes-256-cfb &

where $SERVER_IP and $PASSWORD are refered to the ones of the server and $CLIENT_PORT is refered to the one of the client.

Note that different from vpn global proxy,sslocal(shadowsocks on linux) creates a tunnel of SOCKS5 proxy and only works for browsers rather than running files or applications such as curl, wget or any other commands in bash.

There are two ways to hook calls from local programs to sockets and redirect it through one or more socks/http proxies.

  1. Use proxychains4 to force applications to go through SOCKS5 proxy. link

  2. Use privoxy to convert SOCKS5 into HTTP proxy, and support program using it by exporting the environment variable. link TO BE DONE BY SELF

  • check running status
systemctl status rc-local.service
  • use shadowsockssr JSON config file to run
git clone
vim config.json
cd shadowsocks/
./ -c ../config.json


Here I demonstrate to download a large file from google drive.


  • installation

For the non-root user of the server, using Linuxbrew to install it is recommended

/usr/bin/ruby -e "$(curl -fsSL"

  • modify the config file

vim ~/.linuxbrew/etc/proxychains.conf

change the last line of the config file to your specified port with SOCKS5, e.g. socks5 $LOCAL_PORT

where the $LOCAL_PORT is same to the one in SOCKS5 creating command as sslocal ... $CLIENT_PORT link


Comparing curl to proxychains4 curl

Fetching file

Each file shared on Google drive has a unique id ($FILE_ID), like 1UFaRl_3EGK0hSBlmB6AJVeZZ_r-SHQLm in

There are two ways to download it

  • By wget command (Recommand)

append proxychains4 as the prefix to any commands

proxychains4 wget --no-check-certificate -r '$FILE_ID' -O $FILE_NAME

where $FILE_ID is an identifier to the shared file, and $FILE_NAME is the name of the output file.

Note that -r is crucial and decide wget in recursive mode.

  • By script

proxychains4 curl | proxychains4 bash -s $FILE_ID

This solution is unstable in some cases, the above $FILE_ID mentioned is a case.



  • install the apache package

apt install apache2

  • adjust the firewall, check the available ufw application profiles

ufw app list

  • Check your Web Server, Check with the systemd init system to make sure the service is running by typing:

systemctl status apache2

config your own index.html

  • build an index.html file in anywhere, take ‘/home/test_html/’ as an example

vim /home/test_html/index.html

add the below in the file

  Hello world!
  • append the specified path to the virtualHost statement

vim /etc/apache2/sites-available/000-default.conf

modify the listen port into 7146 in the first line as an example

<VirtualHost *:7146>

change the virtualHost root to the specified path in DocumentRoot

for the above example, we modify the below,

DocumentRoot /home/test_html

  • config the port.conf to add listen port

vim /etc/apache2/ports.conf

then add Listen ports into the config file

  • modify the main config file to add your own web file

for the default security mode, apache doesn’t allow access outside of /usr/share and /var/www.

therefore we should add an item to allow the specified path to be accessed.

vim /etc/apache2/apache2.conf

for the above example, we add the below,

<Directory /home/test_html/>
        Options Indexes FollowSymLinks
        AllowOverride None
        Require all granted
  • then restart the apache service after modify any config files

systemctl restart apache2

  • view apache log, if there is something wrong

tail -f /var/log/apache2/error.log