Introduced Jupyter Notebook to CentOS 7

1. Purpose

Install Jupyter Notebook in CentOS on the virtual machine. In addition, it will be possible to access and execute with a browser from a remote machine such as a host machine.

2. Environment

The installation method of VirtualBox and CentOS is omitted.

3. Introduction

3.1. Update existing packages with Yum

Yum Update

$ sudo yum update -y

3.2. Introduce what you need with Anaconda

Introduced Anaconda

#Get Anaconda with curl
$ curl https://repo.continuum.io/archive/Anaconda3-4.3.1-Linux-x86_64.sh -O
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100  474M  100  474M    0     0  3431k      0  0:02:21  0:02:21 --:--:-- 3837k

#Introduced bzip2
#Required for Anaconda installation
#If it is minimal, it is not installed
$ sudo yum install bzip2 -y

#run sh file
$ bash ./Anaconda3-4.3.1-Linux-x86_64.sh 

#Check Python version
$ python --version
Python 3.6.0 :: Anaconda 4.3.1 (64-bit)

#Check Jupyter version
$ jupyter --version
4.2.1

3.3. Running JupyterNotebook

Run Jupyter Notebook

$ jupyter notebook

Make sure that Jupyter Notebook can be started. However, in this state, the Jupyter Notebook cannot be accessed from the remote machine. Therefore, edit the config file so that it can be connected from the outside.

4. Allow remote connection

4.1. Creating a config file

#Make sure your home directory is in the path of the Config file
$ jupyter --path

#Create a config file in your home directory
$ mkdir ~/.jupyter
$ touch ~/.jupyter/jupyter_notebook_config.py

4.2. Editing the config file

py3:~/.jupyter/jupyter_notebook_config.py

c = get_config()

c.NotebookApp.ip = '*'
c.NotebookApp.open_browser = False
c.NotebookApp.port = 8888

--Accept connections from all IP addresses. --Jupyter starts on port 8888. --The browser does not start when Jupyter Notebook starts.

4.3. Editing firewalld

Add a rule to firewalld to allow access to port 8888 from the outside.

Open port 8888

$ sudo firewall-cmd --add-port=8888/tcp --zone=public --permanent
success
$ sudo firewall-cmd --reload
success
$ sudo firewall-cmd --list-all --zone=public
public (active)
  target: default
  icmp-block-inversion: no
  interfaces: enp0s3 enp0s8
  sources: 
  services: dhcpv6-client http ssh
  ports: 8888/tcp
  protocols: 
  masquerade: no
  forward-ports: 
  sourceports: 
  icmp-blocks: 
  rich rules: 

4.4. Setting login password

4.4.1. Generate iPython encrypted strings

#Launch iPython
$ ipython

#Load library to generate encrypted strings
In [1]: from notebook.auth import passwd

#Enter password to generate encrypted string
In [2]: passwd()
Enter password: 
Verify password: 
Out[2]: 'sha1:(Encrypted string)'

#Get out of iPython
In [3]:                                                                                                                                               
Do you really want to exit ([y]/n)? y

4.4.2. Write the encrypted string to the Jupyter config file

py3:~/.jupyter/jupyter_notebook_config.py

c.NotebookApp.password = u'sha1:（4.4.1.(Character string generated by)'

4.5. Run Jupyter Notebook

Run Jupyter Notebook

$ jupyter notebook

After starting Jupyter Notebook, access the IP address 8888 assigned to the running machine.

You will be prompted to enter the password, so enter the password you set earlier.

If you log in successfully, a list of files in your home directory is displayed.

5. Make JupyterNotebook a service with systemd

5.1. Create Unit definition file

#Become root
$ su root

#Check Jupyter path
\# which jupyter

#Create Unit definition file
\# touch /etc/systemd/system/notebook.service

5.2. Editing the Unit definition file

[Unit]
Description = Jupyter Notebook

[Service]
Type=simple
PIDFile=/var/run/jupyter-notebook.pid
ExecStart='Absolute path to Jupyter' notebook
WorkingDirectory='The path of the home directory of the user who installed Anaconda'
User='Username that introduced Anaconda'
Group='Username that introduced Anaconda'
Restart=always

[Install]
WantedBy = multi-user.target

--In ExecStart, enter the command with notebook as an argument in the Jupyter path confirmed in 5.1. --Specify the path of the home directory of the user who installed Anaconda in WorkingDirectory. --In User and Group, enter the name of the user who installed Anaconda.

5.3. Get a list of registered Units

\# systemctl list-unit-files --type=service

notebook.service                            disabled

Confirm that the Unit file defined earlier exists in the output list.

5.4. Set to run at startup

#Set Jupyter to start automatically at startup
\# systemctl enable notebook

#Start service
\# systemctl start notebook

#Check if Jupyter is working properly
\# systemctl status notebook
● notebook.service - Jupyter Notebook
   Loaded: loaded (/etc/systemd/system/notebook.service; enabled; vendor preset: disabled)
   Active: active (running)since month 2017-03-20 21:27:49 JST; 2s ago

#Reboot and make sure Jupyter is running
\# reboot

6. Apply SSL

6.1. Create SSL server certificate and private key using OpenSSL

In Using SSL for encrypted communication, you can create a self-signed certificate and private key. The following command is given as an example.

$ openssl req -x509 -nodes -days 365 -newkey rsa:1024 -keyout mykey.key -out mycert.pem
Generating a 1024 bit RSA private key

Regarding the meaning of the arguments of the OpenSSL command, [Create a private key and CSR using CentOS7.2 64bit OpenSSL and create a self-signed SSL server certificate](http://www.kakiro-web.com/ You can find out by reading "Create a self-signed SSL server certificate" in linux / ssl.html).

For the above command

--Create private key and CSR at once --X509 certificate standard --Certificate is valid for 365 days

You can see that.

6.2. Editing the config file

Place the certificate and private key created in 6.1. In the _ ~ / .jupyter_ folder.

Moving self-signed certificate and private key file

$ mv mycert.pem ~/.jupyter/
$ mv mykey.key ~/.jupyter/

Edit the certificate and private key location in the config file.

py3:~/.jupyter/jupyter_notebook_config.py

c.NotebookApp.certfile = u'${Absolute path to home directory}/.jupyter/mycert.pem'
c.NotebookApp.keyfile = u'${Absolute path to home directory}/.jupyter/mykey.key'

6.3. Restart the service and access via HTTPS

$ sudo systemctl restart notebook

$ sudo systemctl status notebook
● notebook.service - Jupyter Notebook
   Loaded: loaded (/etc/systemd/system/notebook.service; enabled; vendor preset: disabled)
   Active: active (running)since month 2017-03-20 22:15:08 JST; 6s ago

7. Change the default working directory

Change the working directory of Jupyter Notebook, which currently has a home directory specified by default, to another location.

7.1. Editing the config file

Create a directory for Jupyter directly under your home directory

$ mkdir ~/jupyter_files

Edit the config file.

py3:~/.jupyter/jupyter_notebook_config.py

c.NotebookApp.notebook_dir = u'${Absolute path to the directory created earlier}'

Restart Jupyter Notebook

$ sudo systemctl restart notebook

7.2. Confirm changes

Access Jupyter Notebook. You can see that the "anaconda3" directory that was displayed earlier is no longer displayed.

If you create a file in the working directory specified earlier in the terminal, you can check the changes in jupyter notebook.

$ cd ~/jupyter_files/
$ touch TEST_FILE.py

8. Reference website

• Jupyter Running the Notebook
• Config file and command line options
• Using SSL for encrypted communication -Connect IPython notebook remotely -[Note] Jupyter notebook -How to open the port on CentOS7. -Try to make your own command into a service using Systemd -Notes on OpenSSL command -Public key cryptography and digital signature method using OpenSSL command