Unattended Python3 installation on Centos and RHEL VMs

If you are a Centos or RHEL Linux administrator or a developer using these OS platform then you know that the python version which ships with the original OS build image is 2.7.X

To do unattended / automated installation of Python 3.x and additional packages on multiple Linux nodes; use the below Ansible Playbook.

---
  # Python 3.6.x and packages unattended installation playbook
  # Make sure Repos are configured properly before running this playbook
  # Check internet connectivity and sufficient disk space on managed nodes
  - name: Install Python 3 and packages along with dependencies on Centos/Redhat
    hosts: all
    gather_facts: no
    become: true
    tasks:

      # Task1 - Install the dependent packages to build Python 3.6 from source
      - name: Install dependencies before downloading python
        yum:
          name: "{{ item }}"
          state: latest
        with_items:
          - zlib-devel
          - gcc
          - openssl-devel
          - bzip2-devel
          - python-setuptools

      # Task2 - Check if Python ver 3.6 exists on the managed nodes
      - name: Check Python 3.6 exists or not
        shell:
          /usr/local/bin/python3.6 -V
        register: output
        ignore_errors: yes
        changed_when: false
      - debug:
          var: output

      # Task3 - Get Python source file from the internet (3.6.8)
      - name: Copy Python source archive file only if not installed already from the internet
        get_url:
          url: https://www.python.org/ftp/python/3.6.8/Python-3.6.8.tgz
          dest: /usr/src
          mode: 0644
        register: download
        ignore_errors: yes

      # Task4 - Copy manually already downloaded file if Task3 failed
      - name: Copy of the manually already downloaded file if Task3 failed due to internet connectivity issue
        copy:
          src: "Python-3.6.8.tgz"
          dest: /usr/src
          mode: 0644
        when: output.rc > 0 and download.rc > 0
        ignore_errors: yes

      # Task5 - Check file downloaded to managed node or not alternatively
      # unarchive can be used to copy and unzip on managed nodes directly
      - name: Check file downloaded successfully or not
        stat:
          path: /usr/src/Python-3.6.8.tgz
        register: python_archive

      # Task6 - Unarchive the Python source file to destination /usr/src
      - name: Unarchive Python source archive file
        unarchive:
          src: /usr/src/Python-3.6.8.tgz
          dest: /usr/src
          remote_src: yes
        when: python_archive.stat.exists

      # Task7 - Using stat module to check unarchive was success or not
      - name: Check python unzipped to folder successfully
        stat:
          path: /usr/src/Python-3.6.8
        register: python_dir

      # Task8 - Command module is not idempotent to skip this task added
      # when condition to check if Python exists or not before install
      - name: Install Python by building it from source
        command: chdir=/usr/src/Python-3.6.8 {{ item }}
        with_items:
          - ./configure
          - make altinstall
        when: python_dir.stat.exists == true and output.rc > 0

      # Task9 - After building python from source check whether its installed and available in the right path
      - name: Check Python installed successfully or not
        stat:
          path: /usr/local/bin/python3.6
        register: check
      - debug:
          msg: "Python3.6 installed installed successfully"
        when: check.stat.exists


      # Task10 - Install Python packages using pip
      - name: pip
        pip:
          name: "{{ item }}"
          executable: /usr/local/bin/pip3.6
        with_items:
          - paramiko
          - netmiko
          - requests
          - xlrd
          - xlsxwriter
          - slacker
          - pandas
          - mysql-connector
          - Flask
          - Flask-Mail
          - duallog
          - emails
          - pymysql
          - sqlalchemy
        ignore_errors: yes
        when: check.stat.exists
        register: pip_install
        tags:
          - pip
      - debug:
          var: pip_install
# Playbook End #
...

Note: I’m using Ansible 2.3 so you may use loop or list method instead of item for installing packages using yum module.

Explanation of the above playbook:

There are 10 tasks in the above playbooks.

• Task 1 – Download and install dependency packages using YUM module
• Task 2 – Check Python 3 already installed or not using SHELL and REGISTER modules
• Task 3 – Download Python3 source tarball from Python FTP/ Hosting site using GET_URL module. Ignore errors if unable to download.
• Task 4 – If Step 3 fails use the already downloaded source tarball located in Control node and copy to manage nodes using COPY module
• Task 5 – Check Source tarball file downloaded successfully or not using STAT and DEBUG module. Both these modules are used multiple times along with WHEN conditional validation
• Task 6 – Unarchive the source tarball into /usr/src directory using UNARCHIVE module. We can use UNARCHIVE to do both copy and extract but here Task5 is doing the copy part.
• Task 7 – Check unarchive succeeded or not using STAT module
• Task 8 – Using the Python Source files configure and build the Python using SHELL module
• Task 9 – Check Python build succeeded or not and in the right directory path using STAT module
• Task10 – Install additional Python packages {add your list of packages if not there already) using PIP module only when Python3 successfully installed.

Thanks for stopping by, please share your comments and ideas to improve this blog. Keep watching for more use cases using Ansible and Python automation.

Advertisement

Cisco Nexus Switches Automation using Ansible

This is the success story of automating 700+ Cisco Nexus and IOS switches configuration changes in less than 4 hours. Even I never imagined this could be accomplished but by using Rundeck, Python and Ansible we made it successful.

There was a PCI vulnerability audit findings report shared with Network team manager which showed that close to 700+ switches configuration need to be tweaked to meet the PCI Compliance and Standards. So the network engineers identified and they sleeved up to apply required configuration changes manually. An estimation 3 days ETA was given by the team.

In the meantime network manager reached out to our team to check whether this process can be automated or not.

Ansible and Python’s Netmiko came to our rescue… we could automate all required configuration changes in a matter of few hours. Not to forget that we used SSHUTTLE to connect to all the switches spread across the globe from a single control node. Using Rundeck, we could push the Python scripts to worker nodes to execute the configuration commands using netmiko module.

Here is the simple yet powerful Ansible YAML configuration file which was used to implement the configuration changes on the production switches.

---
  - name: Playbook to remediate PCI Audit Findings on Cisco NXOS
    hosts: all
    gather_facts: no
    tasks:

      - name: Configure switch to disable services and console logging
        nxos_config:
          lines:
            - line console 
            - exec-timeout 10
            - line vty
            - "{{ tm }}"
            - no logging console
            - no logging monitor
            - logging logfile messages 6 size 16384
            - logging timestamp milliseconds
            - wr
            - end
          match: none
          save_when: always
        register: config
        
      - name: Check output
        debug:
          var: config   

nexus.yml – gather_facts used to fetch the device information. In this case we have disabled it to speed up the execution of playbook. There are two tasks in the above playbook 1. nxos_config and 2. debug.

nxos_config is the module developed by Core Ansible team which applies configuration changes on the nexus switches. lines – each line is a command which will be executed in configuration terminal mode. match – if match is set to none, the module will not attempt to compare the source configuration with the running configuration on the remote device. save_when is set to always to set the running config to startuo config. register is the keyword to save the output of the nxos_config to a variable called ‘config’. debug is a module used to display output or messages. Variable ‘config’ has the output of nxos_config which would be shown after playbook execution.

Ansible.cfg – Ansible configuration file.

[defaults]
inventory=inventory
log_path = ansible.log
ansible_debug=true
[persistent_connection]
log_messages = True
command_timeout=60
connect_retry_timeout = 60
[paramiko_connection]
host_key_auto_add = True
#auth_timeout = 300
#timeout = 300

inventory is the file having list of switch IP’s (or FQDN if switches are discoverable by DNS). log_path is the path of the log file to store all logs of the tasks being executed by the above playbook. ansible_debug set to true and its a best practice to enable this value for any network related automation. log_messages is to fetch the verbose logging info from the switches. command_timeout and connect_retry_timeout are mandatory to give more time to reach out to the remotely located devices. host_key_auto_add is set to true to automatically add the RSA keys to avoid prompting or failure of SSH connection. I’ve commented out auth_timeout and timeout but if you encounter delay or failure of logging due to network lag please uncomment them.

group_vars/nxos.yml – Group variables file contain credentials and other critical information. Please use ansible vault to encrypt this information

ansible_connection: local
ansible_network_os: nxos
ansible_user: <username>
ansible_password: <password>
tm: exec-timeout 10

It took about 3 hours to test the playbook. After successful test results, ran the playbook on prod switches which took about an hour to complete!! Later we randomly logged on to few switches to confirm the configuration changes made were successful or not. There were few switches which were failed to execute the commands in the playbook due to connectivity or credential errors.

SYNTAX:

ansible-playbook nexus.yml –syntax-check {Checks the YAML file syntax}

ansible-playbook nexus.yml -C {Dry run}

ansible-playbook nexus.yml {execute playbook}

Here are the screenshots – Output of ansible-playbook execution

Notice that changed is set to 1 and unreachable and failed is 0 indicating successful execution

Switches in red color failed to apply config changes due to credentials or connectivity issue

SSHUTTLE – Connect to various subnets from the jumphost to all the switches from the control node where Rundeck, Python scripts and Ansible are installed.

SYNTAX:

sshuttle -r <username>@<hostname or IP> <Subnet 1 IP> <Subnet 2 IP> <Subnet 3 IP> <Subnet n IP> -x <hostname or IP>

Notice that iptables rules are added automatically to enable SSH connectivity to switches

Here is the Python script used for Cisco IOS switches configuration automation along with Rundeck.

'''
This is Python script to amend changes to IOS XR switches as per PCI audit remediation

To run this script please make sure nexus switch is reachable via SSH port

'''
__author__ = "Vinay Umesh"
__copyright__ = "Copyright 2019, Virtustream, Dell Technologies."
__version__ = "1.0.0"
__maintainer__ = "Core Services Engineering"
__email__ = "vinay.umesh@virtustream.com"
__status__ = "Development"

from netmiko import ConnectHandler  # connect to cisco switches and execute cmds
from datetime import datetime  # Date and time module
import os  # Native OS operations and management
import logging  # Default Python logging module
import argparse  # Pass arguments
import getpass  # get password

# create a log file with system date and time stamp
logfile_ = datetime.now().strftime('nexus_switches_remediation_%H_%M_%d_%m_%Y.log')
date_ = datetime.now().strftime('%H_%M_%d_%m_%Y')


def check_arg(args=None):
    parser = argparse.ArgumentParser(description='Script to amend changes to  \
                                     Nexus switches as per PCI audit remediation')
    parser.add_argument('-s', '--source',
                        help='Source filename in CSV format required',
                        required='True')
    parser.add_argument('-u', '--user',
                        help='Username required', required='True')
    results = parser.parse_args(args)
    return (results.source, results.user)


src, user = check_arg()
# Get password
try:
    pwd = getpass.getpass()
except Exception as error:
    print('ERROR', error)

logger = logging.getLogger('IOSXR_PCI_Audit')
# set logging level
# logging.basicConfig(level=logging.INFO) # Python 2.x syntax
# toggle between DEBUG and INFO to see the difference
logger.setLevel(logging.DEBUG)
# logger.setLevel(logging.INFO)

# create file handler which logs even debug messages

# Create 'logs' folder if not exists. Change the path of logdir as your git folder

logdir = "logs/"
if not os.path.exists(logdir):
    os.makedirs(logdir)

fh = logging.FileHandler(logdir + logfile_)
fh.setLevel(logging.DEBUG)

# create formatter and add it to the handlers
formatter = logging.Formatter('%(asctime)s | %(name)s | %(levelname)s | %(message)s')
fh.setFormatter(formatter)

# add the handlers to the logger
logger.addHandler(fh)

logger.info('IOS XR switch PCI Audit remediation script started @ {}'.format(date_))

# Open the file having list of IPs of Cisco IOS switches
with open(src, 'r') as lines:
    logger.info('Read source file with device details')
    lines = list(lines)  # convert file object to list object
    del(lines[0])  # skip the header row
    for line in lines:
        value = line.split(',')
        dc = value[0]  # dc
        sw = value[1]  # switch name
        ip = value[2]  # ip
        print('Switch DC: {}'.format(dc))
        print('Switch Name: {}'.format(sw))
        print('Switch IP: {}'.format(ip))
        logger.info(' DC - {}  Switch Name - {} IP - {}'.format(dc, sw, ip))
        un = user
        pw = pwd

        #  Connecting to the switch
        try:
            # Default value is 2 else change to 4
            net_connect = ConnectHandler(device_type='cisco_ios', ip=ip, username=un, password=pw,
                                         global_delay_factor=2)
            #  show version of the switch
            ver = net_connect.send_command("show version")
            logger.info('Switch version details :\n {}'.format(ver))
            print('Show version command executed:\n{}'.format(ver))
            #  Change to config term mode
            net_connect.config_mode()
            #  Configuration config_commands
            config_commands = ['no service ipv4 tcp-small-servers',
                               'no service ipv4 udp-small-servers',
                               'no service ipv6 tcp-small-servers',
                               'no service ipv6 udp-small-servers',
                               'no http server',
                               'no tftp ipv4 server',
                               'no tftp ipv6 server',
                               'no dhcp ipv4',
                               'no dhcp ipv6',
                               'line console exec-timeout 10 0',
                               'logging console disable',
                               'logging monitor disable',
                               'no ipv4 source−route',
                               'end', 'copy running-config startup-config']
            #  Run config commands
            config = net_connect.send_config_set(config_commands)
            config += net_connect.send_command('\n', expect_string=r'#', delay_factor=2)
            logger.info('Switch Configuration Output :\n {}'.format(config))
            print('Config commands executed successfully:\n{}'.format(config))

            #  Show history of commands for IOS XR
            history2 = net_connect.send_command("show history")
            print('Show history for nexus command executed successfully: \n {}'.format(history2))
            logger.info('Switch-Nexus history output :\n {}'.format(history2))

            #  Exit from the switch
            net_connect.disconnect()

        except Exception as e:
            print('Error Occured while connecting to switch - {}  IP - {}: \n {}'.format(sw, ip, e))
            logger.info('Unable to connect to the switch - {} IP - {} :\n {}'.format(sw, ip, e))

SYNTAX:

python3 ios.py -s <inventory> -u <switch username>

Script prompts for password

We can use either Ansible or Python Netmiko way to automate Cisco switches configuration changes.

One of the SMEs of Network Engineering team reacted to this automation as under. I’m glad to see that folks are embracing and embarking towards AUTOMATION.

Hope this use case help to understand how to automate Cisco switches configuration and operational tasks. Please leave your feedback if you found this blog useful and share suggestions in the below Comments section.

Image Courtesy and References:

https://blogs.cisco.com/datacenter/ansible-support-for-ucs-and-nexus

https://docs.ansible.com/ansible/latest/modules/nxos_config_module.html

https://docs.ansible.com/ansible/2.3/ios_config_module.html