#pexpect

In the event that you wish to run a single command against multiple network devices from a jump box.

NB: You'll need to install the pexpect Python module for this script to work.

Script:

#!/usr/bin/env python # bulk-ssh-cmd.py (2,000 character output limit per device). # Usage: python bulk-ssh-cmd.py 'command string' '/path/file.txt' # Import required module(s). Only pexpect needs to be downloaded and installed, all others should be present. import sys import os import getpass import pexpect # Validate correct number of command line parameters. argList = sys.argv[1:] if len(argList) 2: sys.exit("Please specify a command string AND a valid source file.") # Set RSA key fingerprint expected output string for new connections. cliCMD = sys.argv[1] ssh_newkey = 'Are you sure you want to continue connecting' # Obtain and read in hosts from file specified in CLI and validate filetype and existence. inputFile = sys.argv[2] print inputFile if not inputFile.endswith(".txt"): sys.exit("Error: %s is not a valid text file." % inputFile) elif os.path.exists(inputFile): hostFile = open(inputFile, "r") else: sys.exit("Error: %s file not found." % inputFile) # Get credentials. userID = os.getenv("USER") openSsme = getpass.getpass(prompt="What is your password? \n") # Handle RSA key fingerprint, login prompt and EOF output. for hostBox in hostFile: # Specify connection string & expected responses. connStr=pexpect.spawn('ssh %s@%s' % (userID, hostBox), timeout = 5) resVal=connStr.expect([ssh_newkey,'assword:',pexpect.EOF]) if resVal==0: print "Sending yes to accept RSA key fingerprint." connStr.sendline('yes') resVal=connStr.expect([ssh_newkey,'assword:',pexpect.EOF]) if resVal==1: print "Password sent. Command:", connStr.sendline(openSsme) connStr.expect("#") connStr.sendline(cliCMD) connStr.expect("#") connStr.sendline("exit") print connStr.before , "\n" # Print the command output. else: print "Key or connection timeout received." pass

Data file (can be IP addresses or hostnames):

admin@linux-host (~/scripts)$ cat hosts.txt 192.168.42.5 192.168.42.6 192.168.42.7

Usage:

python bulk-ssh-cmd.py 'command string' '/path/file.txt'

Error handling:

Too few command line parameters:

admin@linux-host (~/scripts)$ python bulk-ssh-cmd.py 'sh ip int br vrf management' Please specify a command string AND a valid source file.

Too many command line parameters:

admin@linux-host (~/scripts)$ python bulk-ssh-cmd.py 'sh ip int br vrf management' '/home/admin/scripts/hostings.txt' test Please specify a command string AND a valid source file.

Invalid file extension:

admin@linux-host (~/scripts)$ python bulk-ssh-cmd.py 'sh ip int br vrf management' '/home/admin/scripts/test.csv' /home/admin/scripts/test.csv Error: /home/admin/scripts/test.csv is not a valid text file.

Non-existant file:

admin@linux-host (~/scripts)$ python bulk-ssh-cmd.py 'sh ip int br vrf management' '/home/admin/scripts/hostings.txt' /home/admin/scripts/hostings.txt Error: /home/admin/scripts/hostings.txt file not found.

!/usr/bin/env python

-- coding: utf-8 --

import pexpect from sys import argv

p=pexpect.spawn('telnet %s' % argv[1])

p.sendline('LOGIN')

p.sendline('PASSWORD')

p.interact()

Automatic testing - in all it's glory - has been one of my problems for some time now.

The result for the automated testing of the current Fedora Cloud image is:

A screencast can be found here.

tl;dr

Not to long ago I wrote about using pexpect to interact with virtual machines. This time I extend the last experiment and add gherkin (through python-behave) to describe the features to be tested.

I like gherkin, because it enables non-programmers to write test cases, to describe their needs, the features they want to so. Much has been written about it elsewhere and there are nice recommendations of how good features should be written.

A feature to describe the basic bootability of a cloud image could look like:

Feature: Basic bootability Cover the basic boot process. Ensure that we see a bootloader and a login prompt is displayed. Background: Some virtual machine with Fedora Given a default VM and the latest Fedora cloud image and that the VM is turned on Scenario: A bootloader is expected right after boot When we wait for 30 seconds at most Then we expect the bootloader prompt to appear Scenario: The kernel and systemd is started When we wait for 60 seconds at most Then we expect the kernel to be loaded and systemd to be running Scenario: A login prompt is displayed When we wait for 480 seconds at most Then we expect a login prompt

(The feature is suboptimal written.)

How each of this lines is achieved is up to the programmer / the implementation.

The implementation in this case is the same as I've previously written about. qemu and pexpect ironed together.

As the implementation is quite standalone (only python and qemu are required), it was just logical to run it on Travis CI.

The result is that it passes (after a bit of forth and back). Look at the history to see, that failures are also catched.

The .travis.yml file is surprisingly clean:

language: python python: - "2.7" before_install: - "sudo apt-get update" - "sudo apt-get install qemu curl" - "pip install -r requirements.txt" script: - "curl -L -O http://download.fedoraproject.org/pub/fedora/linux/updates/20/Images/x86_64/Fedora-x86_64-20-20140407-sda.qcow2" - "behave" - "rm -vf *.qcow2"

All of this can be found in this repository. Feel free to test, fork and improve!

Things I like about this approach:

Separation of concern - Implementation is independent of specification

Works well for local setups and testing in the cloud

Igor does his job, testing Node images on virtual machines and real hardware. But one drawback is that it's "big". It is not huge, but it is not directly usable out of the box.

Out of the blue - while playing with some ideas - I coupled pexpect (now updated to 3.1 in Fedora) with qemu (and a stdio serial).

What I got is a simple mechanism to control a VM. With much less assumptions then Igor has.

The drawback is obviously that only VMs can be tested, but the benefit is that it is very simple, and there no high requirements. And this is also limited to the console - which is optimal for Node.

Long story short, the snippet below takes an image, boot's it and tries to login via the serial console.

This can further be extended by using 9pfs to exchange data between the host and guest.

A minimal functional test is now also in imgbased to test some basic functionality at runtime.

В процессе изучения Python’а возникла задача применения новоприобретенных навыков в повседневной работе. Одним из аспектов последней является работа с активным сетевым оборудованием различных вендоров. Очень уж лениво стало вводить логин и пароль каждый раз для того, чтобы попасть на свитч. Изначально в качестве решения был выбран специализированный под подобную рутину язык Expect, который в свою очередь является веткой TCL. Однако в последствии было принято решение переписать код с использованием более гибкого в применении языка. Выбор пал на Python и его модуль pexpect. На данный момент работаю над распознаванием вендора оборудования средствами SNMP плюс генерирование конфы под каждого из них.