Execute a command on Remote Machine in Python

Sure, there are several ways to do it!

Let's say you've got a Raspberry Pi on a raspberry.lan host and your username is irfan.

subprocess

It's the default Python library that runs commands.
You can make it run ssh and do whatever you need on a remote server.

scrat has it covered in his answer. You definitely should do this if you don't want to use any third-party libraries.

You can also automate the password/passphrase entering using pexpect.

paramiko

paramiko is a third-party library that adds SSH-protocol support, so it can work like an SSH-client.

The example code that would connect to the server, execute and grab the results of the ls -l command would look like that:

import paramiko

client = paramiko.SSHClient()
client.set_missing_host_key_policy(paramiko.AutoAddPolicy())
client.connect('raspberry.lan', username='irfan', password='my_strong_password')

stdin, stdout, stderr = client.exec_command('ls -l')

for line in stdout:
    print line.strip('\n')

client.close()

fabric

You can also achieve it using fabric.
Fabric is a deployment tool which executes various commands on remote servers.

It's often used to run stuff on a remote server, so you could easily put your latest version of the web application, restart a web-server and whatnot with a single command. Actually, you can run the same command on multiple servers, which is awesome!

Though it was made as a deploying and remote management tool, you still can use it to execute basic commands.

# fabfile.py
from fabric.api import *

def list_files():
    with cd('/'):  # change the directory to '/'
        result = run('ls -l')  # run a 'ls -l' command
        # you can do something with the result here,
        # though it will still be displayed in fabric itself.

It's like typing cd / and ls -l in the remote server, so you'll get the list of directories in your root folder.

Then run in the shell:

fab list_files

It will prompt for an server address:

No hosts found. Please specify (single) host string for connection: [email protected]

A quick note: You can also assign a username and a host right in a fab command:

fab list_files -U irfan -H raspberry.lan

Or you could put a host into the env.hosts variable in your fabfile. Here's how to do it.


Then you'll be prompted for a SSH password:

[[email protected]] run: ls -l
[[email protected]] Login password for 'irfan':

And then the command will be ran successfully.

[[email protected]] out: total 84
[[email protected]] out: drwxr-xr-x   2 root root  4096 Feb  9 05:54 bin
[[email protected]] out: drwxr-xr-x   3 root root  4096 Dec 19 08:19 boot
...

You may use below method with linux/ Unix 's built in ssh command.

   import os
   os.system('ssh username@ip  bash < local_script.sh >> /local/path/output.txt 2>&1')
   os.system('ssh username@ip  python < local_program.py >> /local/path/output.txt 2>&1')

Paramiko module can be used to run multiple commands by invoking shell. Here I created class to invoke ssh shell

class ShellHandler:

def __init__(self, host, user, psw):
    logger.debug("Initialising instance of ShellHandler host:{0}".format(host))
    try:
        self.ssh = paramiko.SSHClient()
        self.ssh.set_missing_host_key_policy(paramiko.AutoAddPolicy())
        self.ssh.connect(host, username=user, password=psw, port=22)
        self.channel = self.ssh.invoke_shell()
    except:
        logger.error("Error Creating ssh connection to {0}".format(host))
        logger.error("Exiting ShellHandler")
        return
    self.psw=psw
    self.stdin = self.channel.makefile('wb')
    self.stdout = self.channel.makefile('r')
    self.host=host
    time.sleep(2)

    while not self.channel.recv_ready():
        time.sleep(2)
    self.initialprompt=""
    while self.channel.recv_ready():

        rl, wl, xl = select.select([ self.stdout.channel ], [ ], [ ], 0.0)
        if len(rl) > 0:
            tmp = self.stdout.channel.recv(24)
            self.initialprompt=self.initialprompt+str(tmp.decode())



def __del__(self):
    self.ssh.close()
    logger.info("closed connection to {0}".format(self.host))

def execute(self, cmd):
    cmd = cmd.strip('\n')
    self.stdin.write(cmd + '\n')
    #self.stdin.write(self.psw +'\n')
    self.stdin.flush()
    time.sleep(1)
    while not self.stdout.channel.recv_ready():
        time.sleep(2)
        logger.debug("Waiting for recv_ready")

    output=""
    while self.channel.recv_ready():
        rl, wl, xl = select.select([ self.stdout.channel ], [ ], [ ], 0.0)
        if len(rl) > 0:
            tmp = self.stdout.channel.recv(24)
            output=output+str(tmp.decode())
    return output

If creating different shell each time does not matter to you then you can use method as below.

def run_cmd(self,cmd):
    try:
        cmd=cmd+'\n'
        #self.ssh.settimeout(60)
        stdin,stdout,stderr=self.ssh.exec_command(cmd)
        while not stdout.channel.eof_received:
           time.sleep(3)
           logger.debug("Waiting for eof_received")
        out=""
        while stdout.channel.recv_ready():
            err=stderr.read()
            if err:
                print("Error: ",my_hostname, str(err))
                return False 

            out=out+stdout.read()
        if out:
               return out 

    except:
        error=sys.exc_info()
        logger.error(error)
        return False 

Simple example from here:

import subprocess
import sys

HOST="www.example.org"
# Ports are handled in ~/.ssh/config since we use OpenSSH
COMMAND="uname -a"

ssh = subprocess.Popen(["ssh", "%s" % HOST, COMMAND],
                       shell=False,
                       stdout=subprocess.PIPE,
                       stderr=subprocess.PIPE)
result = ssh.stdout.readlines()
if result == []:
    error = ssh.stderr.readlines()
    print >>sys.stderr, "ERROR: %s" % error
else:
    print result

It does exactly what you want: connects over ssh, executes command, returns output. No third party library needed.