Question

I wrote several small Python scripts to start clusters inside cloud infrastructures with distributed filesystems. Now, I integrated the code into a single command-line application but the code quality is very bad.

http://code.google.com/p/diffuser/source/browse/trunk/diffuser.py

The application needs to send lots of commands via SSH with paramiko. Each command that is send via SSH needs three lines of code:

stdin, stdout, stderr = client.exec_command("<command>")
print stdout.readlines()
print stderr.readlines()

How can I improve the code quality?

Answer 1

Commands are executed in a shell, so you can use regular shell syntax to combine them together. As a first step, I’d issue a batch of commands in a single exec_command:

  stdin, stdout, stderr = client.exec_command(
           "sudo hostname;"
           "sudo apt-get update;"
           "sudo apt-get -y install nfs-common nfs-kernel-server;"
           "echo y | sudo mkfs.ext3 /dev/sdc;"
           "sudo mkdir /mnt/export;"
           "sudo mount /dev/sdc /mnt/export/;"
           "sudo chmod o+wx /etc/exports;")
  print stdout.readlines()
  print stderr.readlines()

In addition, I consider it unnecessary to start a fresh sudo for each of them. So I’d rather write

  stdin, stdout, stderr = client.exec_command(
           "sudo /bin/sh -c '"
           "hostname;"
           "apt-get update;"
           "apt-get -y install nfs-common nfs-kernel-server;"
           "echo y | mkfs.ext3 /dev/sdc;"
           "mkdir /mnt/export;"
           "mount /dev/sdc /mnt/export/;"
           "chmod o+wx /etc/exports;"
           "'")
  print stdout.readlines()
  print stderr.readlines()