How do I iterate through the mount points of a Linux system using Python? I know I can do it using df command, but is there an in-built Python function to do this?
Also, I'm just writing a Python script to monitor the mount points usage and send email notifications. Would it be better / faster to do this as a normal shell script as compared to a Python script?
Thanks.
The Python and cross-platform way:
pip install psutil # or add it to your setup.py's install_requires
And then:
import psutil
partitions = psutil.disk_partitions()
for p in partitions:
print(p.mountpoint, psutil.disk_usage(p.mountpoint).percent)
psutil is a great package for getting system information that works across all major platforms and deserves to be better known.
/proc don't show up in this tool
Running the mount command from within Python is not the most efficient way to solve the problem. You can apply Khalid's answer and implement it in pure Python:
with open('/proc/mounts','r') as f:
mounts = [line.split()[1] for line in f.readlines()]
import smtplib
import email.mime.text
msg = email.mime.text.MIMEText('\n'.join(mounts))
msg['Subject'] = <subject>
msg['From'] = <sender>
msg['To'] = <recipient>
s = smtplib.SMTP('localhost') # replace 'localhost' will mail exchange host if necessary
s.sendmail(<sender>, <recipient>, msg.as_string())
s.quit()
where <subject>, <sender> and <recipient> should be replaced by appropriate strings.
mount -o bind within the chroot, /proc/mounts will not list any of the chrooted mounts. However, the mount command (within the chroot) will. df won't either which is why my preferred one is to run mount.mount -o bind on a Linux system generally doesn't show in /proc/mounts (at least not any Linux kernel/distro I know of yet). However, mount will show you the mounts done with -o bind as well./proc/mounts does list filesystems mounted with mount -o bind. However, it does so in a confusing manner (the 'device' field in the physical device, not the bound directory, and the filesystem type / mount options do not indicate its a bind mount). Well, at least, on Fedora 20, kernel 3.16.2 it does.I don't know of any library that does it but you could simply launch mount and return all the mount points in a list with something like:
import commands
mount = commands.getoutput('mount -v')
mntlines = mount.split('\n')
mntpoints = map(lambda line: line.split()[2], mntlines)
The code retrieves all the text from the mount -v command, splits the output into a list of lines and then parses each line for the third field which represents the mount point path.
If you wanted to use df then you can do that too but you need to remove the first line which contains the column names:
import commands
mount = commands.getoutput('df')
mntlines = mount.split('\n')[1::] # [1::] trims the first line (column names)
mntpoints = map(lambda line: line.split()[5], mntlines)
Once you have the mount points (mntpoints list) you can use for in to process each one with code like this:
for mount in mntpoints:
# Process each mount here. For an example we just print each
print(mount)
Python has a mail processing module called smtplib, and one can find information in the Python docs
getoutput isn't much of an option. getoutput specifically returns everything from stdin and stderr and dumps it to a string (in this case a string called mount). Very convenient if all you want is all the output in one large text string. You can use Subprocess.popen to do the same job with more code.The bash way to do it, just for fun:
awk '{print $2}' /proc/mounts | df -h | mail -s `date +%Y-%m-%d` "[email protected]"
open('/proc/mounts').read() is a start for Python. (Awesome answer btw)