5 obscure but useful Linux commands for sysadmins

How many Linux commands do you know? How many do you need to know? You probably have a dozen or so that you use on a regular basis and a handful more that you use so sporadically that you have to Google them. This article makes it even worse because, in it, I explore five commands you might never have heard of or had a use for, but now you will.

I recently read that Shakespeare had a vocabulary of approximately 66,000 words. I've also read that the average English speaker's vocabulary consists of 10,000 to 20,000 words. And, although I've never heard an estimate, my best guess is that the average Linux system administrator knows 20 to 30 commands, and very few, if any sysadmins, know all of the possible switches and options for any single command. After all, that's what man pages are for, right?

[ You might also like: How to use the Linux mtr command ]

So, in the spirit of expanding your sysadmin vocabulary, here are those five obscure but useful commands that every sysadmin should know in alphabetical order.

nmtui

One of my all-time favorite programs to run was the old Red Hat Linux (text-based user interface) TUI application, setup. It was the best ever all-in-one program to perform configurations on your Linux system. It has been replaced by others such as NetworkManager TUI or nmtui. However, nmtui is specifically for setting up a network connection.

For servers, nmtui is your hero because you don't have the nice graphical NetworkManager GUI to click and fill in the fields with info. For purists, you can edit the /etc/sysconfig/network-scripts/ifcfg-eth0 or whatever your network device is called.

os-prober

The os-prober command is next up and is an awesome find for those of you who inherit systems from other sysadmins—especially laptops and "test" systems. The os-prober tells you if there are other operating systems (OSs) installed on a computer.

From repoquery: This package detects other OSs available on a system and outputs the results in a generic machine-readable format. Support for new OSs and Linux distributions can be added easily. While I don't have a system with multiple OSs installed on it, the following is an example of an os-prober scan where you have a foreign OS installed alongside your Linux one:

/dev/sda1:Windows 10:Windows:chain

shuf

The shuf command is a shortened term for shuffle. Formally, the shuf command generates random permutations from input. You can think of shuf as a dice rolling or a "pulling names out of a hat" program. An example provides a better illustration.

Create a file with numbers from 2 to 12, one number on each line (dice roll possibilities), and save it as dice.txt.

2
3
4
5
6
7
8
9
10
11
12

Then run the following shuf command to extract one random number from the list, as if you are rolling dice:

$ shuf -n 1 dice.txt 
3
$ shuf -n 1 dice.txt 
7
$ shuf -n 1 dice.txt 
5
$ shuf -n 1 dice.txt 
12
$ shuf -n 1 dice.txt 
8
$ shuf -n 1 dice.txt 
7
$ shuf -n 1 dice.txt 
3
$ shuf -n 1 dice.txt 
11

If you write shell scripts, your mind is now permutating all the possibilities for this one. You can use shuf with strings, numbers, or a combination of strings and numbers.

split

The split command would seem to be most useful for programmers. It's used to split a large file into smaller chunks, with 1000 lines being the default chunk size. Because 1000 lines are far too long to show in this article, here's what I did to demonstrate the split command for myself:

This listing is 5568 lines:

$ ls -lR /etc > listing.txt

$ split listing.txt

$ ls

listing.txt  xaa  xab  xac  xad  xae  xaf

If you don't specify a prefix with the command, those are the filenames you get for the 1000 line splits. Here's an example of the split command with a prefix (blah-):

$ split listing.txt blah-

$ ls

blah-aa  blah-ab  blah-ac  blah-ad  blah-ae  blah-af  listing.txt

If you want to limit the results to something other than 1000 lines per file, use the (-l #LINES) option. For example:

$ split -l 500 listing.txt file-

file-aa  file-ab  file-ac  file-ad  file-ae  file-af  file-ag
file-ah  file-ai  file-aj  file-ak  file-al  listing.txt

There are several other options for which you can search the man page, such as changing the suffix (aa, ab, ac, etc.) to a number. Splitting a file is handy because you can hand out chunks to other developers for debugging or for other testing. It's also a great way to examine source code from another project.

watch

For scripting, the watch command is one of my favorites. Using it, you can execute a command or script periodically to perform a task or to collect information. The following is a worthless but illustrative example. Create a new file named, createfile.sh, and make it executable (chmod u+x createfile.sh) with the following contents.

#!/bin/bash

# This script creates an empty file with the current time as the filename.

touch `date +%T`.txt

Issue the watch command to run the createfile.sh script every 10 seconds.

watch -n 10 /home/khess/createfiles.sh

Give the full path to the script so that the watch command can find it. Here, the path is /home/khess/createfile.sh. Allow the command to run for a minute or so and then stop its execution with CTRL-C and list the files it created.

$ ls
20:21:16.txt  20:21:36.txt  20:21:56.txt  20:22:16.txt  20:22:36.txt  20:22:56.txt
20:21:26.txt  20:21:46.txt  20:22:06.txt  20:22:26.txt  20:22:46.txt  createfiles.sh

Have you thought of some uses for the watch command yet?

[ Free cheat sheet: Get a list of Linux utilities and commands for managing servers and networks. ] 

Wrap up

These five obscure commands won't make you a better sysadmin. I won't even promise that they'll make your life easier. What these five commands will do is help you creatively solve some of your nagging problems with pizzazz—especially when your coworkers see your handiwork and say, "Hey, how did you do that?" It's totally worth it. And, if nothing else, you've expanded your sysadmin vocabulary by five. Take that, Shakespeare.