The Linux Upskill Challenge is something I’ve been following on
Reddit for a while and have been wanting to do. At this point, I’m pretty OK with computers, but I’d like to (re)learn
some fundamentals of Linux and Unix systems.

This post contains my notes as I underwent the process.

Day 0: Setup

Link

They recommend provisioning a VPC somewhere in the cloud. I chose AWS – signed up a for a new account on the free tier
and created a t2.micro Ubuntu 20.04 LTS EC2 instance and an RSA key pair for it.

Changed the perms on the private key to chmod 600 and added the following to my ~/.ssh/config:

Host linux-upskill
  Hostname 18.xxx.xxx.xxx
  User ubuntu
  IdentityFile ~/.ssh/linux-upskill

Then running ssh linux-upskill dropped me in to a shell.

The article briefly mentions how AWS provisions a “tiny” VPS:

In a datacentre somewhere, a single physical server running Linux will be split into a dozen or more Virtual servers,
using the KVM (Kernel-based Virtual Machine) feature that’s been part of Linux since early 2007.

Well this is very cool! From this article and other places:

A hypervisor, also known as a virtual machine monitor or VMM, is software that creates and runs virtual machines (VMs)

[In KVM] Every VM runs as a separate Linux process under systemd, with dedicated virtual hardware resources
attached.

The main difference between Type 1 vs. Type 2 hypervisors is that Type 1 (e.g. KVM) runs on bare metal and Type 2
(e.g. Virtualbox) runs on top of an operating system.

Day 1: Get to know your server

Link

Nothing too wild here, but some fun tidbit learnings:

  • prompt with $ is non-root, # is root
  • free and uname -a are nice utilities

The further reading goes in to SSH port forwarding a bit. I’ve seen some pretty cool remote dev setups in the past that
used port forwarding, and was able to get local port forwarding working using nginx:

ssh linux-upskill
sudo apt-get install nginx -y
sudo systemctl start nginx.service
sudo systemctl status nginx.service
exit

ssh -L localhost:8000:localhost:80 linux-upskill
open http://localhost:8000

The http traffic went through my running ssh tunnel and voila, the nginx landing page!

I assume I could do the reverse, running nginx locally and using reverse port forwarding to curl the nginx server from
my remote server through an ssh tunnel running up with ssh -R ....

Another thing I learned:

Ports numbers less than 1024 are privileged ports and can be used only by root

A Recurser pointed out that a list of all packages installed in a given distribution may be shared in a
public .manifest file,
e.g. Ubuntu 22.04 Desktop.

Day 2: Basic navigation

Link

Pretty basic, cd, mkdir, etc. Tidbit learnings:

  • Boolean arguments to cli tools are called “switches”, e.g. the -l in ls -l
  • The -rt switches to ls sort by time, reversed such that oldest modified files show up first
  • You can see the man page of man with man man – and there are actually a lot of options!
  • Prank your coworkers by putting this in a shell script at work: PS1="$PS1"💩. Every invocation of the script will add
    another emoji to their prompt
  • This actually seems pretty handy for shell scripts at work: export PS4='$0.$LINENO+ '

Day 3: Power trip!

Link

Roles and permissions!

  • setting up non-root accounts with permission to run sudo for admins is standard
  • /etc/shadow is where hashed passwords are kept
  • sudo -i for running a series of commands as root
  • /var/log/auth.log logs all uses of sudo (grep "sudo" /var/log/auth.log)
  • hostnamectl to rename the server
  • /etc/cloud/cloud.cfg has a preserve_hostname across reboots
  • timedatectl for time, timezone, and date controls
  • john the ripper and hashcat are tools that can be used to crack passwords once you have /etc/passwd
    and /etc/shadow

From man sudo:

-i, –login: Run the shell specified by the target user’s password database entry as a login shell…The command is
run with an environment similar to the one a user would receive at log in. Note that most shells behave differently
when a command is specified as compared to an interactive session.

From man su:

su allows to run commands with a substitute user and group ID.

When called without arguments, su defaults to running an interactive shell as root.

Interestingly, the prompt is different for sudo -i and sudo su:

ubuntu@ip-172-31-24-204:~$ sudo -i
root@ip-172-31-24-204:~# logout

ubuntu@ip-172-31-24-204:~$ sudo su
root@ip-172-31-24-204:/home/ubuntu# exit

After some discussion with some Recursers we figured out that sudo -i and sudo su -
both cd to the home directory, whereas sudo su without the dash leaves you in the prior home
directory (/home/ubuntu).

Day 4: Installing software, exploring the file structure

Link

Learned a good amount about the linux directory structure – man hier is gold!

  • apt-get is a full-featured but simplified interface to dpkg, and apt is a more user-friendly but slightly
    stripped-back version of apt-get
  • apt search to find installable packages
  • man hier to see a man page on the directory hierarchy
  • /sbin contains commands needed to boot the system (“system binaries”)
  • /etc contains static configuration files
  • /var contains files that may change in size, e.g. log files
  • /etc/apt/sources.list contains the repository locations, usually mirror servers closer to my server than the main
    servers are

Notably, these directory meanings shouldn’t be treated as gospel. It looks like sbin and bin are symlinks to
their /usr equivalents, e.g.

ubuntu@ip-172-31-24-204:~$ ls -la /sbin
lrwxrwxrwx 1 root root 8 Dec  1 21:32 /sbin -> usr/sbin

Further note on apt vs dpkg (source):

With APT, you can retrieve a file from a remote repository and install it, all in one command. This saves you from the
work of manually finding and downloading the package before installation.

With dpkg, you can only install local files you’ve already downloaded yourself. It can’t search remote repositories or
pull packages from them.

Day 5: More or less…

Link

Mishmash of cli utilities, dotfiles, etc.

  • more is a worse version of less and I can probably always stick with less
  • history and running repeating commands with ! is very nice, although I’ll likely stick
    to Ctrl + r
  • nano is sweet, but I’ll prefer vi or vim when it’s installed

Day 6: Editing with “vim”

Link

I use Vim keybindings in most places I edit text already, but there were some new tidbits for me here:

  • Learned about the existence of pico, joe, and jed text editors
  • vi --version to see if vi is actually vim on your machine
  • cp -v for verbose copy
  • dd deletes lines, whereas d{motion} deletes text that {motion} moves over
  • vi is required by the Single Unix Specification and POSIX
  • vi uses hjkl for motion because of the ADM-3A terminal keyboard
  • neovim has it’s own :Tutor command, whereas vimtutor is also always available for vanilla vim introduction

A Recurser friend noted that another way to check the vi version/binary that is installed is using readlink
and which:

1
readlink -f $(which vi)

Day 7: The server and its services

Link

Install and run Apache2 web server, similar to my nginx experiments on Day 1.

  • since I allow all http traffic to my server on port 80, http://18.236.102.243/ is available
    to the public (might not work anymore if you are reading this in the future)
  • a virtual host is a contained site or system on a web server
  • I can edit the default page with sudo vim /var/www/html/index.html
  • systemctl manages systemd services
  • systemctl list-units [--all --state --type] shows units, i.e. resources that systemd knows how to manage

Notably, Apache may change its name soon.

Since the web server is public, I got this fun entry in the /var/log/apache2/access.log the following day:

198.235.24.13 - - [10/Jan/2023:21:12:16 +0000]
"GET / HTTP/1.0" 200 11155 "-" "Expanse, a Palo Alto Networks company,
searches across the global IPv4 space multiple times per day to identify
customers' presences on the Internet.

Day 8: The infamous “grep” and other text processors

Link

A good day full of useful tools. Some new things I learned:

  • ssh runs it’s own service, sshd (Secure Shell Daemon) sudo systemctl status sshd
  • /var/log/auth.log contains login and sudo usage information
  • tail -f to follow

Use the cut command to select out most interesting portions of each line by specifying “-d” for delimiter and “-f”
for field – like: grep "authenticating" /var/log/auth.log | grep "root" | cut -f 10- -d" "
(field 10 onwards, where the delimiter between field is the ” ” character).

  • use grep -v XXX to invert the output (“all lines that do not contain XXX”)
  • it would be nice to get better at awk and sed one day

Day 9: Diving into networking

Link

Ok, I learned a LOT of stuff today!

ports and network devices

a port is a number assigned to uniquely identify a connection endpoint and to direct data to a specific service

  • ss, “socket status”, is one tool to examine open ports, replacing netstat
  • nmap is another, non-default tool for port examination
ubuntu@ip-172-31-24-204:~$ sudo ss -ltpn
State      Recv-Q     Send-Q         Local Address:Port           Peer Address:Port     Process
LISTEN     0          4096           127.0.0.53%lo:53                  0.0.0.0:*         users:(("systemd-resolve",pid=417,fd=13))
LISTEN     0          128                  0.0.0.0:22                  0.0.0.0:*         users:(("sshd",pid=653,fd=3))
LISTEN     0          511                        *:80                        *:*         users:(("apache2",pid=29833,fd=4),("apache2",pid=29832,fd=4),("apache2",pid=26021,fd=4))
LISTEN     0          128                     [::]:22                     [::]:*         users:(("sshd",pid=653,fd=4))

After sudo apt install nmap:

ubuntu@ip-172-31-24-204:~$ nmap localhost
Starting Nmap 7.80 ( https://nmap.org ) at 2023-01-12 19:40 UTC
Nmap scan report for localhost (127.0.0.1)
Host is up (0.00011s latency).
Not shown: 998 closed ports
PORT   STATE SERVICE
22/tcp open  ssh
80/tcp open  http

Nmap done: 1 IP address (1 host up) scanned in 0.07 seconds
  • nmap localhost may show services that are only bound to localhost (the loopback network device), so can first
    get actual network card IP address:
ubuntu@ip-172-31-24-204:~$ ip address
1: lo:  mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
       valid_lft forever preferred_lft forever
    inet6 ::1/128 scope host
       valid_lft forever preferred_lft forever
2: eth0:  mtu 9001 qdisc fq_codel state UP group default qlen 1000
    link/ether 02:32:a6:75:13:17 brd ff:ff:ff:ff:ff:ff
    inet 172.31.24.204/20 brd 172.31.31.255 scope global dynamic eth0
       valid_lft 3099sec preferred_lft 3099sec
    inet6 fe80::32:a6ff:fe75:1317/64 scope link
       valid_lft forever preferred_lft forever

ubuntu@ip-172-31-24-204:~$ nmap 172.31.24.204
Starting Nmap 7.80 ( https://nmap.org ) at 2023-01-12 19:49 UTC
Nmap scan report for ip-172-31-24-204.us-west-2.compute.internal (172.31.24.204)
Host is up (0.00011s latency).
Not shown: 998 closed ports
PORT   STATE SERVICE
22/tcp open  ssh
80/tcp open  http

Nmap done: 1 IP address (1 host up) scanned in 0.07 seconds
  • so in this case, the exposed services are the same ones bound to localhost, but this may not always be true

host firewall

  • a firewall is a network security system that monitors and controls incoming and outgoing network traffic based on
    predetermined security rules
  • the linux kernel has built-in firewall functionality called netfilter
  • iptables and nftables are utilities for configuring netfilter
  • ufw (uncomplicated firewall) is a less complicated tool
  • checking what rules are in place:
ubuntu@ip-172-31-24-204:~$ sudo iptables -L
Chain INPUT (policy ACCEPT)
target     prot opt source               destination

Chain FORWARD (policy ACCEPT)
target     prot opt source               destination

Chain OUTPUT (policy ACCEPT)
target     prot opt source               destination
  • this^ is equivalent to “no firewall”
  • after sudo apt install ufw, can deny http traffic:
ubuntu@ip-172-31-24-204:~$ sudo ufw allow ssh
Rules updated
Rules updated (v6)

ubuntu@ip-172-31-24-204:~$ sudo ufw deny http
Rules updated
Rules updated (v6)

ubuntu@ip-172-31-24-204:~$ sudo ufw enable
Command may disrupt existing ssh connections. Proceed with operation (y|n)? y
Firewall is active and enabled on system startup
  • sudo iptables -L now has MANY rules, including
Chain ufw-user-input (1 references)
target     prot opt source               destination
ACCEPT     tcp  --  anywhere             anywhere             tcp dpt:ssh
DROP       tcp  --  anywhere             anywhere             tcp dpt:http
sudo ufw allow http
sudo ufw enable
  • in general, not running unnecessary services is a good enough practice for protection depending on the server
  • changing the standard port, e.g. the ssh standard port 22 by editing /etc/ssh/sshd_config, is a fine “security by
    obscurity” (or not, because
    you’re not actually hiding the mechanism) thing to do

Day 10: Getting the computer to do your work for you

Link

  • crontab -l shows user crontab entry, add sudo for root user
  • system crontab at /etc/crontab
ubuntu@ip-172-31-24-204:~$ cat /etc/crontab
# /etc/crontab: system-wide crontab
# Unlike any other crontab you don't have to run the `crontab'
# command to install the new version when you edit this file
# and files in /etc/cron.d. These files also have username fields,
# that none of the other crontabs do.

SHELL=/bin/sh
PATH=/usr/local/sbin:/usr/local/bin:/sbin:/bin:/usr/sbin:/usr/bin

# Example of job definition:
# .---------------- minute (0 - 59)
# |  .------------- hour (0 - 23)
# |  |  .---------- day of month (1 - 31)
# |  |  |  .------- month (1 - 12) OR jan,feb,mar,apr ...
# |  |  |  |  .---- day of week (0 - 6) (Sunday=0 or 7) OR sun,mon,tue,wed,thu,fri,sat
# |  |  |  |  |
# *  *  *  *  * user-name command to be executed
17 *    * * *   root    cd / && run-parts --report /etc/cron.hourly
25 6    * * *   root    test -x /usr/sbin/anacron || ( cd / && run-parts --report /etc/cron.daily )
47 6    * * 7   root    test -x /usr/sbin/anacron || ( cd / && run-parts --report /etc/cron.weekly )
52 6    1 * *   root    test -x /usr/sbin/anacron || ( cd / && run-parts --report /etc/cron.monthly )
#
  • there are hourly, daily, weekly, and monthly scripts run in /etc/cron.hourly, /etc/cron.daily, etc.
  • run-parts runs scripts or programs in a directory in alphabetical order
  • logrotate rotates, compresses, and mails system logs
  • systemd can be used to run specific tasks at times using timers:
ubuntu@ip-172-31-24-204:~$ systemctl list-timers
NEXT                        LEFT          LAST                        PASSED       UNIT                         ACTIVATES
Sat 2023-01-14 00:00:00 UTC 3h 7min left  Fri 2023-01-13 00:01:30 UTC 20h ago      logrotate.timer              logrotate.service
Sat 2023-01-14 00:00:00 UTC 3h 7min left  Fri 2023-01-13 00:01:30 UTC 20h ago      man-db.timer                 man-db.service
Sat 2023-01-14 01:00:13 UTC 4h 7min left  Fri 2023-01-13 10:59:30 UTC 9h ago       fwupd-refresh.timer          fwupd-refresh.service
...
  • anacron, or “anachronistic cron”, is good for e.g. laptops that are off some of the time

Day 11: Finding things…

Link

How to find files and text in files efficiently.

  • locate (sudo apt install mlocate) is a very nifty version of find / -name 2>/dev/null
  • may want to run sudo updatedb to update the index locate searches (usually runs daily through cron)
  • find /home -mtime -3 finds all the files modified in the last 3 days
  • recursively search text with grep -R -i "PermitRootLogin" /etc/* 2>/dev/null
  • zless and zgrep for viewing/searching compressed (e.g. gzip’d) files
  • find flags and options:
    • -iname to ignore case for name
    • -o for “or”
    • -size for size filtering, e.g. -size +100M
    • -atime for access time, e.g. -atime 30 accessed in last 30 days
    • -ok like -exec, but ask for confirmation, e.g. find . -type f -ok cat {} ;

Day 12: Transferring files

Link

File sharing protocols other than ftp and scp:

  • SMB for Windows
  • AFP for local network of MacOS
  • WebDAV for http
  • rsync
  • SFTP over SSH – best for our purposes as we’re already ssh’ing
    • There is an sftp utility pre-installed on MacOS

Cyberduck is a nice cloud storage browser with SFTP support, providing a GUI for
uploading/downloading files to/from remote servers. The logo and app icon alone make it well worth it.

A Recurser pointed out magic-wormhole, which is another
very cool program for transferring data securely.

Day 13: Who has permission?

Link

I’ve seen most of these before, but nice to do an actual overview of the tools where I grok what’s going on a bit
better.

  • all files are tagged with the user and group that owns them
  • -rwxr-xr-x are 3 groups of rwx (read/write/execute) for user, group, others respectively
    • first char is a file type indicator (- for regular file)
  • chmod to change perms
    • chmod u-w tuesday.txt to remove write perms from the user
    • chmod g+w tuesday.txt to add write perms to the group
    • chmod o-r tuesday.txt to remove read perms from others
    • chmod u+rwx,go-rwx hello to do multiple at once
  • most linux systems create a group for each user
  • groups to see groups you are a member of
ubuntu@ip-172-31-24-204:~$ groups
ubuntu adm dialout cdrom floppy sudo audio dip video plugdev netdev lxd
  • sudo usermod -a -G group user to add user to group
  • sudo adduser fred to create a user fred
  • sudo su fred to “become” fred
ubuntu@ip-172-31-24-204:~$ sudo su fred
fred@ip-172-31-24-204:/home/ubuntu$ groups
fred
  • sudo usermod -a -G sudo fred to add fred to sudo group
  • umask determines the permissions new files are given
    • directories get 777 – umask, files get 666 – umask
    • example: umask = 0002, dirs get 775, files get 664
  • octal mode of file permissions:
    • r = 2^2 = 4
    • w = 2^1 = 2
    • x = 2^0 = 1
    • so 1 = --x, 7 = rwx, 5 = r-x etc.
    • chmod 664 myfile is rw- for user and group, r-- for others
  • id command to view user/group/etc ids:
ubuntu@ip-172-31-24-204:~$ id
uid=1000(ubuntu) gid=1000(ubuntu) groups=1000(ubuntu),4(adm),20(dialout),24(cdrom),25(floppy),27(sudo),29(audio),30(dip),44(video),46(plugdev),118(netdev),119(lxd)

Link

  • sudo adduser helen to make new user
  • sudo EDITOR=vim visudo to edit /etc/sudoers file, added:
# Allow user "helen" to run "sudo reboot"
# ...and don't prompt for a password
#
helen ALL = NOPASSWD:/sbin/reboot
  • note that we could have run sudo usermod -a -G sudo helen to give helen ability to do ANYTHING with sudo, not just
    reboot
  • remove with sudo gpasswd -d helen sudo
  • /etc/passwd and /etc/group have user and group info respectively
  • created new key pair in AWS using EC2
  • downloaded the private key
  • mv ~/Downloads/helen-linux-upskill.cer ~/.ssh/helen-linux-upskill
  • chmod 400 ~/.ssh/helen-linux-upskill to correct perms
  • ssh-keygen -y -f ~/.ssh/helen-linux-upskill to get public key
  • Then, ssh as ubuntu and ran sudo su helen to become helen
  • mkdir ~/.ssh
  • vim ~/.ssh/authorized_keys and added public key there
  • chmod 600 ~/.ssh/authorized_keys
  • then added the following to my local ~/.ssh/config
* Host helen-linux-upskill
  Hostname 18.236.102.243
  User helen
  IdentityFile ~/.ssh/helen-linux-upskill
  • then can ssh helen-linux-upskill and be helen 🙂

Link

  • linux versions come from release year (e.g. 18.04 = 2018)
  • /etc/apt/sources.list shows urls to repositories that apt references to install things from
  • count installable packages with apt-cache dump | grep "Package:" | wc -l
ubuntu@ip-172-31-24-204:~$ apt-cache dump | grep "Package:" | wc -l
106898
  • to add multiverse repository, uncommented it in sources.list and ran sudo apt update
  • then could install packages from it, e.g. netperf
ubuntu@ip-172-31-24-204:~$ sudo apt install netperf
...
Get:1 http://us-west-2.ec2.archive.ubuntu.com/ubuntu focal/multiverse amd64 netperf amd64 2.6.0-2.1 [550 kB]
...
  • can also add via cli e.g. sudo add-apt-repository ppa:dawidd0811/neofetch && sudo apt update (this is no longer
    valid)

Link

  • create uncompressed tarball in current dir tar -cvf myinits.tar /etc/init.d/
  • tar was short for “tape archive”
  • compress using GnuZip gzip myinits.tar to create myinits.tar.gz
  • all in one step with tar -cvzf myinits.tgz /etc/init.d/
    • -c: create archive
    • -v: verbose
    • -z: compress
    • -f: output file
  • untar with tar -xvf myinits.tgz
  • bzip2 is another compression algo similar to gzip

Link

  • sudo apt install build-essential to install package of compiler tools
  • wget -v https://nmap.org/dist/nmap-7.93.tar.bz2 to get latest source of nmap
  • tar -jxvf nmap-7.93.tar.bz2 to uncompress (-j for bzip2)
  • cat nmap-7.93/INSTALL shows installation instructions
  • after following install directions:
ubuntu@ip-172-31-24-204:~/nmap-7.93$ sudo updatedb
ubuntu@ip-172-31-24-204:~/nmap-7.93$ locate bin/nmap
/usr/bin/nmap
/usr/local/bin/nmap

Link

  • logrotate app for specifying log rotation behavior
  • ls /var/log already shows rotated logs e.g. syslog.2.gz
  • this is due to /etc/cron.daily/logrotate
  • config file for logrotate at /etc/logrotate.conf and subfiles under /etc/logrotate.d

Link

  • the Linux Virtual Filesystem (VFS) sits above other filesystems like ext3, ext4, etc.
  • layer between filename and data on disk = inode
  • ls -i to show inodes
  • can also use stat command to show inode info
  • filenames point to inodes which point to data on disk
  • permissions, ownership, and dates of a file are at the inode, not filename level
  • several filenames can point to the same inode – a “hard link”
    • ln creates a hard link
    • rm‘ing one of the filenames will not remove the underlying inode, data, or other filename
  • more common than hard links are soft links, or symlinks
    • ln -s creates a symlink. First arg is existing file or directory, second (optional) arg is symlink location
    • symlinks:
      • can link to directories (hard links can’t)
      • can reference a file/directory on a different hard disk or volume (hard links can’t)
      • remain if the original is deleted (hard links don’t)
      • will not reference the file anymore if it is moved (hard links will)
      • have their own inodes/physical disk location (hard links reference the same inode)

Link

Nothing really new here for me as I have good familiarity with bash scripting already.

This was a good course, but more hands on and simplified than I was hoping for. I would still recommend it, particularly
early in your career. Interestingly, the bulk of the most interesting content came in the first 15 days or so.

I still have some knowledge gaps, including questions like:

  • What does “everything is a file” mean?
  • What is a kernel, actually?
  • What is userspace vs kernelspace?
  • What is the “GNU” part of Linux?

Time to read and google onwards :).

Read More