"From one bloody orange!" by "Terry Madeley" on Flickr

Making Vagrant install the latest version of Ansible using Pip and run it as root in Ubuntu Virtual Machines

As previously mentioned, I use Ansible a lot inside Virtual machines orchestrated with Vagrant. Today’s brief tip is how to make Vagrant install the absolutely latest version of Ansible on Ubuntu boxes with Pip.

Here’s your Vagrantfile

Vagrant.configure("2") do |config|
  config.vm.box = "ubuntu/focal64"
  config.vm.provision "ansible_local", run: "always" do |ansible|
    ansible.playbook         = "setup.yml"
    ansible.playbook_command = "sudo ansible-playbook"
    ansible.install_mode     = "pip"
    ansible.pip_install_cmd  = "(until sudo apt update ; do sleep 1 ; done && sudo apt install -y python3-pip && sudo rm -f /usr/bin/pip && sudo ln -s /usr/bin/pip3 /usr/bin/pip && sudo -H pip install --upgrade pip) 2>&1 | tee -a /var/log/vagrant-init"

“But, that pip_install_cmd block is huge”, I hear you cry!

Well, yes, but let’s split that out into a slightly more readable code block! (Yes, I’ve removed the “&&” for clarity sake – it just means “only execute the next command if this one worked”)

  # Wait until we get the apt "package lock" released
  until sudo apt update
    # By sleeping for 1 second increments until it works
    sleep 1

  # Then install python3-pip
  sudo apt install -y python3-pip

  # Just in case python2-pip is installed, delete it
  sudo rm -f /usr/bin/pip

  # And symbolically link pip3 to pip
  sudo ln -s /usr/bin/pip3 /usr/bin/pip

  # And then do a pip self-upgrade
  sudo -H pip install --upgrade pip

# And output this to the end of the file /var/log/vagrant-init, including any error messages
) 2>&1 | tee -a /var/log/vagrant-init

What does this actually do? Well, pip is the python package manager, so we’re asking for the latest packaged version to be installed (it often isn’t particularly with older releases of, well, frankly any Linux distribution) – this is the “pip_install_cmd” block. Then, once pip is installed, it’ll run “pip install ansible” – which will give it the latest version available to Pip, and then when that’s all done, it’ll run “sudo ansible-playbook /vagrant/setup.yml”

Featured image is “From one bloody orange!” by “Terry Madeley” on Flickr and is released under a CC-BY license.

"DeBugged!" by "Randy Heinitz" on Flickr

Debugging Bash Scripts

Yesterday I was struggling a bit with a bash script I was writing. I needed to stop it from running flat out through every loop, and I wanted to see what certain values were at key points in the script.

Yes, I know I could use “read” to pause the script and “echo” to print values, but that leaves a lot of mess that I need to clean up afterwards… so I went looking for something else I could try.

You can have extensive debug statements, which are enabled with a --debug flag or environment variable… but again, messy.

You can run bash -x ./myscript.sh – and, indeed, I do frequently do that… but that shows you the commands which were run at each point, not what the outcome is of each of those commands.

If my problem had been a syntax one, I could have installed shellcheck, which is basically a linter for Bash and other shell scripting languages, but no, I needed more detail about what was happening during the processing.

Instead, I wanted something like xdebug (from PHP)… and I found Bash Debug for VSCode. This doesn’t even need you to install any scripts or services on the target machine – it’s interactive, and has a “watch” section, where you either highlight and right-click a variable expression (like $somevar or ${somevar}) to see when it changes. You can see where in the “callstack” you are and see what values are registered by that script.

Shellcheck shows me problems in my code…
But Bash Debug helps me to find out what values are at specific points in the code.

All in all, a worthy addition to my toolbelt!

Featured image is “DeBugged!” by “Randy Heinitz” on Flickr and is released under a CC-BY license.

"Honey pots" by "Nicholas" on Flickr

Adding MITM (or “Trusted Certificate Authorities”) proxy certificates for Linux and Linux-like Environments

In some work environments, you may find that a “Man In The Middle” (also known as MITM) proxy may have been configured to inspect HTTPS traffic. If you work in a predominantly Windows based environment, you may have had some TLS certificates deployed to your computer when you logged in, or by group policy.

I’ve previously mentioned that if you’re using Firefox on your work machines where you’ve had these certificates pushed to your machine, then you’ll need to enable a configuration flag to make those work under Firefox (“security.enterprise_roots.enabled“), but this is talking about Linux (like Ubuntu, Fedora, CentOS, etc.) and Linux-like environments (like WSL, MSYS2)

Late edit 2021-05-06: Following a conversation with SiDoyle, I added some notes at the end of the post about using the System CA path with the Python Requests library. These notes were initially based on a post by Mohclips from several years ago!

Start with Windows

From your web browser of choice, visit any HTTPS web page that you know will be inspected by your proxy.

If you’re using Mozilla Firefox

In Firefox, click on this part of the address bar and click on the right arrow next to “Connection secure”:

Clicking on the Padlock and then clicking on the Right arrow will take you to the “Connection Security” screen.
Certification Root obscured, but this where we prove we have a MITM certificate.

Click on “More Information” to take you to the “Page info” screen

More obscured details, but click on “View Certificate”

In recent versions of Firefox, clicking on “View Certificate” takes you to a new page which looks like this:

Mammoth amounts of obscuring here! The chain runs from left to right, with the right-most blob being the Root Certificate

Click on the right-most tab of this screen, and navigate down to where it says “Miscellaneous”. Click on the link to download the “PEM (cert)”.

The details on the Certificate Authority (highly obscured!), but here is where we get our “Root” Certificate for this proxy.

Save this certificate somewhere sensible, we’ll need it in a bit!

Note that if you’ve got multiple proxies (perhaps for different network paths, or perhaps for a cloud proxy and an on-premises proxy) you might need to force yourself in into several situations to get these.

If you’re using Google Chrome / Microsoft Edge

In Chrome or Edge, click on the same area, and select “Certificate”:

This will take you to a screen listing the “Certification Path”. This is the chain of trust between the “Root” certificate for the proxy to the certificate they issue so I can visit my website:

This screen shows the chain of trust from the top of the chain (the “Root” certificate) to the bottom (the certificate they issued so I could visit this website)

Click on the topmost line of the list, and then click “View Certificate” to see the root certificate. Click on “Details”:

The (obscured) details for the root CA.

Click on “Copy to File” to open the “Certificate Export Wizard”:

In the Certificate Export Wizard, click “Next”
Select “Base-64 encoded X.509 (.CER)” and click “Next”
Click on the “Browse…” button to select a path.
Name the file something sensible, and put the file somewhere you’ll find it shortly. Click “Save”, then click “Next”.

Once you’ve saved this file, rename it to have the extension .pem. You may need to do this from a command line!

Copy the certificate into the environment and add it to the system keychain

Ubuntu or Debian based systems as an OS, or as a WSL environment

As root, copy the proxy’s root key into /usr/local/share/ca-certificates/<your_proxy_name>.crt (for example, /usr/local/share/ca-certificates/proxy.my.corp.crt) and then run update-ca-certificates to update the system-wide certificate store.

RHEL/CentOS as an OS, or as a WSL environment

As root, copy the proxy’s root key into /etc/pki/ca-trust/source/anchors/<your_proxy_name>.pem (for example, /etc/pki/ca-trust/source/anchors/proxy.my.corp.pem) and then run update-ca-trust to update the system-wide certificate store.

MSYS2 or the Ruby Installer

Open the path to your MSYS2 environment (e.g. C:\Ruby30-x64\msys64) using your file manager (Explorer) and run msys2.exe. Then paste the proxy’s root key into the etc/pki/ca-trust/source/anchors subdirectory, naming it <your_proxy_name>.pem. In the MSYS2 window, run update-ca-trust to update the environment-wide certificate store.

If you’ve obtained the Ruby Installer from https://rubyinstaller.org/ and installed it from there, assuming you accepted the default path of C:\Ruby<VERSION>-x64 (e.g. C:\Ruby30-x64) you need to perform the above step (running update-ca-trust) and then copy the file from C:\Ruby30-x64\mysys64\etc\pki\ca-trust\extracted\pem\tls-ca-bundle.pem to C:\Ruby30-x64\ssl\cert.pem

Using the keychain

Most of your Linux and Linux-Like environments will operate fine with this keychain, but for some reason, Python needs an environment variable to be passed to it for this. As I encounter more environments, I’ll update this post!

The path to the system keychain varies between releases, but under Debian based systems, it is: /etc/ssl/certs/ca-certificates.crt while under RedHat based systems, it is: /etc/pki/tls/certs/ca-bundle.crt.

Python “Requests” library

If you’re getting TLS errors in your Python applications, you need the REQUESTS_CA_BUNDLE environment variable set to the path for the system-wide keychain. You may want to add this line to your /etc/profile to include this path.


Featured image is “Honey pots” by “Nicholas” on Flickr and is released under a CC-BY license.

My Fujitsu Stylistic V727

Review of my Fujitsu Stylistic V727 Laptop/Tablet.

TL;DR: Linux is usually awesome, but it doesn’t work for my niche case.

Why was I in the market for a new computer?

October 2019 my beloved (but 7 year old) Acer V5-171, “Minilith” (so named because it was smaller than it’s predecessor, a 17″ monster of a black slab that was named “Monolith”) started exhibiting signs of having a dead battery. I replaced the battery with an 3rd party replacement, and while it charged OK for a few runs, it stopped charging all together (I could get a maximum of 5% charge), so I put the old battery in, and it started working better. Huzzah. All was going well until around 6 months ago when the hard drive failed, so I replaced it with an SSD, and that gave it a new lease of life… and this month, well, it just wouldn’t boot reliably. I finally decided that it was time to let it go and play with Timmy the dog at the farm, and replace it with something newer.

The back of "Minilith", my 7 year old Laptop.
The back of “Minilith”, my 7 year old Laptop.
Minilith's Keyboard and Screen
Minilith’s Keyboard and Screen

Fortunately, this co-incided with a small win on the company social Prize Draw of a reasonable sized pay out, enough to consider looking at the Ex-Demo staff sales list made available to me by dint of my employer.

Making my choice.

There weren’t a lot of options, to be fair, but one item stood out to me. A Stylistic Tablet Computer. I’d previously had an Asus Transformer TF300T – a tablet-come-computer which had a detachable keyboard. I’d loved that, even though it didn’t really do what I wanted it for (and, I think I’d paid quite a bit over what it was worth, really)… but what I really wanted to do was have a tablet I could use for computing… Hence, the Stylistic.

Image of the TF300T, a tablet model I’d previously owned, from the Asus Transformer Marketing Pages
Der obere Teil des Stylistic V727 ist ein Tablet.
Image of the tablet view of the Stylistic V727 computer from a German blog.

Fujitsu are in a bit of an odd place, at least in the UK (I’ve not looked elsewhere) for personal computers – we sell quite well (apparently) to business, but we moved out of the “selling to the public” market probably around 2010, and so it was pretty hard to gauge how well this laptop performed. Oh, and of course, being a “Linux Enthusiast”, I wanted to be able to run Ubuntu, Fedora or others on it.

Because it was an internal sale, and I wanted to test Ubuntu on it before I bought it, I was able to get the sales team to let me evaluate it before I bought it.

It arrives!

It arrived as the tablet and keyboard, with a dock for setting it on your desk. I tested it with Windows, where the dock worked well, but the keyboard by itself didn’t so much. You see, the keyboard is an optional accessory, and had been sold with the laptop, all good thus far. Except what you also need to get, when you get the keyboard, is the case. The case gives you the sturdy back to give the “laptop” a frame. It’s basically the hinge that the top-heavy screen needs to keep itself upright.

A screen capture of the Fujitsu Stylistic V727 from the datasheet. Note this image shows the optional keyboard and the optional case.

The sales team were very understanding, and found a case to ship to me as well, but it wouldn’t come for a few days, so I was left to try out the rest of the hardware.

What do you get for your money?

The processor is an Intel i5-7Y57 dual core CPU with four threads, running at 1.2GHz.

It has 8GB of RAM and a 256GB m2 SATA drive.

The 12.3″ touch-or-pen (included) enabled screen has a maximum resolution of 1920×1280 pixels. The surface of the tablet is considered a WACOM tablet, and the pen can be sensed from a reasonable distance away. There are two buttons on the side of the pen, which turns a tap on the screen from a “normal” left click to a right click, or a center click.

On the rear is a fingerprint sensor.

From a network perspective, the WIFI supports 802.11ac, Bluetooth 4.2, and under the battery there is a LTE module onboard (although, I’ve not tested that).

On the side is a USB3.1 A connector, and a USB C connector (the specs sheet I linked to above suggests there is a single USB 3 and a USB 2 interface, but I doubt the USB C is USB 2).

There is also a MicroSD slot, which is detected by the booted OS as an MMC device, but it is not detected as a bootable device.

There’s a combined 3.5mm audio in and out jack, which I’ve not tried and a power socket.

There are two cameras, a 5 megapixel front-facing camera and an 8 megapixel rear facing camera with a flash.

The detachable keyboard has an integrated touchpad. It’s all good, and compared to my poor Acer V5, it’s a massive step up ❤

When you add in the desk dock (where, to be fair, it’s spent most of it’s time since I got it), the connections also then include Gigabit Ethernet, a Display Port interface, a VGA port and three USB 3 A interfaces, and a power socket.

The OS Comparison starts

Windows first

I booted it in Windows, and found it really rather responsive, especially once I’d reinstalled Windows without all the customizations the demo team had put on…

My previous install of Windows on Minilith had been the Home edition, and I’d found the semi-constant nagging to install games and the like rather annoying. I’ve had a couple of Windows 10 Professional installs at work, and, while those builds came with their own fair share of corporately mandated bloat (after all, their threat models are somewhat different to mine) they usually felt more slim than the Windows 10 home install I’d had, so when I saw this had Windows 10 Professional, I was looking forward to seeing something a bit leaner… and I wasn’t let down. All the hardware worked fine, I had the fingerprint reader working, no worries and the dock was great.

Docking and undocking is relatively seamless, although the first try was a bit tricky, I’ve got used to it. I had two screens attached via my work-supplied Fujitsu PR08 DisplayLink adaptor, plugged through the dock, and again, that all worked fine.

I could use the pen in the tablet mode really well. It makes selecting items on the screen easy, and if you don’t want to use the virtual keyboard, in some cases, it pops up a handwriting recognition box, although the time I showed this to my wife (where I’d been using it successfully for some time), it didn’t recognise half the words I wrote… but I’m sure that’s just my dreadful scrawl, and not the tablet’s fault!

Even using the tablet without the pen worked really well. Tapping the screen is a left click, and a long press on an area is a right click, similar to how Android handles left-and-right clicks in RDP and VNC sessions. The keyboard has several “modes” – a reduced character set, a thumb typing set or a full keyboard. The reduced set has a control key and an escape key, but no alt, windows or arrow keys. I didn’t try the thumb typing set (this thing is 12″ across!) but the full keyboard is an “ISO layout” 75% keyboard (I discovered by matching the image to this website!) which means I still get my Control, Alt, and Windows keys.

Next, Ubuntu

I booted from a USB stick that had the Ubuntu 20.04 installer on it. Ubuntu booted fine, allowed me to repartition the Windows partition into approximately half the drive, and install away. During the install, I was asked to provide a password to setup the SecureBoot keys, and instructed that it’d prompt me for it on the next reboot. Most of the hardware worked fine. Dock, keyboard, Wifi, Bluetooth… all good. The fingerprint sensor wasn’t detected, and still isn’t, but I’m OK with that, it was always just a nice-to-have. The install worked fine, and yes, on reboot, I got a blue screen asking me to set up my “MOK” (which, I guessed eventually, was the SecureBoot setup). I realised that the SecureBoot install stage of the Linux install copies a private key to the UEFI space, and on the next boot, it spots there’s a key there and asks you to unlock that private key, so it can install it into the boot keys. All good!

I was working away on it with the tablet in the dock. I tried using it with the detachable keyboard, but it was a bit tricky to use without the rigid back, so I kept it in the dock. The pen works a treat too.

The problem came when I tried to use it as a tablet.

You see, where Windows has a selection of keyboard layouts for their “On Screen Keyboard” (OSK), the Gnome one only lets you use this layout:

Screenshot taken from an article at OMG Ubuntu.

While this is passable for tapping stuff into a URL bar in your browser, entering a password for logging in, or typing simple statements into dialogue boxes, there are some key things it’s missing. The first (for me) is a Control, Alt or Super (Windows) key. This means I can’t do any programming, of any sort, in Tablet mode. Note, this just works on Windows, and is possible on Android with an extended keyboard called “Hacker Keyboard”. There are also no cursor keys, which seems like it’s less important, but it makes editing things you’ve typed (or mistyped) MUCH harder.

“Well, OK then, let’s have a look around and see what our options are?”

I’d heard good things about “OnBoard”, a predecessor to Gnome OSK, but because OSK is registered as “The” on screen keyboard, and runs as a system process, and OnBoard is a user process, Gnome OSK pops up any time you want to do on screen keyboard things, even if you’ve got OnBoard loaded. Ahah! I found an extension which blocks Gnome OSK… except that stops it from being able to be used for logging in.

You see, that whole “system” versus “user” process thing I mentioned before. The Gnome lock screen is considered a system process, not a user one, which means that if you’ve disabled Gnome OSK, then you can’t put your password in, but equally, if you’re typing in a box with OnBoard, change focus and change back again, up pops Gnome OSK.

Breaking down and turning it around.

I should confess, I didn’t spend a lot of time wondering about this. I booted a Kubuntu environment instead, and found that this really didn’t work for me either (although I now don’t remember what stopped me from liking it – I might have to revisit this!) By this point, I’d spent several hours “messing” around with this, and I just wanted to give something a try. So I booted back into Windows.

I gave the on screen keyboard another try. It worked great. I tried doing some sketches in Paint 3D (the replacement for Windows Paint) with the pen, and it was very easy (so much so, I need to work out how to use it for my next design call with work!)

All the familiar tools I use in my work or personal environment are there.

  • VSCode. Check.
  • A usable shell (via Windows Subsystem for Linux). Check.
  • File synchronization (via Syncthing). Check.
  • Web browser (Firefox). Check.
  • Audio recording software (Audacity). Check.
  • Image editor (GIMP). Check.
  • Voice chat for the podcast software (Mumble). Check.
  • Screencasting software (OBS). Check.
  • Virtual Machine software (VirtualBox, Vagrant, Terraform). Check.

And the fingerprint reader works… so I stuck with Windows 10.

The only last catch, whether it was Windows or Linux? There’s no HDMI or VGA out without the dock… so I need to start looking into “cheap” display adaptors that I can use for presenting things, whenever we get back to “normal” and I can start attending and speaking at conferences again.

What about the case?

Oh yehr, so a few days after I get the computer, the case turns up. It attaches to the back of the computer with tape, and feels like leather (although, I’m sure it’s not leather). It definitely makes it feel like a “quality” product 😀. It’s a little bit more tricky to drop into the dock, but it makes it feel like a Laptop when you’re using it like one. The detachable keyboard is interesting. I’ve used it in the car, waiting for children to finish activities, and it’s fine, because it goes flat. I’ve detached the keyboard from the screen to just do tablet-y things with it, and that’s fine too.

So in summary

I think if I didn’t want it to be a tablet as much as a computer, I’d have been fine.

If you want a Windows Tablet that turns into a Laptop, it’s fine. If you want a low-profile desktop computer (in a dock) that can become a laptop, it’s fine.

But until Gnome or one of the other flavours gets a handle on how to do a reasonable on-screen keyboard… I don’t think I’ll be using Linux on here (because it’s also a tablet) for the next few months… and I think that’s going to be OK.

All of that said, if you use any Linux distributions with a tablet style mode, and you’ve got a working OSK, please contact me (via one of the links at the top of the site) to let me know what and how you did it, and I’ll give it a try too!

"Main console" by "Steve Parker" on Flickr

Running services (like SSH, nginx, etc) on Windows Subsystem for Linux (WSL1) on boot

I recently got a new laptop, and for various reasons, I’m going to be primarily running Windows on that laptop. However, I still like having a working SSH server, running in the context of my Windows Subsystem for Linux (WSL) environment.

Initially, trying to run service ssh start failed with an error, because you need to re-execute the ssh configuration steps which are missed in a WSL environment. To fix that, run sudo apt install --reinstall openssh-server.

Once you know your service runs OK, you start digging around to find out how to start it on boot, and you’ll see lots of people saying things like “Just run a shell script that starts your first service, and then another shell script for the next service.”

Well, the frustration for me is that Linux already has this capability – the current popular version is called SystemD, but a slightly older variant is still knocking around in modern linux distributions, and it’s called SystemV Init, often referred to as just “sysv” or “init.d”.

The way that those services work is that you have an “init” file in /etc/init.d and then those files have a symbolic link into a “runlevel” directory, for example /etc/rc3.d. Each symbolic link is named S##service or K##service, where the ## represents the order in which it’s to be launched. The SSH Daemon, for example, that I want to run is created in there as /etc/rc3.d/S01ssh.

So, how do I make this work in the grander scheme of WSL? I can’t use SystemD, where I could say systemctl enable --now ssh, instead I need to add a (yes, I know) shell script, which looks in my desired runlevel directory. Runlevel 3 is the level at which network services have started, hence using that one. If I was trying to set up a graphical desktop, I’d instead be looking to use Runlevel 5, but the X Windows system isn’t ported to Windows like that yet… Anyway.

Because the rc#.d directory already has this structure for ordering and naming services to load, I can just step over this directory looking for files which match or do not match the naming convention, and I do that with this script:

#! /bin/bash
function run_rc() {
  base="$(basename "$1")"
  if [[ ${base:0:1} == "S" ]]
    "$1" start
    "$1" stop

if [ "$1" != "" ] && [ -e "$1" ]
  run_rc "$1"
  if [ "$1" != "" ] && [ -e "/etc/rc${$1}.d/" ]
  for digit1 in {0..9}
    for digit2 in {0..9}
      find "/etc/rc${rc}.d/" -name "[SK]${digit1}${digit2}*" -exec "$0" '{}' \; 2>/dev/null

I’ve put this script in /opt/wsl_init.sh

This does a bit of trickery, but basically runs the bottom block first. It loops over the digits 0 to 9 twice (giving you 00, 01, 02 and so on up to 99) and looks in /etc/rc3.d for any file containing the filename starting S or K and then with the two digits you’ve looped to by that point. Finally, it runs itself again, passing the name of the file it just found, and this is where the top block comes in.

In the top block we look at the “basename” – the part of the path supplied, without any prefixed directories attached, and then extract just the first character (that’s the ${base:0:1} part) to see whether it’s an “S” or anything else. If it’s an S (which everything there is likely to be), it executes the task like this: /etc/rc3.d/S01ssh start and this works because it’s how that script is designed! You can run one of the following instances of this command: service ssh start, /etc/init.d/ssh start or /etc/rc3.d/S01ssh start. There are other options, notably “stop” or “status”, but these aren’t really useful here.

Now, how do we make Windows execute this on boot? I’m using NSSM, the “Non-sucking service manager” to add a line to the Windows System services. I placed the NSSM executable in C:\Program Files\nssm\nssm.exe, and then from a command line, ran C:\Program Files\nssm\nssm.exe install WSL_Init.

I configured it with the Application Path: C:\Windows\System32\wsl.exe and the Arguments: -d ubuntu -e sudo /opt/wsl_init.sh. Note that this only works because I’ve also got Sudo setup to execute this command without prompting for a password.

Here I invoke C:\Windows\System32\wsl.exe -d ubuntu -e sudo /opt/wsl_init.sh
I define the name of the service, as Services will see it, and also the description of the service.
I put in MY username and My Windows Password here, otherwise I’m not running WSL in my user context, but another one.

And then I rebooted. SSH was running as I needed it.

Featured image is “Main console” by “Steve Parker” on Flickr and is released under a CC-BY license.

"#security #lockpick" by "John Jones" on Flickr

Auto-starting an SSH Agent in Windows Subsystem for Linux

I tend to use Windows Subsystem for Linux (WSL) as a comprehensive SSH client, mostly for running things like Ansible scripts and Terraform. One of the issues I’ve had with it though is that, on a Linux GUI based system, I would start my SSH Agent on login, and then the first time I used an SSH key, I would unlock the key using the agent, and it would be cached for the duration of my logged in session.

While I was looking for something last night, I came across this solution on Stack Overflow (which in turn links to this blog post, which in turn links to this mailing list post) that suggests adding the following stanza to ~/.profile in WSL. I’m running the WSL version of Ubuntu 20.04, but the same principles apply on Cygwin, or, probably, any headless-server installation of a Linux distribution, if that’s your thing.

function start_agent {
    echo "Initialising new SSH agent..."
    /usr/bin/ssh-agent | sed 's/^echo/#echo/' > "${SSH_ENV}"
    echo succeeded
    chmod 600 "${SSH_ENV}"
    . "${SSH_ENV}" > /dev/null
# Source SSH settings, if applicable
if [ -f "${SSH_ENV}" ]; then
    . "${SSH_ENV}" > /dev/null
    ps -ef | grep ${SSH_AGENT_PID} | grep ssh-agent$ > /dev/null || {

Now, this part is all well-and-good, but what about that part where you want to SSH using a key, and then that being unlocked for the duration of your SSH Agent being available?

To get around that, in the same solution page, there is a suggestion of adding this line to your .ssh/config: AddKeysToAgent yes. I’ve previously suggested using dynamically included SSH configuration files, so in this case, I’d look for your file which contains your “wildcard” stanza (if you have one), and add the line there. This is what mine looks like:

Host *
  AddKeysToAgent yes
  IdentityFile ~/.ssh/MyCurrentKey

How does this help you? Well, if you’re using jump hosts (using ProxyJump MyBastionHost, for example) you’ll only be prompted for your SSH Key once, or if you typically do a lot of SSH sessions, you’ll only need to unlock your session once.

BUT, and I can’t really stress this enough, don’t use this on a shared or suspected compromised system! If you’ve got a root account which can access the content of your Agent’s Socket and PID, then any protections that private key may have held for your system is compromised.

Featured image is “#security #lockpick” by “John Jones” on Flickr and is released under a CC-BY-ND license.

"Fishing line and bobbin stuck on tree at Douthat State Park" by "Virginia State Parks" on Flickr

Note to self: Linux shell scripts don’t cope well with combined CRLF + LF files… Especially in User-Data / Custom Data / Cloud-Init scripts

This one is more a nudge to myself. On several occasions when building Infrastructure As Code (IAC), I split out a code sections into one or more files, for readability and reusability purposes. What I tended to do, and this was more apparent with the Linux builds than the Windows builds, was to forget to set the line terminator from CRLF to LF.

While this doesn’t really impact Windows builds too much (they’re kinda designed to support people being idiots with line endings now), Linux still really struggles with CRLF endings, and you’ll only see when you’ve broken this because you’ll completely fail to run any of the user-data script.

How do you determine this is your problem? Well, actually it’s a bit tricky, as neither cat, less, more or nano spot this issue. The only two things I found that identified it were file and vi.

The first part of the combined file with mixed line endings. This part has LF termination.
The second part of the combined file with mixed line endings. This part has CRLF termination.
What happens when we cat these two parts into one file? A file with CRLF, LF line terminators obviously!
What the combined file looks like in Vi. Note the blue ^M at the ends of the lines.

So, how to fix this? Assuming you’re using Visual Studio Code;

A failed line-ending clue in Visual Studio Code

You’ll notice this line showing “CRLF” in the status bar at the bottom of Code. Click on that, which brings up a discrete box near the top, as follows:

Oh no, it’s set to “CRLF”. That’s not what we want!

Selecting LF in that box changes the line feeds into LF for this file, but it’s not saved. Make sure you save this file before you re-run your terraform script!

Notice, we’re now using LF endings, but the file isn’t saved.

Fantastic! It’s all worked!

In Nano, I’ve opened the part with the invalid line endings.

Oh no! We have a “DOS Format” file. Quick, let’s fix it!

To fix this, we need to write the file out. Hit Ctrl+O. This tells us that we’re in DOS Format, and also gives us the keyboard combination to toggle “DOS Format” off – it’s Alt+D (In Unix/Linux world, the Alt key is referred to as the Meta key – hence M not A).

This is how we fix things

So, after hitting Alt+D, the “File Name to write” line changes, see below:

Yey, no pesky “DOS Format” warning here!

Using either editor (or any others, if you know how to solve line ending issues in other editors), you still need to combine your script back together before you can run it, so… do that, and your file will be fine to run! Good luck!

Featured image is “Fishing line and bobbin stuck on tree at Douthat State Park” by “Virginia State Parks” on Flickr and is released under a CC-BY license.

Opening to my video: Screencast 001 - Ansible and Inspec using Vagrant

Screencast 001: Ansible and Inspec with Vagrant and Git (a mentoring style video)

If you’ve ever wondered how I use Ansible and Inspec, or wondered why some of my Vagrant files look like they do, well, I want to start recording some “mentor” style videos… You know how, if you were sitting next to someone who’s a mentor to you, and you watch how they build a solution.

The first one was released last night!

I recently saw a video by Chris Hartjes on how he creates his TDD (Test driven development) based PHP projects, and I really wanted to emulate that style, but talking about the things I use.

This was my second attempt at recording a mentoring style video yesterday, the first was shown to the Admin Admin Podcast listeners group on Telegram, and then sacrificed to the demo gods (there were lots of issues in that first video) never to be seen again.

From a tooling perspective, I’m using a remote virtual machine running Ubuntu Mate 18.04 over RDP (to improve performance) with xrdp and Remmina, OBS is running locally to record the content, and I’m using Visual Studio Code, git, Vagrant and Virtualbox, as well as Ansible and Inspec.

Late edit 2020-02-29: Like videos like this, hate YouTube? It’s also on archive.org: https://archive.org/details/JonTheNiceGuyScreencast001

Late edit 2020-03-01: Popey told me about LBRY.tv when I announced this on the Admin Admin Podcast telegram channel, and so I’ve also copied the video to there: https://lbry.tv/@JonTheNiceGuy:b/Screencast001-Ansible-and-Inspec-with-Vagrant:8

"vieux port Marseille" by "Jeanne Menjoulet" on Flickr

Networking tricks with Multipass in Virtualbox on Windows (Bridged interfaces and Port Forwards)

TL;DR? Want to “just” bridge one or more interfaces to a Multipass instance when you’re using Virtualbox? See the Bridging Summary below. Want to do a port forward? See the Port Forward section below. You will need the psexec command and to execute this as an administrator. The use of these two may be considered a security incident on your computing environment, depending on how your security processes and infrastructure are defined and configured.

Ah Multipass. This is a tool created by Canonical to create a “A mini-cloud on your Mac or Windows workstation.” (from their website)…

I’ve often seen this endorsed as the tool of choice from Canonical employees to do “stuff” like run Kubernetes, develop tools for UBPorts (previously Ubuntu Touch) devices, and so on.

So far, it seems interesting. It’s a little bit like Vagrant with an in-built cloud-init Provisioner, and as I want to test out the cloud-init files I’m creating for AWS and Azure, that’d be so much easier than actually building the AWS or Azure machines, or finding a viable cloud-init plugin for Vagrant to test it out.

BUT… Multipass is really designed for Linux systems (running LibVirt), OS X (running HyperKit) and Windows (running Hyper-V). Even if I were using Windows 10 Pro on this machine, I use Virtualbox for “things” on my Windows Machine, and Hyper-V steals the VT-X bit, which means that VirtualBox can’t run x64 code…. Soooo I can’t use the Hyper-V mode.

Now, there is a “fix” for this. You can put Multipass into Virtualbox mode, which lets you run Multipass on Windows or OS X without using their designed-for hypervisor, but this has a downside, you see, VirtualBox doesn’t give MultiPass the same interface to route networking connections to the VM, and there’s currently no CLI or GUI options to say “bridge my network” or “forward a port” (in part because it needs to be portable to the native hypervisor options, apparently). So, I needed to fudge some things so I can get my beloved bridged connections.

I got to the point where I could do this, thanks to the responses to a few issues I raised on the Multipass Github issues, mostly #1333.

The first thing you need to install in Windows is PsExec, because Multipass runs it’s Virtual Machines as the SYSTEM account, and talking to SYSTEM account processes is nominally hard. Get PsExec from the SysInternals website. Some IT Security professionals will note the addition of PsExec as a potential security incident, but then again, they might also see the running of a virtual machine as a security incident too, as these aren’t controlled with a central image. Anyway… Just bear it in mind, and don’t shout at me if you get frogmarched in front of your CISO.

I’m guessing if you’re here, you’ve already installed Multipass, (but if not, and it seems interesting – it’s over at https://multipass.run. Get it and install it, then carry on…) and you’ve probably enabled the VirtualBox mode (if not – open a command prompt as administrator, and run “multipass set local.driver=virtualbox“). Now, you can start sorting out your bridges.

Sorting out bridges

First things first, you need to launch a virtual machine. I did, and it generated a name for my image.

C:\Users\JON>multipass launch
Launched: witty-kelpie

Fab! We have a running virtual machine, and you should be able to get a shell in there by running multipass shell "witty-kelpie" (the name of the machine it launched before). But, uh-oh. We have the “default” NAT interface of this device mapped, not a bridged interface.

C:\Users\JON>multipass shell "witty-kelpie"
Welcome to Ubuntu 18.04.3 LTS (GNU/Linux 4.15.0-76-generic x86_64)

 * Documentation:  https://help.ubuntu.com
 * Management:     https://landscape.canonical.com
 * Support:        https://ubuntu.com/advantage

  System information as of Thu Feb  6 10:56:38 GMT 2020

  System load:  0.3               Processes:             82
  Usage of /:   20.9% of 4.67GB   Users logged in:       0
  Memory usage: 11%               IP address for enp0s3:
  Swap usage:   0%

0 packages can be updated.
0 updates are security updates.

To run a command as administrator (user "root"), use "sudo <command>".
See "man sudo_root" for details.


So, exit the machine, and issue a multipass stop "witty-kelpie" command to ask Virtualbox to shut it down.

So, this is where the fun[1] part begins.
[1] The “Fun” part here depends on how you view this specific set of circumstances 😉

We need to get the descriptions of all the interfaces we might want to bridge to this device. I have three interfaces on my machine – a WiFi interface, a Ethernet interface on my laptop, and an Ethernet interface on my USB3 dock. At some point in the past, I renamed these interfaces, so I’d recognise them in the list of interfaces, so they’re not just called “Connection #1”, “Connection #2” and so on… but you should recognise your interfaces.

To get this list of interfaces, open PowerShell (as a “user”), and run this command:

PS C:\Users\JON> Get-NetAdapter -Physical | format-list -property "Name","DriverDescription"

Name              : On-Board Network Connection
DriverDescription : Intel(R) Ethernet Connection I219-LM

Name              : Wi-Fi
DriverDescription : Intel(R) Dual Band Wireless-AC 8260

Name              : Dock Network Connection
DriverDescription : DisplayLink Network Adapter NCM

For reasons best known to the Oracle team, they use the “Driver Description” to identify the interfaces, not the name assigned to the device by the user, so, before we get started, find your interface, and note down the description for later. If you want to bridge “all” of them, make a note of all the interfaces in question, and in the order you want to attach them. Note that Virtualbox doesn’t really like exposing more than 8 NICs without changing the Chipset to ICH9 (but really… 9+ NICs? really??) and the first one is already consumed with the NAT interface you’re using to connect to it… so that gives you 7 bridgeable interfaces. Whee!

So, now you know what interfaces you want to bridge, let’s configure the Virtualbox side. Like I said before you need psexec. I’ve got psexec stored in my Downloads folder. You can only run psexec as administrator, so open up an Administrator command prompt or powershell session, and run your command.

Just for clarity, your commands are likely to have some different paths, so remember that wherever “your” PsExec64.exe command is located, mine is in C:\Users\JON\Downloads\sysinternals\PsExec64.exe, and wherever your vboxmanage.exe is located, mine is in C:\Program Files\Oracle\VirtualBox\vboxmanage.exe.

Here, I’m going to attach my dock port (“DisplayLink Network Adapter NCM”) to the second VirtualBox interface, the Wifi adaptor to the third interface and my locally connected interface to the fourth interface. Your interfaces WILL have different descriptions, and you’re likely not to need quite so many of them!

C:\WINDOWS\system32>C:\Users\JON\Downloads\sysinternals\PsExec64.exe -s "c:\program files\oracle\virtualbox\vboxmanage" modifyvm "witty-kelpie" --nic2 bridged --bridgeadapter2 "DisplayLink Network Adapter NCM" --nic3 bridged --bridgeadapter3 "Intel(R) Dual Band Wireless-AC 8260" --nic4 bridged --bridgeadapter4 "Intel(R) Ethernet Connection I219-LM"

PsExec v2.2 - Execute processes remotely
Copyright (C) 2001-2016 Mark Russinovich
Sysinternals - www.sysinternals.com

c:\program files\oracle\virtualbox\vboxmanage exited on MINILITH with error code 0.

An error code of 0 means that it completed successfuly and with no issues.

If you wanted to use a “Host Only” network (if you’re used to using Vagrant, you might know it as “Private” Networking), then change the NIC you’re interested in from --nicX bridged --bridgeadapterX "Some Description" to --nicX hostonly --hostonlyadapterX "VirtualBox Host-Only Ethernet Adapter" (where X is replaced with the NIC number you want to swap, ranged between 2 and 8, as 1 is the NAT interface you use to SSH into the virtual machine.)

Now we need to check to make sure the machine has it’s requisite number of interfaces. We use the showvminfo flag to the vboxmanage command. It produces a LOT of content, so I’ve manually filtered the lines I want, but you should spot it reasonably quickly.

C:\WINDOWS\system32>C:\Users\JON\Downloads\sysinternals\PsExec64.exe -s "c:\program files\oracle\virtualbox\vboxmanage" showvminfo "witty-kelpie"

PsExec v2.2 - Execute processes remotely
Copyright (C) 2001-2016 Mark Russinovich
Sysinternals - www.sysinternals.com

Name:                        witty-kelpie
Groups:                      /Multipass
Guest OS:                    Ubuntu (64-bit)
NIC 1:                       MAC: 0800273CCED0, Attachment: NAT, Cable connected: on, Trace: off (file: none), Type: 82540EM, Reported speed: 0 Mbps, Boot priority: 0, Promisc Policy: deny, Bandwidth group: none
NIC 1 Settings:  MTU: 0, Socket (send: 64, receive: 64), TCP Window (send:64, receive: 64)
NIC 1 Rule(0):   name = ssh, protocol = tcp, host ip = , host port = 53507, guest ip = , guest port = 22
NIC 2:                       MAC: 080027303758, Attachment: Bridged Interface 'DisplayLink Network Adapter NCM', Cable connected: on, Trace: off (file: none), Type: 82540EM, Reported speed: 0 Mbps, Boot priority: 0, Promisc Policy: deny, Bandwidth group: none
NIC 3:                       MAC: 0800276EA174, Attachment: Bridged Interface 'Intel(R) Dual Band Wireless-AC 8260', Cable connected: on, Trace: off (file: none), Type: 82540EM, Reported speed: 0 Mbps, Boot priority: 0, Promisc Policy: deny, Bandwidth group: none
NIC 4:                       MAC: 080027042135, Attachment: Bridged Interface 'Intel(R) Ethernet Connection I219-LM', Cable connected: on, Trace: off (file: none), Type: 82540EM, Reported speed: 0 Mbps, Boot priority: 0, Promisc Policy: deny, Bandwidth group: none
NIC 5:                       disabled
NIC 6:                       disabled
NIC 7:                       disabled
NIC 8:                       disabled

Configured memory balloon size: 0MB

c:\program files\oracle\virtualbox\vboxmanage exited on MINILITH with error code 0.

Fab! We now have working interfaces… But wait, let’s start that VM back up and see what happens.

C:\Users\JON>multipass shell "witty-kelpie"
Welcome to Ubuntu 18.04.3 LTS (GNU/Linux 4.15.0-76-generic x86_64)

 * Documentation:  https://help.ubuntu.com
 * Management:     https://landscape.canonical.com
 * Support:        https://ubuntu.com/advantage

  System information as of Thu Feb  6 11:31:08 GMT 2020

  System load:  0.1               Processes:             84
  Usage of /:   21.1% of 4.67GB   Users logged in:       0
  Memory usage: 11%               IP address for enp0s3:
  Swap usage:   0%

0 packages can be updated.
0 updates are security updates.

Last login: Thu Feb  6 10:56:45 2020 from
To run a command as administrator (user "root"), use "sudo <command>".
See "man sudo_root" for details.


Wait, what….. We’ve still only got the one interface up with an IP address… OK, let’s fix this!

As of Ubuntu 18.04, interfaces are managed using Netplan, and, well, when the VM was built, it didn’t know about any interface past the first one, so we need to get Netplan to get them enabled. Let’s check they’re detected by the VM, and see what they’re all called:

ubuntu@witty-kelpie:~$ ip link
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN mode DEFAULT group default qlen 1000
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
2: enp0s3: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc fq_codel state UP mode DEFAULT group default qlen 1000
    link/ether 08:00:27:3c:ce:d0 brd ff:ff:ff:ff:ff:ff
3: enp0s8: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN mode DEFAULT group default qlen 1000
    link/ether 08:00:27:30:37:58 brd ff:ff:ff:ff:ff:ff
4: enp0s9: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN mode DEFAULT group default qlen 1000
    link/ether 08:00:27:6e:a1:74 brd ff:ff:ff:ff:ff:ff
5: enp0s10: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN mode DEFAULT group default qlen 1000
    link/ether 08:00:27:04:21:35 brd ff:ff:ff:ff:ff:ff

If you compare the link/ether lines to the output from showvminfo we executed before, you’ll see that the MAC address against enp0s3 matches the NAT interface, while enp0s8 matches the DisplayLink adapter, and so on… So we basically want to ask NetPlan to do a DHCP lookup for all the new interfaces we’ve added to it. If you’ve got 1 NAT and 7 physical interfaces (why oh why…) then you’d have enp0s8, 9, 10, 16, 17, 18 and 19 (I’ll come back to the random numbering in a tic)… so we now need to ask Netplan to do DHCP on all of those interfaces (assuming we’ll be asking for them all to come up!)

If we want to push that in, then we need to add a new file in /etc/netplan called something like 60-extra-interfaces.yaml, that should contain:

      optional: yes
      dhcp4: yes
        route-metric: 10
      optional: yes
      dhcp4: yes
        route-metric: 11
      optional: yes
      dhcp4: yes
        route-metric: 12
      optional: yes
      dhcp4: yes
        route-metric: 13
      optional: yes
      dhcp4: yes
        route-metric: 14
      optional: yes
      dhcp4: yes
        route-metric: 15
      optional: yes
      dhcp4: yes
        route-metric: 16

Going through this, we basically ask netplan not to assume the interfaces are attached. This stops the boot process for waiting for a timeout to configure each of the interfaces before proceeding, so it means your boot should be reasonably fast, particularly if you don’t always attach a network cable or join a Wifi network on all your interfaces!

We also say to assume we want IPv4 DHCP on each of those interfaces. I’ve done IPv4 only, as most people don’t use IPv6 at home, but if you are doing IPv6 as well, then you’d also need the same lines that start dhcp4 copied to show dhcp6 (like dhcp6: yes and dhcp6-overrides: route-metric: 10)

The eagle eyed of you might notice that the route metric increases for each extra interface. This is because realistically, if you have two interfaces connected (perhaps if you’ve got wifi enabled, and plug a network cable in), then you’re more likely to want to prioritize traffic going over the lower numbered interfaces than the higher number interfaces.

Once you’ve created this file, you need to run netplan apply or reboot your machine.

So, yehr, that gets you sorted on the interface front.

Bridging Summary

To review, you launch your machine with multipass launch, and immediately stop it with multipass stop "vm-name", then, as an admin, run psexec vboxmanage modifyvm "vm-name" --nic2 bridged --bridgedadapter2 "NIC description", and then start the machine with multipass start "vm-name". Lastly, ask the interface to do DHCP by manipulating your Netplan configuration.

Interface Names in VirtualBox

Just a quick note on the fact that the interface names aren’t called things like eth0 any more. A few years back, Ubuntu (amongst pretty much all of the Linux distribution vendors) changed from using eth0 style naming to what they call “Predictable Network Interface Names”. This derives the names from things like, what the BIOS provides for on-board interfaces, slot index numbers for PCI Express ports, and for this case, the “geographic location of the connector”. In Virtualbox, these interfaces are provided as the “Geographically” attached to “port 0” (so enp0 are all on port 0), but for some reason, they broadcast themselves as being attached to the port 0 at “slots” 3, 8, 9, 10, 16, 17, 18 and 19… hence enp0s3 and so on. shrug It just means that if you don’t have the interfaces coming up on the interfaces you’re expecting, you need to run ip link to confirm the MAC addresses match.

Port Forwarding

Unlike with the Bridging, we don’t need to power down the VM to add the extra interfaces, we just need to use psexec (as an admin again) to execute a vboxmanage command – in this case, it’s:

C:\WINDOWS\system32>C:\Users\JON\Downloads\sysinternals\PsExec64.exe -s "c:\program files\oracle\virtualbox\vboxmanage" controlvm "witty-kelpie" --natpf1 "myport,tcp,,1234,,2345"

OK, that’s a bit more obscure. Basically it says “Create a NAT rule on NIC 1 called ‘myport’ to forward TCP connections from port 1234 attached to any IP associated to the host OS to port 2345 attached to the DHCP supplied IP on the guest OS”.

If we wanted to run a DNS server in our VM, we could run multiple NAT rules in the same command, like this:

C:\WINDOWS\system32>C:\Users\JON\Downloads\sysinternals\PsExec64.exe -s "c:\program files\oracle\virtualbox\vboxmanage" controlvm "witty-kelpie" --natpf1 "TCP DNS,tcp,,53,,53" --natpf1 "UDP DNS,udp,,53,,53"

If we then decide we don’t need those NAT rules any more, we just (with psexec and appropriate paths) issue: vboxmanage controlvm "vm-name" --natpf1 delete "TCP DNS"

Using ifupdown instead of netplan

Late Edit 2020-04-01: On Github, someone asked me how they could use the same type of config with netplan, but instead on a 16.04 system. Ubuntu 16.04 doesn’t use netplan, but instead uses ifupdown instead. Here’s how to configure the file for ifupdown:

You can either add the following stanzas to /etc/network/interfaces, or create a separate file for each interface in /etc/network/interfaces.d/<number>-<interface>.cfg (e.g. /etc/network/interfaces.d/10-enp0s8.cfg)

allow-hotplug enp0s8
iface enp0s8 inet dhcp
  metric 10

To re-iterate, in the above netplan file, the interfaces we identified were: enp0s8, enp0s9, enp0s10, enp0s16, enp0s17, enp0s18 and enp0s19. Each interface was incrementally assigned a route metric, starting at 10 and ending at 16, so enp0s8 has a metric of 10, while enp0s16 has a metric of 13, and so on. To build these files, I’ve created this brief shell script you could use:

export metric=10
for int in 8 9 10 16 17 18 19
  echo -e "allow-hotplug enp0s${int}\niface enp0s${int} inet dhcp\n  metric $metric" > /etc/network/interfaces.d/enp0s${int}.cfg

As before, you could reboot to make the changes to the interfaces. Bear in mind, however, that unlike Netplan, these interfaces will try and DHCP on boot with this configuration, so boot time will take longer if every interface attached isn’t connected to a network.

Using NAT Network instead of NAT Interface

Late update 2020-05-26: Ruzsinsky contacted me by email to ask how I’d use a “NAT Network” instead of a “NAT interface”. Essentially, it’s the same as the Bridged interface above, with one other tweak first, we need to create the Net Network, with this command (as an Admin)

C:\WINDOWS\system32>C:\Users\JON\Downloads\sysinternals\PsExec64.exe -s "c:\program files\oracle\virtualbox\vboxmanage" natnetwork add --netname MyNet --network

Next, stop your multipass virtual machine with multipass stop "witty-kelpie", and configure your second interface, like this:

C:\WINDOWS\system32>C:\Users\JON\Downloads\sysinternals\PsExec64.exe -s "c:\program files\oracle\virtualbox\vboxmanage" modifyvm "witty-kelpie" --nic2 natnetwork --nat-network2 "MyNet"

PsExec v2.2 - Execute processes remotely
Copyright (C) 2001-2016 Mark Russinovich
Sysinternals - www.sysinternals.com

c:\program files\oracle\virtualbox\vboxmanage exited on MINILITH with error code 0.

Start the vm with multipass start "witty-kelpie", open a shell with it multipass shell "witty-kelpie", become root sudo -i and then configure the interface in /etc/netplan/60-extra-interfaces.yaml like we did before:

      optional: yes
      dhcp4: yes
        route-metric: 10

And then run netplan apply or reboot.

What I would say, however, is that the first interface seems to be expected to be a NAT interface, at which point, having a NAT network as well seems a bit pointless. You might be better off using a “Host Only” (or “Private”) network for any inter-host communications between nodes at a network level… But you know your environments and requirements better than I do :)

Featured image is “vieux port Marseille” by “Jeanne Menjoulet” on Flickr and is released under a CC-BY-ND license.

"Unnatural Love" by "Keith Garner" on Flickr

Configuring a Remote Desktop (Gnome Shell) for Ubuntu

I started thinking a couple of weeks ago, when my coding laptop broke, that it would be really useful to have a development machine somewhere else that I could use.

It wouldn’t need a lot of power (after all, I’m mostly developing web apps and not compiling stuff), but it does need to be a desktop OS, as I rather like being able to open code editors and suchlike, while I’ve got a web browser open.

I have an Android tablet, which while it’s great for being a tablet, it’s not much use as a desktop, and … yes, I’ve got a work laptop, but I don’t really want to install software on that (and I don’t think my admin team would be happy if I did).

Also, I quite like Linux.

Some time ago, I spotted that AWS has a “Virtual Desktop” environment, and I think that’s kinda what I’m after. Something I can spin up, run for a bit and then shut it down, so I thought I’d build something like that… but not pesky Windows, after all… who likes Windows, eh? ;)

So, I built a Virtual Desktop Environment (VDE) in AWS, using Terraform and a bit of shell script!

I start from an Ubuntu 18.04 server image, and, after the install is complete, I run this user-data script inside it. Yes, I know I could be doing this with Ansible, but… eh, I wanted it to be a quick deployment ;)

Oh, and there’s a couple of Terraform managed variables in here – ${aws_eip.vde.public_ip} is the AWS public IP address assigned to this host., ${var.firstuser} is the username we want to rename “ubuntu” (the stock server username) to. ${var.firstgecos} is the user’s “real name” which the machine identifies the user as (like “Log out Jon Spriggs” and so on). ${var.userpw} is either the password you want it to use, OR (by default) pwgen 12 which generates a 12 character long password. ${var.desktopenv} is the name of the desktop environment I want to install (Ubuntu by default) and … well, ${var.var_start} is a bit of a fudge, because I couldn’t, in a hurry, work out how to tell Terraform not to mangle the bash variable allocation of ${somevar} which is the format that Terraform also uses. D’oh.

#! /bin/bash
# Set Hostname
hostnamectl set-hostname vde.${aws_eip.vde.public_ip}.nip.io
# Change User
if [ ! "$user" == 'ubuntu' ]
  until usermod -c "${var.firstgecos}" -l $user ubuntu ; do sleep 5 ; done
  until groupmod -n $user ubuntu ; do sleep 5 ; done
  until usermod  -d /home/$user -m $user ; do sleep 5 ; done
  if [ -f /etc/sudoers.d/90-cloudimg-ubuntu ]; then
    mv /etc/sudoers.d/90-cloudimg-ubuntu /etc/sudoers.d/90-cloud-init-users
  perl -pi -e "s/ubuntu/$user/g;" /etc/sudoers.d/90-cloud-init-users
if [ '${var.userpw}' == '$(pwgen 12)' ]
  apt update && apt install pwgen
echo "$newpw" > /var/log/userpw
echo -e "$fullpw" | passwd $user
# Install Desktop and RDP
apt-get update
export DEBIAN_FRONTEND=noninteractive
apt-get full-upgrade -yq
apt-get autoremove -y
apt-get autoclean -y
apt-get install -y ${var.desktopenv}-desktop xrdp certbot
# Configure Certbot
echo "#!/bin/sh" > /etc/letsencrypt/merge_cert.sh
echo 'cat ${var.var_start}{RENEWED_LINEAGE}/privkey.pem ${var.var_start}{RENEWED_LINEAGE}/fullchain.pem > ${var.var_start}{RENEWED_LINEAGE}/merged.pem' >> /etc/letsencrypt/merge_cert.sh
echo 'chmod 640 ${var.var_start}{RENEWED_LINEAGE}/merged.pem' >> /etc/letsencrypt/merge_cert.sh
chmod 750 /etc/letsencrypt/merge_cert.sh
certbot certonly --standalone --deploy-hook /etc/letsencrypt/merge_cert.sh -n -d vde.${aws_eip.vde.public_ip}.nip.io -d ${aws_eip.vde.public_ip}.nip.io --register-unsafely-without-email --agree-tos
# Based on https://www.snel.com/support/xrdp-with-lets-encrypt-on-ubuntu-18-04/
sed -i 's~^certificate=$~certificate=/etc/letsencrypt/live/vde.${aws_eip.vde.public_ip}.nip.io/fullchain.pem~; s~^key_file=$~key_file=/etc/letsencrypt/live/vde.${aws_eip.vde.public_ip}.nip.io/privkey.pem' /etc/xrdp/xrdp.ini
# Fix colord remote user issue
# Derived from http://c-nergy.be/blog/?p=12043
echo "[Allow Colord all Users]
ResultActive=yes" > /etc/polkit-1/localauthority/50-local.d/45-allow.colord.pkla
# Configure Desktop
if [ '${var.desktopenv}' == 'ubuntu' ]
  echo "#!/bin/bash" > /tmp/desktop_settings
  echo "gsettings set org.gnome.desktop.input-sources sources \"[('xkb', 'gb')]\"" >> /tmp/desktop_settings
  echo "gsettings set org.gnome.desktop.app-folders folder-children \"['Utilities', 'Sundry', 'YaST']\"" >> /tmp/desktop_settings
  echo "gsettings set org.gnome.desktop.privacy report-technical-problems false" >> /tmp/desktop_settings
  echo "gsettings set org.gnome.desktop.screensaver lock-enabled false" >> /tmp/desktop_settings
  echo "gsettings set org.gnome.desktop.session idle-delay 0" >> /tmp/desktop_settings
  echo "echo yes > /home/${var.firstuser}/.config/gnome-initial-setup-done" >> /tmp/desktop_settings
  sudo -H -u ${var.firstuser} dbus-launch --exit-with-session bash /tmp/desktop_settings
  rm -f /tmp/desktop_settings
# Install VSCode
wget https://vscode-update.azurewebsites.net/latest/linux-deb-x64/stable -O /tmp/vscode.deb
apt install -y /tmp/vscode.deb
rm /var/crash/*
shutdown -r now

Ubuntu 18.04 has a “first login” wizard, that lets you pre-set up things like, what language will you be using. I bypassed this with the gsettings commands towards the end of the script, and writing the string “yes” to ~/.config/gnome-initial-setup-done.

Also, I wanted to be able to RDP to it. I’m a bit concerned by the use of VNC, especially where RDP is more than capable. It’s just an apt-install away, so… that’s what I do. But, because I’m RDP’ing into this box, I wanted to prevent the RDP session from locking, so I provide two commands to the session: gsettings set org.gnome.desktop.screensaver lock-enabled false which removes the screensaver’s ability to lock the screen, and gsettings set org.gnome.desktop.session idle-delay 0 which stops the screensaver from even starting in the first place.

Now all I need to do is to figure out where I’m going to store my code between boots ;)

So, in summary, I now have a Virtual Machine, which runs Ubuntu 18.04 Desktop, in AWS, with an RDP connection (powered by xRDP), and a disabled screensaver. Job done, I think!

Oh, and if I’m doing it “wrong”, let me know in the comments? :)

Featured image is “Unnatural Love” by “Keith Garner” on Flickr and is released under a CC-BY-SA license.