About David Peach

this is me

Using ansible to prepare a digital ocean droplet to host a static website

Posted on 29th August 2023 by David Peach

Preface

This guide comes logically after the previous one I wrote about setting up a digital ocean server with Terraform.

You can clone my website’s ansible repository for reference.

The main logic for this Ansible configuration happens in the setup.yml file. This file can be called whatever you like as we’ll call it by name later on.

Installing Ansible

You can install Ansible with your package manager of choice.

I install it using pacman on Arch Linux:

Bash

sudo pacman -S ansible

The inventory.yml file

The inventory file is where I have set the relative configuration needed for the playbook.

The all key contains all of the host configurations (although I’m only using a single one).

Within that all key is vars.ansible_ssh_private_key_file which is just the local path to the ssh private key used to access the server.

This is the key I set up with Terraform in the previous guide.

Then the hosts key just contains the hosts I want to be able to target (im using the domain name that I set up in the previous Terraform guide)

The setup.yml file explained

The setup.yml file is what is known as an “Ansible Playbook”.

From my limited working knowledge of Ansible, a playbook is basically a set of tasks that are run against a server or a collection of servers.

In my own one I am currently only running it against a single server, which I am targeting via its domain name of “zet.davidpeach.me”

YAML

- hosts: all
  become: true
  user: root
  vars_files:
    - vars/default.yml

This first section is the setup of the playbook.

hosts:all tells it to run against all hosts that are defined in the ./inventory.yml file.

become:true is saying that ansible will switch to the root user on the server (defined on the next line with user: root) before running the playbook tasks.

The vars_files: part lets you set relative paths to files containing variables that are used in the playbook and inside the file ./files/nginx.conf.j2.

I wont go through each of the variables but hopefully you can see what they are doing.

The Playbook Tasks

Each of the tasks in the Playbook has a descriptive title that hopefully does well in explaining what the steps are doing.

The key value pairs of configuration after each of the task titles are pre-defined settings available to use in ansible.

The tasks read from top to bottom and essentially automate the steps that normally need to be manually done when preparing a server.

Running the playbook

Bash

cd ansible-project

ansible-playbook setup.yml -i inventory.yml

This command should start Ansible off. You should get the usual message about trusting the target host when first connecting to the server. Just answer “yes” and press enter.

You should now see the output for each step defined in the playbook.

The server should now be ready to deploy to.

Testing your webserver

In the ./files/nginx.conf.j2 there is a root directive on live 3. For me this is set to /var/www/{{ http_host }}. (http_host is a variable set in the vars/default.yml file).

SSH on to the server, using the private ssh key from the keypair I am using (see the Terraform guide for reference).

Bash

ssh -i ~/.ssh/id_rsa.davidpeachme zet.davidpeach.me

Then on the server, create a basic index.html file in the website root defined in the default nginx file:

Bash

cd /var/www/zet.davidpeach.me
touch index.html
echo "hello world" > index.html

Now, going to your website url in a browser, you should be able to see the text “hello world” in the top left.

The server is ready to host a static html website.

Next Step

You can use whatever method you prefer to get your html files on to your server.

You could use rsync, scp, an overly-complicated CI pipeline, or – if your’e using lupo – your could have lupo deploy it straight to your server for you.

Setting up a Digital Ocean droplet for a Lupo website with Terraform

Posted on 29th August 2023 by David Peach

Overview of this guide

My Terraform Repository used in this guide

Terraform is a program that enables you to set up all of your cloud-based infrastructure with configuration files. This is opposed to the traditional way of logging into a cloud provider’s dashboard and manually clicking buttons and setting up things yourself.

This is known as “Infrastructure as Code”.

It can be intimidating to get started, but my aim with this guide is to get you to the point of being able to deploy a single server on Digital Ocean, along with some surrounding items like a DNS A record and an ssh key for remote access.

This guide assumes that you have a Digital Ocean account and that you also have your domain and nameservers setup to point to Digital Ocean.

You can then build upon those foundations and work on building out your own desired infrastructures.

The Terraform Flow

As a brief outline, here is what will happen when working with terraform, and will hopefully give you a broad picture from which I can fill in the blanks below.

Firstly we write a configuration file that defines the infrastructure that we want.
Then we need to set up any access tokens, ssh keys and terraform variables. Basically anything that our Terraform configuration needs to be able to complete its task.
Finally we run the terraform plan command to test our infrastructure configuration, and then terraform apply to make it all live.

Installing the Terraform program

Terraform has installation instructions, but you may be able to find it with your package manager.

Here I am installing it on Arch Linux, by the way, with pacman

Bash

sudo pacman -S terraform

Setting the required variables

The configuration file for the infrastructure I am using requires only a single variable from outside. That is the do_token.

This is created manually in the API section of the Digital Ocean dashboard. Create yours and keep its value to hand for usage later.

Terraform accepts variables in a number of ways. I opt to save my tokens in my local password manager, and then use them when prompted by the terraform command. This is slightly more long-winding than just setting a terraform-specific env in your bashrc. However, I recently learned off rwxrob how much of a bad idea that is.

Creating an ssh key

In the main.tf file, I could have set the ssh public key path to my existing one. However, I thought I’d create a key pair specific for my website deployment.

Bash

ssh-keygen -t rsa

I give it a different name so as to not override my standard id_rsa one. I call it id_rsa.davidpeachme just so I know which is my website server one at a glance.

Describing your desired infrastructure with code

Terraform uses a declaritive language, as opposed to imperetive.

What this means for you, is that you write configuration files that describe the state that you want your infrastructure to be in. For example if you want a single server, you just add the server spec in your configuration and Terraform will work out how best to create it for you.

You dont need to be concerned with the nitty gritty of how it is achieved.

I have a real-life example that will show you exactly what a minimal configuration can look like.

Clone / fork the repository for my website server.

Explaination of my terraform repository

YAML

terraform {
  required_providers {
    digitalocean = {
      source = "digitalocean/digitalocean"
      version = "~> 2.0"
    }
  }
}

variable "do_token" {}

# Variables whose values are defined in ./terraform.tfvars
variable "domain_name" {}
variable "droplet_image" {}
variable "droplet_name" {}
variable "droplet_region" {}
variable "droplet_size" {}
variable "ssh_key_name" {}
variable "ssh_local_path" {}

provider "digitalocean" {
  token = var.do_token
}

The first block tells terraform which providers I want to use. Providers are essentially the third-party APIs that I am going to interact with.

Since I’m only creating a Digital Ocean droplet, and a couple of surrounding resources, I only need the digitalocean/digitalocean provider.

The second block above tells terraform that it should expect – and require – a single variable to be able to run. This is the Digital Ocean Access Token that was obtained above in the previous section, from the Digital Ocean dashboard.

Following that are the variables that I have defined myself in the ./terraform.tfvars file. That tfvars file would normally be kept out of a public repository. However, I kept it in so that you could hopefully just fork my repo and change those values for your own usage.

The bottom block is the setting up of the provider. Basically just passing the access token into the provider so that it can perform the necessary API calls it needs to.

YAML

resource "digitalocean_ssh_key" "ssh_key" {
  name       = var.ssh_key_name
  public_key = file(var.ssh_local_path)
}

Here is the first resource that I am telling terraform to create. Its taking a public key on my local filesystem and sending it to Digital Ocean.

This is needed for ssh access to the server once it is ready. However, it is added to the root account on the server.

I use Ansible for setting up the server with the required programs once Terraform has built it. So this ssh key is actually used by Ansible to gain access to do its thing.

I will have a separate guide soon on how I use ansible to set my server up ready to host my static website.

YAML

resource "digitalocean_droplet" "droplet" {
  image    = var.droplet_image
  name     = var.droplet_name
  region   = var.droplet_region
  size     = var.droplet_size
  ssh_keys = [digitalocean_ssh_key.ssh_key.fingerprint]
}

Here is the meat of the infrastructure – the droplet itself. I am telling it what operating system image I want to use; what size and region I want; and am telling it to make use of the ssh key I added in the previous block.

YAML

data "digitalocean_domain" "domain" {
  name = var.domain_name
}

This block is a little different. Here I am using the data property to grab information about something that already exists in my Digital Ocean account.

I have already set up my domain in Digital Ocean’s networking area.

This is the overarching domain itself – not the specific A record that will point to the server.

The reason i’m doing it this way, is because I have got mailbox settings and TXT records that are working, so i dont want them to be potentially torn down and re-created with the rest of my infrastructure if I ever run terraform destroy.

YAML

resource "digitalocean_record" "record" {
  domain = data.digitalocean_domain.domain.id
  type   = "A"
  name   = "@"
  ttl    = 60
  value  = "${digitalocean_droplet.droplet.ipv4_address}"
}

The final block creates the actual A record with my existing domain settings.

It uses the domain id given back by the data block i defined above, and the ip address of the created droplet for the A record value.

Testing and Running the config to create the infrastructure

If you now go into the root of your terraform project and run the following command, you should see it displays a write up of what it intends to create:

Bash

terraform plan

If the output looks okay to you, then type the following command and enter “yes” when it asks you:

Bash

terraform apply

This should create the three items of infrastructure we have defined.

Next Step

Next we need to set that server up with the required software needed to run a static html website.

I will be doing this with a program called Ansible.

I’ll be writing up those steps in a zet very soon.

Beyond Aliases — define your development workflow with custom bash scripts

Posted on 25th August 2023 by David Peach

Being a Linux user for just over 10 years now, I can’t imagine my life with my aliases.

Aliases help with removing the repetition of commonly-used commands on a system.

For example, here’s some of my own that I use with the Laravel framework:

Bash

alias a="php artisan"
alias sail='[ -f sail ] && bash sail || bash vendor/bin/sail'
alias stan="./vendor/bin/phpstan analyse"

You can set these in your ~/.bashrc file. See mine in my dotfiles as a fuller example.

However, I recently came to want greater control over my development workflow. And so, with the help of videos by rwxrob, I came to embrace the idea of learning bash, and writing my own little scripts to help in various places in my workflow.

A custom bash script

For the example here, I’ll use the action of wanting to “exec” on to a local docker container.

Sometimes you’ll want to get into a shell within a local docker container to test / debug things.

I found I was repeating the same steps to do this and so I made a little script.

Here is the script in full:

Bash

#!/bin/bash

docker container ls | fzf | awk '{print $1}' | \
xargs -o -I % docker exec -it % bash

Breaking it down

In order to better understand this script I’ll assume no prior knowledge and explain some bash concepts along the way.

Sh-bang line.

the first line is the “sh-bang”. It basically tells your shell which binary should execute this script when ran.

For example you could write a valid php script and add #!/usr/bin/php at the top, which would tell the shell to use your php binary to interpret the script.

So #!/usr/bash means we are writing a bash script.

Pipes

The pipe symbol: |.

In brief, a “pipe” in bash is a way to pass the output of the left hand command to the input of the right hand command.

So the order of the commands to be ran in the script is in this order:

docker container ls
fzf
awk ‘{print $1}’
xargs -o -I % docker exec -it % bash

docker container ls

This gives us the list of currently-running containers on our system. The output is the list like so (I’ve used an image as the formatting gets messed up when pasting into a post as text) :

fzf

So the output of the docker container ls command above is the table in the image above, which is several rows of text.

fzf is a “fuzzy finder” tool, which can be passed a list of pretty much anything, which can then be searched over by “fuzzy searching” the list.

In this case the list is each row of that output (header row included)

When you select (press enter) on your chosen row, that row of text is returned as the output of the command.

In this image example you can see I’ve typed in “app” to search for, and it has highlighted the closest matching row.

awk ‘{print $1}’

awk is an extremely powerful tool, built into linux distributions, that allows you to parse structured text and return specific parts of that text.

'{print $1}' is saying “take whatever input I’m given, split it up based on a delimeter, and return the item that is 1st ($1).

The default delimeter is a space. So looking at that previous image example, the first piece of text in the docker image rows is the image ID: `”df96280be3ad” in the app image chosen just above.

So pressing enter for that row from fzf, wil pass it to awk, which will then split that row up by spaces and return you the first element from that internal array of text items.

xargs -o -I % docker exec -it % bash

xargs is another powerful tool, which enables you to pass what ever is given as input, into another command. I’ll break it down further to explain the flow:

The beginning of the xargs command is as so:

Bash

xargs -o -I %

-o is needed when running an “interactive application”. Since our goal is to “exec” on to the docker container we choose, interactive is what we need. -o means to “open stdin (standard in) as /dev/tty in the child process before executing the command we specify.

Next, -I % is us telling xargs, “when you next see the ‘%’ character, replace it with what we give you as input. Which in this case will be that docker container ID returned from the awk command previously.

So when you replace the % character in the command that we are giving xargs, it will read as such:

Bash

docker exec -it df96280be3ad bash

This is will “exec” on to that docker container and immediately run “bash” in that container.

Goal complete.

Put it in a script file

So all that’s needed now, is to have that full set of piped commands in an executable script:

Bash

#!/bin/bash

docker container ls | fzf | awk '{print $1}' | xargs -o -I % docker exec -it % bash

My own version of this script is in a file called d8exec, which after saving it I ran:

Bash

chmod +x ./d8exec

Call the script

In order to be able to call your script from anywhere in your terminal, you just need to add the script to a directory that is in your $PATH. I keep mine at ~/.local/bin/, which is pretty standard for a user’s own scripts in Linux.

You can see how I set my own in my .bashrc file here. The section that reads $HOME/.local/bin is the relevant piece. Each folder that is added to the $PATH is separated by the : character.

Feel free to explore further

You can look over all of my own little scripts in my bin folder for more inspiration for your own bash adventures.

Have fun. And don’t put anything into your scripts that you wouldn’t want others seeing (api keys / secrets etc)

Strange Things are Afoot

Aside

Complete a task for a stranger
Red Dead Redemption

Clemency Pays

Aside

Capture a bounty alive.
— Red Dead Redemption

That Government Boy

Aside

Complete “Exodus in America”
— Red Dead Redemption

Foxglove at Pipe Hall Farm

Image

nice flower.

Setting up a GPG Key with git to sign your commits

Posted on 5th June 2023 by David Peach

Signing your git commits with GPG is really easy to set up and I’m always surprised by how many developers I meet that don’t do this.

Of course it’s not required to push commits and has no baring on quality of code. But that green verified message next to your commits does feel good.

Essentially there are three parts to this:

Create your GPG key
Tell git to use your GPG key to sign your commits
Upload the public part of your GPG key to Gitlab / Github / etc

Creating the GPG key if needed

gpg --full-generate-key

In the interactive guide, I choose:

(1) RSA and RSA (default)
4096 bits long
Does not expire
Fill in Name, Email, Comment and Confirm.
Enter passphrase when prompted.

Getting the Key ID

This will list all of your keys:

gpg --list-secret-keys --keyid-format=long

Example of the output:

sec   rsa4096/THIS0IS0YOUR0KEY0ID 2020-12-25 [SC]
      KGHJ64GHG6HJGH5J4G6H5465HJGHJGHJG56HJ5GY
uid                 [ultimate] Bob GPG Key<mail@your-domain.co.uk>

In that example, the key id that you would need next is “THIS0IS0YOUR0KEY0ID” from the first line, after the forward slash.

Tell your local git about the signing key

To set the gpg key as the signing key for all of your git projects, run the following global git command:

git config --global user.signingkey THIS0IS0YOUR0KEY0ID

If you want to do it on a repository by repository basis, you can run it from within each project, and omit the --global flag:

git config user.signingkey THIS0IS0YOUR0KEY0ID

Signing your commits

You can either set commit signing to true for all projects as the default, or by a repo by repo basis.

# global
git config --global commit.gpgsign true

# local
git config commit.gpgsign true

If you wanted to, you could even decide to sign commits per each commit, by not setting it as a config setting, but passing a flag on every commit:

git commit -S -m "My signed commit message"

Adding your public key to gitlab / github / wherever

Firstly export the public part of your key using your key id. Again, using the example key id from above:

# Show your public key in terminal
gpg --armor --export THIS0IS0YOUR0KEY0ID

# Copy straight to your system clipboard using "xclip"
gpg --armor --export THIS0IS0YOUR0KEY0ID | xclip -sel clipboard

This will spit out a large key text block begining and ending with comments. Copy all of the text that it gives you and paste it into the gpg textbox in your git forge of choice – gitlab / github / gitea / etc.

How I use vimwiki in neovim

Posted on 1st April 2023 by David Peach

This post is currently in-progress, and is more of a brain-dump right now. But I like to share as often as I can otherwise I’d never share anything 🙂

Please view the official Vimwiki Github repository for up-to-date details of Vimwiki usage and installation. This page just documents my own processes at the time.

Installation

Add the following to plugins.lua

use "vimwiki/vimwiki"

Run the following two commands separately in the neovim command line:

:PackerSync
:PackerInstall

Close and re-open Neovim.

How I configure Vimwiki

I have 2 separate wikis set up in my Neovim.

One for my personal homepage and one for my commonplace site.

I set these up by adding the following in my dotfiles, at the following position: $NEOVIM_CONFIG_ROOT/after/plugin/vimwiki.lua. So for me that would be ~/.config/nvim/after/plugin/vimwiki.lua.

You could also put this command inside the config function in your plugins.lua file, where you require the vimwiki plugin. I just tend to put all my plugin-specific settings in their own “after/plugin” files for organisation.

vim.cmd([[
  let wiki_1 = {}
  let wiki_1.path = '~/vimwiki/website/'
  let wiki_1.html_template = '~/vimwiki/website_html/'
  let wiki_2 = {}
  let wiki_2.path = '~/vimwiki/commonplace/'
  let wiki_2.html_template = '~/vimwiki/commonplace_html/'
  let g:vimwiki_list = [wiki_1, wiki_2]
  call vimwiki#vars#init()
]])

The path keys tell vimwiki where to plave the root index.wiki file for each wiki you configure.

The html_template keys tell vimwiki where to place the compiled html files (when running the :VimwikiAll2HTML command).

I keep them separate as I am deploying them to separate domains on my server.

When I want to open and edit my website wiki, I enter 1<leader>ww.

When I want to open and edit my commonplace wiki, I enter 2<leader>ww.

Pressing those key bindings for the first time will ask you if you want the directories creating.

How I use vimwiki

At the moment, my usage is standard to what is described in the Github repository linked at the top of this page.

When I develop any custom workflows I’ll add them here.

Deployment

Setting up a server to deploy to is outside the scope of this post, but hope to write up a quick guide soon.

I run the following command from within vim on one of my wiki index pages, to export that entire wiki to html files:

:VimwikiAll2HTML

I then SCP the compiled HTML files to my server. Here is an example scp command that you can modify with your own paths:

scp -r ~/vimwiki/website_html/* your_user@your-domain.test:/var/www/website/public_html

For the best deployment experience, I recommend setting up ssh key authentication to your server.

For bonus points I also add a bash / zsh alias to wrap that scp command.

General plugins I use in Neovim

Posted on 21st March 2023 by David Peach

I define a “general plugin” as a plugin that I use regardless of the filetype I’m editing.

These will add extra functionality for enhancing my Neovim experience.

I use Which-key for displaying keybindings as I type them. For example if I press my <leader> key and wait a few milliseconds, it will display all keybindings I have set that begin with my <leader> key.

It will also display any marks and registers I have set, when only pressing ' or @ respectively.

use "folke/which-key.nvim"

Vim-commentary makes it super easy to comment out lines in files using vim motions. So in normal mode you can enter gcc to comment out the current line; or 5gcc to comment out the next 5 lines.

You can also make a visual selection and enter gc to comment out that selected block.

use "tpope/vim-commentary"

Vim-surround provides me with an extra set of abilities on text objects. It lets me add, remove and change surrounding elements.

For example I can place my cursor over a word and enter ysiw" to surround that word with double quotes.

Or I can make a visual selection and press S" to surround that selection with double quotes.

use "tpope/vim-surround"

Vim-unimpaired adds a bunch of extra mappings that tpope had in his own vimrc, which he extracted to a plugin.

They include mappings for the [ and ] keys for previous and next items. For example using [b and ]b moves backwards and forwards through your open buffers. Whilst [q and ]q will move you backwards and forwards respectively through your quickfist list items.

use "tpope/vim-unimpaired"

Passive plugins I use in Neovim

Posted on 21st March 2023 by David Peach

These plugins I use in Neovim are ones I consider “passive”. That is, they just sit there doing their thing in the background to enhance my development experience.

Generally they wont offer extra keybindings or commands I will use day to day.

You can view all the plugins I use in my plugins.lua file in my dotfiles.

Vim-lastplace will remember the last edit position of each file you’re working with and place your cursor there when re-entering.

use "farmergreg/vim-lastplace"

Nvim-autopairs will automatically add closing characters when opening a “pair”, such as {, [ and (. It will then place your cursor between the two.

use "windwp/nvim-autopairs"

Neoscroll makes scrolling smooth in neovim.

use "karb94/neoscroll.nvim"

Vim-pasta will super-charge your pasting in neovim to preserve indents when pasting contents in with “p” and “P“.

use({
  "sickill/vim-pasta",
  config = function()
    vim.g.pasta_disabled_filetypes = { 'fugitive' }
  end,
})

Here I am passing a config function to disable vim-pasta for “fugitive” filetypes. “Fugitive” is in reference to the vim-fugitive plugin that I will explain in another post.

Nvim-colorizer will highlight any colour codes your write out.