by Michael Cropper | Feb 8, 2023 | Developer |
Note: This blog post is entirely generated by AI with the prompt “Write a blog post titled “What is Docker” which is around 2000 words in length and include sub-headings to make the content easy to read. ” ….. Leave a comment on what you think…..
Docker is an open-source platform that automates the deployment, scaling, and management of applications inside containers. In simple terms, it provides a way for developers to package their applications and dependencies into a container, which can then run consistently on any system that has Docker installed.
Containers vs Virtual Machines
One of the key differences between Docker and traditional virtualization methods like Virtual Machines (VMs) is that containers share the host system’s operating system (OS) kernel, while VMs run on a full copy of the host’s OS. This makes containers much lighter and more efficient than VMs, as they don’t require the same amount of system resources or disk space.
Another important difference is that containers are isolated from one another, but share the host’s OS. This means that each container runs its own application and dependencies, but they are all running on the same underlying system. In contrast, VMs are completely isolated from one another and run their own OS, which can lead to compatibility issues between different systems.
Why use Docker?
Docker provides several benefits over traditional virtualization methods:
- Portability: Docker containers can run on any system that has Docker installed, making it easy to move applications between development, testing, and production environments.
- Scalability: Docker makes it easy to scale applications by adding or removing containers as needed, without having to reconfigure the underlying system.
- Isolation: Each Docker container runs in its own isolated environment, which helps to ensure that applications don’t interfere with one another.
- Consistency: Docker ensures that applications run consistently across different systems, as they are packaged with all of their dependencies.
- Ease of use: Docker provides a simple and straightforward way to manage containers, making it easy for developers to focus on writing code instead of dealing with complex infrastructure.
Docker Architecture
Docker uses a client-server architecture, with the Docker client communicating with the Docker daemon, which is responsible for building, running, and managing containers.
The Docker client and daemon can run on the same system, or they can communicate over a network. The Docker client sends commands to the Docker daemon, which then performs the requested actions.
Docker Hub
Docker Hub is a public repository for Docker images, which are pre-built packages that include everything needed to run a specific application or service.
Developers can use images from Docker Hub to quickly get started with building and deploying their applications, or they can build their own images and upload them to Docker Hub for others to use.
Dockerfile
A Dockerfile is a script that contains instructions for building a Docker image. It specifies the base image to use, the application and dependencies to include, and any configuration information that the image needs.
Once the Dockerfile has been created, developers can use the Docker build command to build the image and create a container from it.
Getting Started with Docker
Getting started with Docker is straightforward, as it only requires a few basic steps:
- Install Docker: To start using Docker, you’ll need to install it on your system. Docker is available for a wide range of platforms, including Windows, macOS, and various Linux distributions.
- Download a Docker image: You can either download a pre-built image from Docker Hub or build your own image using a Dockerfile.
- Run a container: Once you have an image, you can use the Docker run command to create a container from it and start it running.
Conclusion
Docker is a powerful platform that makes it easy to
Commentary on my opinion of the quality of the ChatGPT generated blog post
Honestly….. Not bad. Technically accurate too. But….. It’s only ~500 words in length, the content is very surface level and nothing really makes sense unless you know this stuff already which kind of makes it a bit pointless. The challenge here is that these topics are complex, the documentation is complex, there are lots of moving parts and actually the whole benefit to writing blog posts is not to write a high level overview of the topic, but to help readers truly understand so they can anchor their existing knowledge towards a new topic so they can get up to speed quickly.
Very gimmicky, entertaining but when it comes to something I would deem as high quality, it’s a long way off. I do like the headings it introduced though (formatted manually), but in reality, nothing that you can’t easily create the structure of a blog post with a tiny bit of research on the topic before jumping straight in.
by Michael Cropper | Sep 26, 2022 | Developer |
There are many ways to skin this cat, so this blog post is going to look at some of the ways that you can help simplify Apache Virtual Host configuration on Linux by breaking things down into manageable self-contained chunks with bounded contexts, aka. all around a domain name which is what 99% of people will be aiming to do.
The difficulty with the official documentation for Apache Virtual Hosts is that it provides many different examples to follow, but gives very little contextual information around use case and instead goes very deep into the art of the possible without guiding you to where you should look. And I guess that’s fine to an extent for official documentation, but it’s also not very useful at the same time as different configurations require different levels of complexity.
So in this blog post we’re going to focus on the common setup for what the majority of use cases for Apache Virtual Hosts are and how you should probably set this up to make your life easy.
System Architecture
Firstly, it’s important to understand the situation we find ourselves in. Below is a basic hierarchy of layers where this sits. Naturally this is going to differ in reality for most situations, but conceptually in most cases, this tends to be the setup that is ultimately in place in one way or another with a few nuances along the way.
- Virtual Machine
- Operating
- Firewall
- Apache / HTTPd
- Default
- Default Document Root
- Default httpd.conf
- example.com
- example.com Document Root
- example.com httpd.conf
- another-example.com
- another-example.com Document Root
- another-example.com httpd.conf
Operating System Firewall
Keep an eye on this, this is one of the steps that causes a lot of confusion. People often start working at the Apache HTTPd layer, yet haven’t opened the correct ports on the Operating System Firewall. By default many modern operating systems are out of the box configured as a deny-all setup so are likely to block traffic before it’s even reached your Apache HTTPd installation, usually the default open ports are only port 22 for SSH access.
Remember, the layers of your system are essential to understand in detail to make your life easier debugging problems.
The operating system firewall is beyond the scope of this blog post, so we’ll cover this off in a subsequent blog post, but always keep this in mind.
Apache HTTPd Installation
When we install Apache HTTPd via tools such as Yum it will create folders, files and scripts throughout the system. The important ones being;
- /etc/httpd/ – For the configuration aspects of Apache HTTPd
- /var/www/html – For the files that need to be served to users accessing web applications
This is the basics.
For single website hosting this can be more than fine to work with out of the box with zero additional configuration. In reality, most Apache HTTPd installations are hosting many websites, essentially acting as a mini-reverse proxy inside the virtual machine to host multiple websites on the same virtual machine and ultimately the same underlying infrastructure.
Document Roots
Ok, so getting back to basics here. What is a Document Root? In simple terms, this is the home directory for a specific configuration. To put this into context, most people run Windows computers for personal and corporate use. In this example, your “Document Root” is the equivalent of “My Documents”.
So when User 1 logs into a Windows computer they go to “My Documents” and see their own files.
When User 2 logs into a Windows computer they go to “My Documents” and see their own files.
The concept of a Document Root is essentially the under-the-hood configuration that makes this possible.
So in the same way when we are hosting multiple websites this essentially allows us to direct example-one.com to /example-one/index.html, and example-two.com to /example-two/index.html.
Conceptually this is what we’re working with, despite the terminology and underlying configurations being fairly complex using lots of terms that aren’t familiar to 99.9% of us and having to search the web to gather these pieces of information.
So, let’s dig deeper…
Default Document Root – /var/www/html
The default Document Root that is created on Apache HTTPd installation usually lives at /var/www/html. You can confirm this on first installation if you setup Apache HTTPd and then try to access the IP address of the virtual machine. If you have no advanced configuration sitting in the way, you’ll see a successful page that shows a message confirming that Apache HTTPd has been successfully installed. Awesome
Domain Specific Document Root – /var/www/example.com/public_html
But, what if you want to host multiple websites using Apache HTTPd, you need to segment this into a separate Document Root for ease of management. Essentially two separate folders for two different domains;
- example-one.com > /var/www/example-one.com/public_html/index.html
- example-two.com > /var/www/example-two.com/public_html/index.html
Create a basic “Hello World” index.html file for Example One and Example Two so you can easily identify the two and you’re good to go.
You’ll notice that if you check these locations after initial setup that they don’t exist. You need to create these folders and files using Linux commands such as mkdir and nano index.html if you’re not too familiar with these commands. These commands are the equivalent on Windows of right clicking and selecting the Create Folder menu item and the Create File item respectively.
Here we have now created two separate folders, aka. Document Roots, that we can then use to configure the Apache HTTPd Configuration.
Apache HTTPd Configuration Files
Now that we’ve covered off the Document Roots which is where your files live, the next step is to cover off how to configure Apache HTTPd Virtual Hosts properly to ensure your hostnames can root to the correct Document Root.
Default HTTPd.conf – /etc/httpd/conf/httpd.conf
The default configuration file that comes out of the box with Apache HTTPd is located at /etc/httpd/conf/httpd.conf. This is the global single file that rules them all. What is important though is that this file can be added to and also extended or overridden. This is where things get interesting.
There are many way to skin a cat, and this is one example. Fundamentally there are two ways to extend the Apache HTTPd configuration, one of them is by extending this main configuration file. The other is what we’re going to cover off next.
While is it possible to extend the main Apache httpd.conf file, it’s generally bad practice to do so when you are configuring virtual hosts. Mainly because it makes things significantly more difficult to manage and maintain.
If you do want to add Virtual Hosts configuration to the primary httpd.conf file then you simply add these details;
[XML]
ServerName mk1.contrado.cloud
DocumentRoot /var/www/mk1.contrado.cloud/public_html
ErrorLog /etc/httpd/logs/mk1.contrado.cloud-error.log
CustomLog /etc/httpd/logs/mk1.contrado.cloud-access.log combined
[/XML]
Ultimately though, whatever you do in your httpd.conf file, can be over ruled by a separate domain specific configuration file. This is what we’re going to cover next.
It’s not best practice to add Virtual Hosts to your httpd.conf file as it keeps every single configuration bound to a single file which can cause problems with dependencies over time.
Domain Specific HTTPd.conf – /etc/httpd/conf.d/example.com.conf
So we’ve talked previously about adding the specific configuration to a separate Apache HTTPd configuration file which is what we’re going to look at next. Apache is a well-established and advanced piece of software which understands parent/child relationships. And this is the case with .conf files.
As we’ve seen earlier around how the core httpd.conf file is located and how the override is configured, let’s look at this;
- /etc/httpd/conf/httpd.conf – Default Apache HTTPD configuration file
- /etc/httpd/conf.d/example.com.conf – Domain specific Apache HTTPD configuration file
It’s not super complex in practice, while under the hood it clearly is. Ultimately it’s a simple scenario that if there is a domain specific configuration file then this takes priority over the general configuration file.
And all this is managed through the configuration of the Apache HTTPd Virtual Hosts syntax.
To manage this effectively, simply create a file using the command;
nano /etc/httpd/conf.d/example.com.conf
Then add in the exact same configuration details that we’ve outlined earlier;
[XML]
ServerName mk1.contrado.cloud
DocumentRoot /var/www/mk1.contrado.cloud/public_html
ErrorLog /etc/httpd/logs/mk1.contrado.cloud-error.log
CustomLog /etc/httpd/logs/mk1.contrado.cloud-access.log combined
[/XML]
What this essentially means is that Apache will take into account these additional configuration files and use them to override the default. This is all accomplished via the primary http.conf file mentioned earlier with the out of the box configuration of;
IncludeOptional conf.d/*.conf
There is nothing specific to configure to make sure this is working.
Virtual Machine or Apache Level SSL Configuration via Let’s Encrypt
There are many way to ‘do’ SSL. In most legacy on-prem setups, you’ll tend to find that SSL is offloaded at the primary on-prem firewall and traffic is unencrypted from that point forward as traffic travels to the correct virtual machine with Apache or Nginx is running etc.
I’m not saying this is bad practice per say, because it works, but yeah, it’s often a very error prone setup with all eggs in one basket and all and often causes significant bottle necks as the primary firewall tends to be heavily restricted and any change is virtually impossible to make without weeks of discussions, planning, forms and more – for what is ultimately a 5 minute job in most cases to implement.
So, for the purpose of this blog post and to provide a full end to end setup, we’re going to assume that you’re using modern Let’s Encrypt technologies to generate your SSL certificates on the fly for free every few months automatically, from the virtual machine where your application lives.
If you need more information about Let’s Encrypt then we have covered off several blog posts on this topic over the years so search around the site, some of the core ones being;
Why it’s important to mention this is because of the previous setups we’ve gone through.
Security Considerations
We’ve not really covered security considerations for any of the above in this blog post as this is a significantly more in-depth discussion which has many nuances based on every individual setup, governance and controls.
For the purpose of this blog post, this has been to look at how to host multiple websites behind a properly configured Apache HTTPd Virtual Host setup for applications that you own, control and can trust 100%.
Things get significantly more complex for other applications when there are multi-tenancy considerations which is ultimately where software such as cPanel and WHM come into play, but that’s a topic for another time.
Summary
Hopefully this blog post has provided you with some insight on how to configure Apache Virtual Hosts using Document Roots and HTTPd.conf files and separate domain specific HTTPd configuration files to help make your life easy to manage and keep configurations segmented.
As with everything Apache and HTTPd related, everything is going to be specific to your individual use case so please treat this blog post as guidance not a rule. Take a look at your own set up and assess how any of this information may or may not apply to your specific setup.
Hope this helps.
by Michael Cropper | Sep 19, 2022 | Developer |
I’ve found that this topic is quite an undocumented one online and lots of assumptions tend to be made. The majority of content online under this topic that I’ve come across tends to direct you down the route of HA Proxy, which can be fine with very specific setups. But, the minute you want to start to do anything more complex than the basics, HA Proxy soon becomes limiting.
Many websites these days have multiple ports open for specific use cases. Take for example one of the most common web hosting platforms cPanel, this requires many open inbound ports at the firewall layer and in other scenarios we want to give the control to the virtual machine’s operating system the power to decide what ports to allow in without having to configure the firewall every time since they may not have access to the firewall.
We did a blog post recently for How to Setup HA Proxy on pfSense to Host Multiple Websites, which is worth a read to understand the differences for what we are going to cover off here. The core difference here is that with HA Proxy, you have to be explicit when configuring it which ties the settings against a single port, which often is too limiting for many applications.
To get the maximum flexibility you need multiple public IP addresses. This allows you to configure things in any way that you require. So this is what we’re going to cover off in this blog post.
System Architecture
To get a high level view of a setup like this we have the basic components that are outlined below. With a modem, connected to a pfSense firewall which has virtual IP addresses, port forwarding, static IP addresses for virtual machines, VLANs for security configured, then connected to a server with a virtual machine running on it.
In reality, there are often a few bits more along the way and significantly more complex in real world environments, but fundamentally, this is the basic architecture for how all this plugs together.
Purchase Additional Public IP Addresses from your ISP
Firstly, you need to purchase a block of IP addresses from your internet provider. What happens when you do this is that any traffic from those public IP address ranges will ultimately route through to your pfSense firewall so you can then determine what happens to that traffic next.
We’ve covered off a blog post a while ago which helps you to Understand Network Private Address Ranges and looks at the difference between public IP address ranges and private IP address ranges. So take a look at that blog post if you aren’t sure of the difference.
Ok, so now you’ve got your additional public IP addresses.
pfSense Firewall Virtual IP Address Configuration
Now it’s time to tell your pfSense firewall about these IP addresses so it knows how to handle the traffic that is coming in. The first step is to setup Virtual IP addresses within pfSense. To do this, simple navigate to Firewall > Virtual IPs.
There is nothing particularly complex for settings these up. Simple setup the type as an IP Alias, the interface would be the WAN, the Addresses would be one of the IP addresses with the /32 CIDR range (aka. single IP address), then give it a friendly name and you’re done.
Configure NAT Port Forwarding Rules in pfSense
Now that we’ve got the Virtual IPs configured, it’s time to configure the port forwarding rules so the traffic gets to where it needs to for that public IP address. To get started with this, simply navigate to Firewall > NAT > Port Forward. Then add in some rules.
As you can see in the screenshot below, the settings are rather basic, with the rule being applied to the WAN interface, for the TCP protocol where the Destination Address is the Public IP Address that you added as a Virtual IP Address previously, then forwarding traffic to the static IP address of the virtual machine whenever traffic is received on any port.
When you do this, pfSense will automatically add a Firewall Rule on the WAN interface.
pfSense Static IPs and VLANs
We’re not going to cover this off within this blog post as this is a separate topic and one that is going to be dependent on your specific setup. The static IP addresses are important to ensure your Virtual Machines always get the same IP address every time they are rebooted so that the firewall rules are still accurate.
Likewise, the VLANs and IP ranges are extremely important from a security perspective to ensure that any inbound traffic is securely segmented from your sensitive internal systems and/or other separate public IP ranges that need to be kept separate from other ones.
Server and Virtual Machines
To keep this bit relatively short and simple, if you have configured all of the above correctly, the traffic that comes into your WAN from specific IP addresses then this will flow through to your virtual machine that you have setup.
So for example, if you have the DNS A Record setup for mk1.contrado.cloud as 62.3.66.193, then this traffic will reach the specific virtual machine where the operating system firewall will then control what traffic it will accept in and what traffic it will deny.
This setup gives you the complete control of the traffic without having to continually play with the pfSense firewall rules.
Summary
Hopefully the above is a handy guide for how to configure pfSense with multiple public IP addresses then use NAT so that you can host multiple websites using multiple ports with minimal configuration within the pfSense firewall.
This is a complex topic, and one that is going to be very different in every single use case. Hopefully this blog post had helped fill in a few gaps in knowledge to get you pointed in the right direction.
by Michael Cropper | Sep 10, 2022 | Developer |
I do this so infrequently, this blog post is mainly as a reminder to myself as a lot of content online around this topic isn’t quite as good as it could be. Anyhow. It’s important to remember how to easily mount a disk in a Linux environment so that it’s possible to control how and where you manage your data storage for your specific use case.
Context and Background
Firstly, understand your own infrastructure you’re working with – in depth. This specific blog post is based on a lab setup to provide guidance, not concrete solutions for production challenges.
The setup in this physical scenario is;
- Type 1 Hypervisor – XCP-ng
- HostOS HDDs/SSDs
- GuestOS HDDs/SSDs
- Virtual Machine running Ubuntu
- Default HDD/SSD Configured
And the scenario that we’re aiming to achieve is an additional disk mount against the ‘Virtual Machine running Ubuntu’. The reason being is that gives us a physically different location to store data against.
In this specific example, we have;
- RAID {X} Array for Type 1 Hypervisor HostOS
- RAID {X} Array for GuestOS Virtual Machines
- RAID {X} Array for Backups
All the above being on the same physical machine.
Partitioning and Formatting
Ensure your disks are partitioned and formatted so they are ready to be used if they are band new. Obviously don’t do this if you have data on the disk that you need. Again, this all comes down to your individual setup, but for ease, make sure your physical disks have been partitioned in the way that you require and have been formatted accordingly.
Attach the Disk to the Virtual Machine
Within XCP-ng Centre, make sure the virtual machine has the disk attached. Interestingly when you do this, this doesn’t mean that it is automatically available to be used interestingly so you can’t navigate to the new disk via the Linux command line, since that is installed on the disk where you installed Linux.
So next, we have to mount the new disk to the Linux installation on the main disk so that it can access that additional storage location.
Find Disk that can be Mounted
Firstly you need to SSH into your Linux virtual machine. To find the available disks, run the command;
fdisk -l
The fdisk utility is designed to manage partitions on hard drives which allows you to create, delete and partition disks for your specific use case. The L flag is for Listing the drives and their partitions. This command is extremely valuable to check if the disk has partitions or not. As you can see below, on the disk in question it has no partitions.
You’ll notice that the top disk in the above screenshot has a lot more information than the second disk which is showing there are no partitions on the disk. So next, we need to get those created.
Create Partition on Disk
As we can see in the previous image, the disk is at /dev/xcdb. So we need to use the fdisk command again to create a partition.
fdisk /dev/xcdb
Here we can see we’ve used a couple of commands and configuration options when doing this which are primarily;
- n = New Partition
- p = Primary Partition
- 1 = Partition Number
- +1TB = Size of Partition
After following this through, we’ve now got a partition successfully created.
Next we need to write the changes to the partition table, aka. save changes, so that it is ready to be used with the w command;
If you run the command fdisk –l again, you’ll see the partition that has been created which is setup as /dev/xvdb1.
Next we need to create the file system using the command;
mkfs .ext4 /dev/xvdb1
Mount the Disk to Linux
Now we’ve got the disk partitioned and with a file system on, it’s time to mount the disk to Linux so that we can then use it. This part is straight forward now that the ground work has been done to the disk. So run the command;
mount /dev/xvdb1 /mnt/backups
Obviously change the disk name and the location folder to your specific use case.
Ensure the Mount Persists after Reboot
Finally, make sure the mount persists on reboot. To do this we use the fstab utility. To configure this, run the command nano /etc/fstab then edit the file as follows;
Save and exit the file. Job done. You can now use that new file system however you want as it is now accessible.
by Michael Cropper | Aug 25, 2022 | Developer |
In this blog post we’re going to look at how you can host multiple websites on separate virtual machines that sit behind a pfSense firewall with HA Proxy installed. This is quite an involved topic and one that is going to vary greatly based on your physical hardware and virtualisation environments, so keep this in mind if you’re trying to follow this.
For reference, the underlying physical hardware behind the below setup is nothing more than;
- Physical Modem
- Physical pfSense Firewall
- Physical Managed Switch
- Physical Server running XCP-ng
- Multiple Virtual Servers running inside XCP-ng
Configure pfSense System > Advanced > Admin Access
By default the pfSense WebGUI runs over port 80 and 443. What this means is that if you want to host a website behind pfSense then you need to re-configure this since your websites are going to be running over either HTTP or HTTPS.
To do this, simply change the TCP Port to an available port and disable the webConfigurator Redirect Rule as can be seen below;
Install HA Proxy via pfSense Package Manager
The first place to get started is to install the latest version of HA Proxy via the pfSense package manager by navigating through to System > Package Manager > Available Packages. Simply install the package and you’ll see this software now available for you to manage and configure.
Whenever you install packages within pfSense you’ll notice different menu items start to appear where you can configure the package and/or view the current use of it. The core menu item for configuring HA Proxy is under Services > HA Proxy;
HA Proxy Terminology and Architecture
Before we jump into the configuration for how to setup HA Proxy to host multiple websites, it’s first important to make sure we understand the overall architecture and how things fit together as the terminology and user interface often isn’t as clear as it could be for someone brand new to this. By understanding this, the configuration we’re going to go through below should hopefully make much more sense why we are configuring the things we are and ultimately will help you work with HA Proxy better in the future.
The below architecture is what we’re going to talk through in this guide for how to setup HA proxy on pfSense to allow you to host multiple websites behind a single IP address using a shared front end;
The core bits of terminology in HA Proxy that may not be as familiar to those coming from a web development background are the concepts of Front Ends and Back Ends, since these mean significantly different things in the world of web development. In web development, Front End generally refers to technologies such as HTML, CSS, JavaScript. And Back End tends to refer to server side technologies such as Java and C# to name a few for the purpose of illustration.
Those more familiar with scalable infrastructure in the web development world will be more used to the terms from cloud platforms such as AWS where there is the use of Application Load Balancers and Target Groups.
How this translates into HA Proxy terminology is that an Application Load Balancer in AWS terminology is essentially a HA Proxy Front End, and the AWS Target Group is equivalent to HA Proxy’s Back End. i.e. it is a configuration space that ultimately points to an upstream server somewhere.
And for those of you familiar with Nginx and using this as a reverse proxy, then the Back End in HA Proxy terminology tends to align closely with Upstream Server in Nginx terminology. All fun and games when every piece of technology calls essentially the same thing something completely different.
Anyhow, hopefully that’s helped to clarify a few bits of the terminology related to HA Proxy and how it loosely aligns with terminology that you may be more familiar with if you’re coming from a cloud native or web development background.
Configure HA Proxy Settings
So far all we’ve done is install the basic package with nothing configured and nothing turned on. Now we understand the architecture as explained earlier, it’s time to jump into some of the settings. First we’ll get started with the overall HA Proxy Settings.
Turn on HA Proxy
Firstly, let’s make sure HA Proxy is turned on;
Turn on HA Proxy Statistics
To do this simply configure a relevant port on the settings page;
What this gives you is an extremely handy report that you can view from Status > HA Proxy Stats page which allows you to view lots of valuable information so you can understand how many sessions are hitting your Front Ends and Back Ends in HA Proxy once these have been set up a little later. This is significantly valuable when trying to debug things if they aren/aren’t working. Jumping ahead a little, but this is worth turning on while you are in this area of the configuration’
Configure the Max SSL Diffie-Hellman Size
Don’t ask me what this is, I genuinely don’t have a clue at this point in time of writing. An error message showed up in HA Proxy/pfSense when configuring future steps below which needed this setting up. So hey, do what it recommends – Seems odd that it wouldn’t automatically populate with best practice and requires a manual configuration but hey, probably one for a future contribution to the open source project by someone;
Summary of Settings Configuration
You’ll notice that we’ve skipped over a lot of the settings that are available to configure and for genuine reason. And that is that you just don’t need the majority of them to get a basic setup in place to start testing this stuff.
Sure, in a production environment you’re going to need to get finer tuning in place, but while you’re learning this then it’s more important to get a basic setup in place that you can play with and improve from there.
Setup Your Virtual Machines
Ok, so for the purpose of this blog post I’m going to assume you already have got some virtual machines set up and working in the way you want with the appropriate security in place to ensure that they are securely segmented for public access.
In our example, we’ve got the three virtual machines that are powering;
- contrado.cloud
- contrado.cloud
- contrado.cloud
In this example they have simply been configured to run Apache / httpd, configured to ensure httpd runs on startup, configured to allow inbound http traffic and added a basic index.html page in /var/www/html/index.html so that we can easily see which server we are on. Beyond that, everything else is out of the box on these virtual machines.
Configure HA Proxy Back Ends
This seems really counter intuitive to do this first, but it’s essential to do this first. This is what we’re going to achieve with this example;
And we’ll dig into a single one as an example, the others are the same though with no differences beyond IP addresses. So configure your first Back End in HA Proxy;
You’ll notice that the IP address is on the 192.168.1.0/24 LAN network which is clearly insecure, as is the Port 80 for insecure HTTP traffic for any real world production environment. But at least this gives an example for how to get this set up as a starting point.
There is nothing really to configure as a basic setup beyond the above. You can go configuring the Health Checking if you like, which is essential in production environments with High Availability Front Ends and disposable back end servers and infrastructure as code, but that’s well beyond the scope of this blog post. So either disable this by setting the option to None or configure as per your needs.
Rinse and repeat for your other virtual machines that you want to expose to the internet.
Configure HA Proxy Shared Front End
Note, if you are only hosting a single website then you can use a basic Front End rather than a shared front end, the configuration steps are very similar. You actually don’t even need to use HA Proxy if you are only hosting a single website as you can use basic Port Forwarding in pfSense as we’ve previously written a basic guide for on how to host a single website behind a pfSense firewall which you should check out if this situation applies to you.
Now that you have all of your Back End HA Proxy configured, it’s time to apply a Front End to handle traffic from the internet in a way that suits your needs. As this blog post is about hosting multiple websites behind pfSense using HA Proxy, the reality is that this is going to be primarily determined by the hostname of the incoming request, i.e. http://helloworld.contrado.cloud or http://hellouniverse.contrado.cloud or http://mk1.contrado.cloud etc.
For context, HA Proxy is really designed for traditional network infrastructure where you essentially don’t expose anything to the internet and anything you do you do minimally and with great caution. In the world of cloud and web development, well, this doesn’t quite meet expectations. There will be future blog posts about other supporting technologies in due course which significantly expand the capabilities and functionality that is explained in this blog post. This is just the basics for now.
What we’re aiming to achieve with a shared Front End in HA Proxy is the following;
What you’ll notice is that this is purely for Port 80. This comes back to the point raised earlier that the default behaviour for pfSense and HA Proxy in general is to disable all and specifically permit inbound rules. In reality, any website that is being hosted is highly likely to require multiple ports to be opened to be managed remotely with ease such as HTTP, HTTPS, SSH, SFTP and many more. That all comes down to your specific use case though, so this blog post is simply designed to get you going in the right direction, not give you all the answers to your specific use case.
Jumping into the specific details of the above to get clarity on how this is configured. Simply give the Front End a Name, Description, make sure it’s Active and listening on the WAN on port 80 then set the Type to ‘http / https (offloading)’ as can be seen in the image below;
The next bit of configuration on your Front End is to configure what rules you need in place to allow your front end to talk to multiple back ends depending on your setup. For this blog post we’ve simply got the 3x virtual machines powering the HelloWorld, HelloUniverse and MK1 sub-domains to see how this works.
Firstly, configure your Access Control Lists which in this example simply gives you a way to map a hostname to a friendly name.
Next, configure the Actions by mapping how each of the Access Control List friendly names maps to a Back End in HA Proxy. The user interface in here isn’t the best if I’m honest. These things should be selectable dropdowns rather than free text field as it causes confusion, but hey.
Finally, select which is the default Back End so that HA Proxy knows where to send traffic when it doesn’t know what to do with it.
And that’s it for configuring HA Proxy. There are so many different settings throughout all those different screens we’ve looked at, but ultimately to get a very basic implementation set up so you can have a baseline to play with and improve up, this is all there is to it.
Configure pfSense Firewall Rules
Last step is to ensure you have a firewall rule on your WAN interface so that inbound traffic to the WAN from the internet can talk to the firewall and hence HA Proxy so that HA Proxy can then direct the inbound traffic to the correct destination based on what you have configured. Note that the top two rules in the screenshot below are out of the box pfSense rules to protect your network.
Check Everything is Working
Now simply navigate to your Sub-Domains and check everything is loading correctly as you expect.
Awesome!
Final Considerations
This blog post is an extremely basic setup looking at only a single port. In reality, websites are complex, particularly when hosting a wide variety of services and even more complicated when you start to look at how SSL certificates are managed and where etc. Things start to get significantly more complex from this point.
Also, don’t forget to configure things such as static IP addresses for your virtual machines, ensure Apache or Nginx is starting on boot and your firewalls on your virtual machines are allowing inbound traffic on the ports you require open as most operating systems these days default to only allowing SSH traffic in by default.
There are a lot of jumps that your connection is going through as you work through the steps, so if something isn’t working for you, be extremely methodical to trace this through step by step and confirm if the traffic is hitting the next hop or not. This is often half the battle getting these things set up properly.
Summary
Hopefully by now, the pfSense and HA Proxy architecture diagram from earlier in this blog post is making sense what we’ve configured and how we’ve configured it. All of this should help get you going in the right direction to then further customise this to your specific use case.
As always, keep security in mind at all times. What has been outlined above is an extremely insecure setup on a flat network and running over HTTP. All of the IPs and custom configurations have naturally been changed for the purpose of this example tutorial as it is designed to get you understanding the basics. We’ve other blog posts on the site for how to configure things such as managed switches and pfSense so take a look around for other helpful content.
Hey, if you’re lucky and you test one of the hostnames that have been mentioned on this site, you may even see the websites up and running. The underlying infrastructure behind this is very much a part timer at the moment, only turned on when I’m having a play around.
by Michael Cropper | Aug 20, 2022 | Developer |
Ok, before start let’s acknowledge that this blog post is an extremely complex topic. That being said, let’s get the foundational understanding in place for the readers before we jump into the complexities.
What is Xen
Xen is simply the type 1 Hypervisor that runs on the bare metal, aka. Xen Project. This is an open source project that has had an enormous amount of contributions from the big players including Citrix, Arm, SuSE, AWS, Alibaba, CentOS, Debian, Oracle, RedHat, Ubuntu, XCP-ng and many more.
What is Citrix Xen Centre
Citrix Xen Centre is the software that is installed on a separate physical machine that is used to control Xen. Such as spinning up new virtual machines, configuring networks and managing storage.
What is XCP-ng Centre
XCP-ng Centre is the software that is installed on a separate physical machine that is used to control Xen. Such as spinning up new virtual machines, configuring networks and managing storage. This software is a fork of Citrix Xen Centre which was created during the transition to fully open source where Citrix released the reigns and control on the project.
What is Xen Orchestra (XO)
Xen Orchestra (XO) is the open source technology for managing Xen infrastructure via a user friendly web based user interface to allow the general user to control everything than Xen can do under the hood. Whereas Citrix Xen Centre and XCP-ng Centre are installed applications on a separate physical device, Xen Orchestra can be run within a host itself.
What is really interesting about Xen Orchestra is that you can actually install it within a virtual machine on a physical server where XCP-ng Server installed on. You may think this is counter intuitive and, well, it kind of is because from a hierarchical perspective, the virtual machine is effectively controlling the parent server for spinning up new virtual machines and more. This is pretty cool, but as always, keep security in mind and plan these things well if you are doing this.
If you’re looking for a guide on how to install Xen Orchestra (XO) from Sources then checkout the link. This guide will talk you through the finer details of how to get Xen Orchestra setup with ease to help you manage your virtual machines with ease and without limitations.
What is Xen Orchestra Appliance (XOA)
Xen Orchestra Appliance (XOA) is the commercial implementation of the open source Xen Orchestra (XO). This is managed by Vates, a French company who is leading the development of the open source project to maturity. Hat’s off to them and Olivier Lambert as the lead.
Architecture
So what does all this mean? Well, essentially that Xen / Xen Server / XCP-ng Server is installed on the physical machine and that this can be controlled / manged using either Citrix Xen Centre, XCP-ng Centre, Xen Orchestra (XO) or Xen Orchestra Appliance (XOA). Simple right?
Think of the control systems (Citrix Xen Centre, XCP-ng Centre, Xen Orchestra and Xen Orchestra Appliance) as input/control devices. In much the same way that a keyboard and mouse act, they are both input devices into a computer where you can achieve the same goals in one way or another.
Naturally for this basic setup you’ll notice that everything is on a flat network, aka. no securely segmented networks. This is rarely a good idea in a production environment, but at least this basic setup helps you to get the base level understanding for how things plug together under the hood and helps to align terminology and understanding.
What is not always commonly understood is that Citrix Xen Centre can control both Citrix Xen Server and XCP-ng Server. Likewise, XCP-ng Centre can control both Citrix Xen Server and XCP-ng Server as XCP-ng was forked from the Citrix setup when it was open sourced. Quite handy to know.
Synchronisation
What is important to understand with all the above in mind is that it is not the tool that is mastering the configuration data, but the tools are there to control the mastering of the configuration data. What this means is that if you have multiple of the control devices installed on your machine then you’ll notice that they all update whenever one of them updates the configuration, as they are all reading/controlling the same physical device.
Summary
Given the above descriptions and architectural diagrams, hopefully it is now clear how you can manage Xen using a multitude of tools based on your individual user preference. What is hopefully clear is that this is a complex topic and one that often ends in a question being answered as “it depends”.
Hopefully this blog post has helped to align understanding of how the different dots connect together to help you in your understanding of what you need to achieve for your specific use case.
