This is a series of build scripts that allow you to build various versions of Tsugi servers on various ubuntu instances.
-
Quick Start documentation as to how to bring a self-contained Tsugi server up on Digital Ocean README. These quick start instructions can be adapted to any basic Linux hosting environment.
-
More detailed documentation on how these scripts can be run inside of a bare ubuntu distribution to prepare the instance to run Tsugi ubuntu README.
-
You can also make a multi-virtual-host environment at virtual host README. The vhost process is manual and easily adopted to AWS, Digital Ocean, etc.
-
How build an Amazon EC2 instance of Tsugi for scalable and hardened production using Aurora Serverless, Elasticache, Simple Email Service, and Elastic File System, Load Balancing, and Auto Scaling Groups AMI README. You can use the
tsugi-public-prod
Community AMIs or build your own AMI AMI BUILD README. -
How to build a multi-virtual hosted envionment where you can host several independent domains on one server with Apache Virtual hosting.
-
Dockerfiles that produce docker images that can be used for development, testing, docker swarm, or Kubernetes Docker README. At this point there is documentation and scripts that produce the images - but there is no swarm or kubernetes documentation.
At this point I don't run any production on docker - I just use Amazon. Not because of Tsugi - that is easy - the hard part is all of the external scalable resources Tsugi needs.
If someone with expertise knew how to set up a production grade swarm with backup and autoscaling - and that worked in on locally owned hardware, Google Cloud or Azure I would be happy to entertain adding documentation for that.
Tsugi is just a collection of tools that you need to organize how to expose to your users. Tsugi will be installed at the "/tsugi" path in the web hierarchy at:
/var/www/html/tsugi
You will need to add some documentation that explains what this server is, who can use it - why it is here, etc. It is the web site that wraps Tsugi and allows you to contextualize Tsugi for your purposes. This outer site will be at the "/" path in the browser and reside at:
/var/www/html
In your server.
Here are some "outer site" samples ranging from the simple to the complex:
-
A empty static HMTL shell documentation - https://github.com/tsugiproject/tsugi-parent
-
A Jekyll based site for https://dev1.tsugicloud.org/ - https://github.com/tsugicloud/dev-jekyll
-
A Grav based site for https://www.tsugicloud.org/ - https://github.com/tsugicloud/website
-
A Koseu-based course for https://www.py4e.com/ - https://github.com/csev/py4e
There is even more flexibility as to where you "embed" Tsugi.
The "outer site" for https://www.learnxp.com is actually a Shopify site that then jumps into https://apps.learnxp.com in order to do its Tsugi functions. For https://openochem.org/ooc/ the "outer site" is actually WordPress.
If you are just getting started, just use the tsugi-parent site and then once things are working you can evolve your outer site to make your Ysugi look exactly the way you like it.
In general there are two phases of this process.
-
Preparation - Install all the software and pre-requisites (scripts named include 'prepare')
-
Startup / Configuration - Customize the instance to be yours - which database to use, etc. (script names include 'startup')
Most of the scripts that do the heavy lifting are in the docker
folder.
We have simple Dockerfile descriptions of how to prepare and start Tsugi
servers that lean heavily on the prepare
and startup
scripts.
The ubuntu
and ami
processes actually read the Dockerfile instructions
and extract the steps and data items that Docker would run but
outside of docker.
This might seem a bit obtuse but the goal was not to have more than one copy of any of the instructions to make sure that no matter how we built an instance it could be configured and function correctly.
Because of how Docker COPY commands in Dockerfiles copy from the local
drive into the docker container, it was easier to make Dockerfiles
that work and then pull out the instructions and do them without
docker
in the ubuntu
and ami
processes using the "crude-but-effective"
ubuntu\fake-docker.sh
script.
I would guess some fancy product or service could do this as well - but the approach I have taken is just to depend on ubuntu and shell scripts.
The result of all this is that if you want to make a major change (like
a new version of PHP). Get it working and tested in the docker
folder
and then test the ubuntu version and then the ami version.
Do not get too tricky in the Dockerfiles - or you wll break the
"highly simplified" way that fake-docker.sh
extracts information
from the Dockerfile. In particular fake-docker.sh
will not handle
multi-line docker commands at all.