Unstoppable Web 2.0 | Akash Network - Your Guide to Decentralized Cloud

Repository repository is located here.

This repository serves as a general guide and proof of concept for deploying a full-stack web application onto blockchain-based decentralized infrastructure.

The repository is structured as a monorepo - with infrastructure configuration, application frontend code, and application backend code all in one repository. This is done so that anyone can clone or fork this one repository and begin to experiment with deploying a decentralized web application.

Decentralized infrastructure

DNS	Frontend	Backend	Redundant Database

Handshake	Skynet	Akash	Postgresql

DNS

Frontend

Backend

Redundant

Database

Web application

The demo application deployed in this guide is a note app - with create, read, delete functionality. It serves as a minimal proof of concept for deploying a multi-tier web application to decentralized infrastructure.

Frontend	Backend	Database
React	Python FastAPI	PostgreSQL

The application structure is bootstrapped using this cookiecutter template. The docker image is based on this image.

The application code, technology choices, and configuration used in this repository should provide a general enough base to fork and adapt, or just reference, for future decentralized web application development.

Deployed Demo

A deployed live demo was deployed using the steps in this guide.

Guide

Step 1 - Buy Handshake domain

There are a few simple options for registering a Handshake domain name:

Use a platform like Namebase or a tool like Bob Wallet to purchase a top-level Handshake domain.

Domain purchases using Handshake are achieved through an auction process. It will take 10 days for your auction to finalize. and a winning bid, before you can make use of the domain.
Use gateway.io to purchase a domain under one of gateway’s existing top-level Handshake domains, like .c or .api. Domains purchased in this way are usable immediately.

Step 2 - Push Docker image

Make sure Docker is installed.
CD to the directory where your Dockerfile is. For this repository, it is in application/backend/Dockerfile.
Build and tag docker image docker build -t/: .
Login to dockerhub docker login --username=. You will be prompted for a password.
Push your docker image to dockerhub docker push. Paste this yaml into the tool.

Step 3 - Deploy back-end to Akash

If you haven’t already, take the time to read through the Akash deployment documentation and familiarize yourself with the concepts and steps involved.

You can deploy to Akash using the standard Akash CLI directly but for the purpose of this guide, I am using Tom Beynon’s Akash Deploy UI which is a great tool built on top of the standard CLI. Steps for deploying to Akash using this tool are below.

Start the Akash Deploy UI tool by running the following:

docker run -v ~/.akash-ui:/root/akash -p 3000:3000 --rm -it tombeynon/akash-deploy

Visit http://localhost:3000 to access the tool
Use the tool to create a new wallet. Make sure to record your mnemonic phrase so that you can restore the wallet if needed in the future.
Fund this new wallet with at least 5 AKT (5,000,000 UAKT) by transferring 5 AKT or more to the displayed wallet address.
Create a new certificate. This certificate only needs to be created once and will be used for all future deployments.
Create a new deployment using akash/deploy-sample.yml as a template. The Akash Stack Definition Language (SDL)
Once you have created your deployment, you will start to receive bids. You can view these in the tool.
Chose a bid and Create a lease, then Send manifest.
At this point, the tool will show you your Web URIs which you can use to access your application.
Update application/frontend/.env.production with the Web URIs.

Step 4 - Deploy front-end to Skynet

Deployment of the front-end to Skynet is handled automatically in this repository, using GitHub Actions. When any update to the application/frontend/ directory is pushed to the master branch, the Action workflow will be run automatically by GitHub. This workflow tests, builds, and deploys the static files to Skynet.

This GitHub workflow is based on the excellent write-up by Karol Wypchło and uses his pre-built Skynet deploy Action.

The result of this GitHub Action is a Skynet registry entry that provides a constant point of reference for your DNS record. The Skylink itself changes with each deployment to Skynet and would require you to constantly update your DNS record if referencing it directly.

Step 5 - Configure Handshake domain

Once your GitHub action completes, in the Action log in GitHub under the Deploy to Skynet step, you will see a link to https://siasky.net/skynet/registry with parameters in the URL for publickey and datakey.

You now need to copy those param values and create your skyns:// URL in the format skyns:///. The registry link will remain constant in the future, so this is a one-time manual step.

If using Namebase, you can now configure your Handshake domain to point to your Skynet hosted frontend by adding a TXT record under the Blockchain DNS Records section on your domain manage page. The value of this record is the skyns:/// URL that you just generated above.

This initial DNS configuration can take up to ~6 hours to be synced. For all future code changes pushed to the master branch, the Deploy to Skynet Action will automatically update your Skynet Registry data and you should see the changes reflected very quickly.

Once the DNS change takes effect, your site will be accessible at https://.hns.siasky.net. For example, the demo application for this guide is accessible here.

Step 6 - Setup HTTPS URL for API

Although you can now make requests directly to your Akash URL over HTTP successfully, if you attempt to make these requests from your frontend that is being served over HTTPS, your API call will be blocked due to Mixed Content.

It is likely possible to enabled an HTTPS connection just using Akash deployment configuration but I was unable to find any documentation on this functionality.

The approach used in this guide is to sign up a free Cloudflare account and set up their flexible SSL/TLS encryption mode on a secondary (non-Handshake) domain in order to provide an HTTPS entrypoint that proxies requests to the Akash API server.

Support for HTTPS with Akash is something that I expect to explore further.

Step 7 - Setup Database Redundancy

Backend Redundant Postgresql infrastructure

The backend runs on Akash. Take a moment to review the Akash config. The backend stores the database on Akash. Currently, Akash does not have a persistent storage solution. This solution provides the capability to perform regular backups and database restoration via Skynet. This reference architecture provides the following capability:

Provide a redundant Postgresql configuration. pg-0 is replicated to pg-1
pg-0 - the postgres master starts up. When it first starts up, it will download an encrypted db backup from Skynet.
pg-0 - decrypts the backup and loads into the database.
pg-1 - the postgres standby starts up. and syncs with pg-0.
pgpool container starts up and connects to both pg-0 and pg-1 and automatically determines which one is master.
web container starts up and connects the the db cluster via pgpool.
If pg-1 or pg-0 goes down, pgpool will automatically re-route to the one that is active.
If the entire cluster goes down, the cluster can be re-deployed to Akash. It will then start up from the last backup.

Implementation Notes

The postgres Docker image will execute any .sh, load and .sql, .sql.gz file in /docker-entrypoint-initdb.d
Note that when testing this locally with Docker there are no volumes. This is by design. Everything persistent should be uploaded to the Skynet.
The pg-0 image initially loads pg-0/dbout.sh before the PG cluster is started. It reads in the following env vars: BACKUP_SYKNET_URL and BACKUP_PASS. The encrypted zip file is retrieved from Skynet and decrypts it with BACKUP_PASS. Finally, the DB backup is loaded into the PG cluster.
BACKUP_SKYNET_URL is used to retrieve the latest backup.

Architecture

Based off the Bitnami pgpool project, this approach sets up three containers. pg-0 as a postgresql master, pg-1 as a postgresql secondary ( running as a hot standby), and pg-pool will determine which postgres container to connect to.

Testing Postgresql cluster in a test environment

# run the backend
$docker-compose up -d
# run the front end
$ cd application/frontend/ && npm install && npm start
# Connect to the browser at http://localhost:3000

Shut down pg-0 container

$ docker-compose stop pg-0
## verify that things are working as planned
# Connect to the browser at http://localhost:3000
# Should see all data still there. Try adding values

Step 8 - Setup database backups

Backup your data locally

To retrieve a live encrypted backup of the data, you can run the following command against the API:

# For testing locally with docker-compose
curl -v http://localhost/getsnapshot > dbout.zip
# For testing against AKASH deployment
# Obtain $AkashHost and $AkashPort from akash deployment (Consult Akash [documentation](https://akash.network/docs/guides/deploy) on how to obtain this)
curl -v http://$AkashHost/getsnapshot:$AkashPort > dbout.zip
#
# Backup against live demo
#
curl https://unstoppablestack.coffeeroaster.me/getsnapshot > dbout.zip

Make sure to update the $BACKUP_PASS env variable in docker-compose.yml, akash/deploy-sample.yml. You will use the password to decrypt the zip file. You can use $BACKUP_PASS env var in akash/deploy-sample.yml to decrypt and examine the backup.

Upload local backup to Skynet

Upload the backup to Skynet with skynet-cli and store URL as BACKUP_SKYNET_URL

$ npm install -g skynet-cli
## send it to Skynet!
$ skynet-cli dbout.zip
## Take special note of the URL. This value will be used as **BACKUP_SKYNET_URL**

Tell Unstoppable Stack to load it up next time.

Update docker-compose.yml (for testing locally) and akash/deploy-sample.yml environment variables section and update BACKUP_SKYNET_URL with the value from the previous step.

Shutdown and destroy your entire deployment

# For test environment
$docker-compose down -v
# For Akash deployment
(Consult Akash [documentation](https://akash.network/docs/guides/deploy)

Start it back up (development env)

$docker-compose up -d
# run the front end
$ cd application/frontend/ && npm start
# Connect to the browser at http://localhost:3000

You should now see the values from the database that you backed up.

Start it back up (production env)

Update akash/deploy-sample.yml with updated ENV variables (namely the BACKUP_SKYNET_URL )
Redeploy backend to Akash following step 3 in this guide.
If using Handshake (Step 5) Use the same Front END URL.

Database replication next steps

Investigate NuCypher for better key management
Look into the SkyNet registry for a simpler way to provide backup / restoration. Right now, each provision to Akash requires updating a new SkyLink URL. Using a Skylink Registry will allow the container to use a consistent skyns URL to retrieve the latest backup for ease of use.

Run demo application locally

To run the application locally, you can follow the steps below.

Stand up FastAPI and PostgreSQL backend
```
cd backend
docker-compose up --build
```
In a separate terminal, stand up the React frontend
```
cd frontend
npm install
npm run start
```

You can now visit http://localhost:3000 in the browser to access the UI. For testing the API, requests can be made to port 80, Example: GET http://localhost:80/api/v1/notes

Sources and resources

Akash

Skynet

Detailed Skynet deployment guide