What do you use to deploy and update your infrastructure?

I use a herokuish like workflow where I git push to a remote and it uses buildpacks to build and deploy my app to a kubernetes cluster.

This is the premise behind my product I am building to compete with Heroku, https://primcloud.com

I'll be honest based on previous threads on this topic I feel like I'm in the minority of not really having much in the way of deployment infrastructure.

I have hosts on Linode running my application via systemd services and my Angular app is served by Apache. I've also got Cloudflare in front of the web app.

I use BitBucket cloud for source control.

When it comes to deployment, I have SSH keys set up and a couple of shell scripts to build, upload, backup, and then restart the specific backend service or frontend app I'm deploying.

Then I just have a manual regression test plan that I run through to confirm application functionality which only takes about two minutes to complete validation.

It's simple but it works and it means I haven't spent a lot of time on engineering bling at this point instead of actually working on bringing my project to market.

I also use a heroku-ish workflow where a git push deploys to a $10 Digital Ocean droplet running Piku. Because Piku is very lightweight you can run a large number of projects (Indie Hacker product experiments) on a single box. I've got 22 running on my one droplet at the moment.

I'm personally using Dokku hosted on a Digital Ocean droplet. It's very easy to install since it's a one-click app and it has a lot of plugins that make things like SSL or DB with backups really easy to setup. It certainly does the job at the beginning, once I'll gain traction I'll think about moving to a more robust solution 😉

Setting up hosting and deployment infrastructure can be quite time-consuming and take focus away from building product.

Server and deployment management is what I'm trying help simplify with my app Cleaver. We're still a bit early on and haven't rounded out everything quite yet, but should be a great start for anyone who wants to ride along and help navigate.

If you are just starting off, I think Heroku + Github Actions will keep sanity imo. For DBs, AWS RDS or Heroku's postgres offerings would be good. If it is frontend only application, Netlify/Vercel should something you should look for. People also suggest using Firebase as a primary DB, but I've not used that :)

Where I work at this stack works decently for us:

Cloud:

• Mix of Heroku and AWS (for backend)
• Netlify (for frontend)

Database and storage:

• AWS RDS (Postgres)
• AWS S3

Infra Management:

• Terraform

CI/CD:

• Github Actions
• Before Github Actions used CircleCI a lot

Github and Heroku.

Firebase take all that for you. The hosting is serve as CDN and it implements different technics to deliver, there is no need of load balance, and the functions get balanced and scaled automatically by them. The deploy is also one command thing

Netlify or Digital Ocean, depending on the project.

• Kubernetes on Google Cloud (GKE)
• Using AWS S3 for file storage and AWS CloudFront as CDN for media files due to a limitation in GCE-Ingress related to Google Cloud CDN and Google Cloud Storage.
• Deploy is a git push that's triggers a Google Cloud Build Job.

It's slightly more expensive than using something static like Digitalocean droplets but have saved me multiple times when some user-generated content got viral since GKE autoscales capacity.

As for the software stack, here's whats powers QuizMe:

• Python
• Django
• Gunicorn
• Nginx
• Postgresql
• Memcached

I have a mostly–immutable blue/green deployment system set up with Terraform and Packer on DigitalOcean.

The actual infrastructure is a load balancer running nginx and 1–2 application servers (one for each “color”). TL;DR on every deploy, Packer builds a new image for the staging color, then Terraform creates a new application server using that image and updates the nginx configuration to point to it. Then I can spin down the other application server whenever I want, and on the next deploy the same process happens for the opposite color. I track which color is live using a tag on the load balancer.

This is all run locally; I haven’t set up continuous integration or anything. The whole set up costs $10/mo for the load balancer + application server and 1.4 cents per deploy.

We use these 3 extensively.

1. Google Cloud (GCE, Firestore, GCF)
2. Cloudflare (CDN, DNS, Firewall)
3. Gitlab (Code and CI/CD)

No, terraform or config management. (Mostly we haven't yet reached that level).

All the systemd configs, packages, installation steps are stored in repo.

On a similar note, we shared our awesome-resources https://github.com/HorizonTechnologies/awesome-resources