Droplet: Docker = laravel (mysql) + astro app [unusable]

Question

I have been developing a Laravel (mysql) + astro frontend application using docker to manage the services locally. Now that I am at the point where I want to test a production setup, I went with a Digital Ocean Droplet and the cheapest $4 configuration. When testing the startup of the docker container for the first time, I see the CPU usage climb to 100% and the droplet eventually becomes unreachable for hours.

So I went back to the drawing board with my docker configuration and tried to ensure that it was more tailored for production, and I also tried another configuration, this time with an intel vcpu and 2GB of RAM. Still, I am seeing the machine climb to 100% CPU usage and the droplet becoming unreachable for hours.

Does anyone else have examples of making this kind of small application setup working on a droplet or do I need to consider using App Platform instead?

One thing I will mention is that composer needs to install laravel and other dependencies on first run as well as node modules need to be installed on first run. I expected that this might take a little time but to be at 100% CPU usage for hours just does not make sense to me.

alexdo · Answer

Heya, @keneucker This behaviour is actually very common on small Droplets and usually isn’t a Docker or DigitalOcean issue. What’s causing it is that CPU-heavy build steps are running at container startup . composer install, npm install, Astro builds, native module compilation, and MySQL initialization are all expensive operations. On a $4 Droplet (and even on 1 vCPU / 2 GB), doing all of this at once will easily peg the CPU at 100% for a long time and make the Droplet appear unreachable, including over SSH. For production, these steps should not run at runtime. The usual approach is to: Run Composer and Node installs during the Docker image build Build frontend assets ahead of time Deploy pre-built images to the Droplet Once installs are moved out of startup, container launch time drops from minutes (or hours) to seconds. More RAM helps a bit, but this is primarily a single-core CPU bottleneck , and Docker adds some overhead on top. App Platform avoids this because builds happen elsewhere, but it’s not required — many people run Laravel + frontend stacks on Droplets successfully once builds are handled in CI or multi-stage Docker builds. Hope that this helps!

Bobby · Answer

Hi there, I might be wrong, but this behavior is pretty common when doing first-time installs on small Droplets. Composer and Node module installs are very CPU and memory intensive, and on a $4 or even a 2 GB Droplet it’s easy to hit 100% CPU and memory pressure, which can make the Droplet feel unresponsive. In many cases this can still work with a bit of tuning. Enabling some swap helps a lot, especially during the initial install. It also helps to avoid running composer install and npm install every time the containers start. A common pattern is to build the Docker images locally or in CI, then deploy prebuilt images to the Droplet. For a small setup, Laravel, MySQL, and a Node based frontend can run fine on a Droplet once everything is built and running, but the build phase needs more resources than people often expect. If you’d rather not deal with that, App Platform handles the build step for you and keeps it isolated from runtime. So it can work on a Droplet with some adjustments, but moving build work out of runtime and adding a bit of swap usually makes a big difference.

Become a contributor for community

DigitalOcean Documentation

Resources for startups and AI-native businesses

The developer cloud

Start building today

Report this

Droplet: Docker => laravel (mysql) + astro app [unusable]

Related questions

Mysql Data disappears intermittently

docker laravel app with astro frontend

Can I create a managed MySQL cluster running a recent version?

how to connect digital ocean app platform app with cpanel mysql database?

What’s the best way to handle downtime during software deployments?

Where to Find Slow Query Logs in Managed Database (Already Enabled)?

Search index command in pimcore 10.5.15 production gives error.

Query taking 1 minute to run

Fullstack Architecture Deployments Recomendations

Mysql CPU alerts and max_execution_time

I can't install MySQL on the server

How to reset database cluster