Personal trainer app

make
make setup-git-hooks # if you want to develop straight to main branch

make dev

Navigate to http://localhost:4000 to see the service in action. You can attach a debugger to it.

If you get something like the following error when running the below commands:

Error: the config for your app is missing an app name, add an app field to the fly.toml file or specify with the -a flag

Then, you need to select the fly app you have deployed:

export FLY_APP=petrapp

This project uses Fly.io for infrastructure and Litestream for SQLite database backups. It's a single instance Dockerized application with a persistent volume. Try fly launch to configure your own. You might also need to add some secrets to with fly secrets.

The container image contains sqlite3 binary to make it easy to manipulate the live production database.

make fly-sqlite3

One way to recover a lost or broken database is to restore it with Litestream. The process could still use some improvements but at least it works. Notably, you need to have a working machine running so that you can run commands on it. Another alternative is to clone the machine with an empty volume and populate it yourself using the fly sftp shell command.

# list databases
fly ssh console --user petrapp -C "/dist/litestream databases"
# list snapshot generations of selected database
fly ssh console --user petrapp -C "/dist/litestream snapshots /data/petrapp.sqlite3"
# restore latest snapshot to /data/petrapp4.sqlite
fly ssh console --user petrapp -C "/dist/litestream restore -o /data/petrapp4.sqlite -generation db5b998e60a203a3 /data/petrapp.sqlite3"

# Edit fly.toml env PETRAPP_SQLITE_URL = "/data/petrapp.sqlite3" before deploying to take new database into use
vim fly.toml

# Deploy the new configuration
fly deploy

Use pprof for perfomance investigation.

Proxy the pprof server to your local machine.

fly proxy 6060:6060

Capture a CPU profile of the running app.

go tool pprof --http=: "http://localhost:6060/debug/pprof/profile?seconds=30"

Capture a goroutine stack traces.

go tool pprof -top "http://localhost:6060/debug/pprof/goroutine"

When a request times out, the app writes a trace to a file and logs something like the following line:

{
  "time": "2025-09-13T10:02:11.604995985+03:00",
  "level": "WARN",
  "msg": "captured timeout trace",
  "service_name": "pr-29-myrjola-petrapp",
  "file": "/data/traces/timeout-20250913-070211.trace",
  "bytes": 709652,
  "trace_id": "HBGYTREFLURSGLEQGR2OX4XEBK",
  "proto": "HTTP/1.1",
  "method": "GET",
  "uri": "/api/test/timeout?sleep_ms=3000"
}

This file can be downloaded with the following replacing FLY_APP and file name with service_name and file from the log line:

FLY_APP=pr-29-myrjola-petrapp fly sftp get /data/traces/timeout-20250913-070211.trace

Once you have the file, you can analyze it with:

go tool trace timeout-20250913-070211.trace

This project uses GitHub Actions for CI/CD.

Prerequisite: Ensure you have Fly set up correctly with fly auth whoami.

Create a new app with a globally unique name.

fly apps create petrapp-staging

Create a bucket for the database backups. This should configure the secrets automatically matching the configuration in litestream.yml.

fly storage create --app petrapp-staging --name petrapp-staging-backup

Now we are ready to deploy the app.

fly deploy --app petrapp-staging

Petra logo made by Martin Yrjölä using Inkscape.