Deploy Pongo to Fly.io for always-on persistent VMs with automatic scheduler and archiver startup.
Quick Start
Install flyctl
# macOS
brew install flyctl
# Linux
curl -L https://fly.io/install.sh | sh
# Windows
iwr https://fly.io/install.ps1 -useb | iex
Clone and navigate to repository
git clone https://github.com/TimMikeladze/pongo.git
cd pongo
Launch the app
This command:
- Detects the Dockerfile and fly.toml configuration
- Creates a new Fly.io app
- Provisions a 1GB volume for SQLite data
- Deploys the application
Set environment variables
# Required: Enable scheduler auto-start
fly secrets set SCHEDULER_ENABLED=true
# Optional: Use PostgreSQL instead of SQLite
fly secrets set DATABASE_URL="postgres://user:pass@host:5432/pongo"
# Optional: Password-protect dashboard
fly secrets set ACCESS_CODE="your-secret-password"
# Optional: Enable archiver
fly secrets set ARCHIVAL_ENABLED=true
Configuration
fly.toml
The included fly.toml configuration:
app = 'pongo-production'
primary_region = 'sjc'
[build]
[[mounts]]
source = 'pongo_data'
destination = '/data'
initial_size = '1gb'
[http_service]
internal_port = 3000
force_https = true
auto_stop_machines = 'off'
auto_start_machines = true
min_machines_running = 1
max_machines_running = 1
[[http_service.checks]]
interval = '30s'
timeout = '5s'
grace_period = '10s'
method = 'GET'
path = '/'
[scale]
count = 1
[[vm]]
memory = '1gb'
cpu_kind = 'shared'
cpus = 1
Key Settings
- Volume mount: Persistent storage at
/data for SQLite database
- Auto-stop disabled: Keeps machine always running for reliable monitoring
- Health checks: Ensures app stays healthy
- Single machine: Cost-effective for most monitoring needs
Environment Variables
Required
| Variable | Description | Default |
|---|
SCHEDULER_ENABLED | Auto-start scheduler in Docker | false |
Optional
| Variable | Description | Default |
|---|
DATABASE_URL | PostgreSQL connection string (omit for SQLite) | file:/data/pongo.db |
ACCESS_CODE | Dashboard password | - (no auth) |
PONGO_REGION | Region identifier for multi-region setups | default |
ARCHIVAL_ENABLED | Auto-start archiver in Docker | false |
S3_BUCKET | S3 bucket for archives | - |
S3_REGION | S3 region | - |
S3_ACCESS_KEY_ID | S3 access key | - |
S3_SECRET_ACCESS_KEY | S3 secret key | - |
Automatic Scheduler Startup
The included docker-entrypoint.js automatically starts the scheduler when SCHEDULER_ENABLED=true:
const schedulerEnabled = process.env.SCHEDULER_ENABLED === "true";
const archiverEnabled = process.env.ARCHIVAL_ENABLED === "true";
if (command === "bun run start") {
// Start scheduler in background if enabled
if (schedulerEnabled) {
spawn("bun", ["run", "src/scheduler/index.ts"], {
shell: true,
stdio: "inherit",
env,
});
}
// Start archiver in background if enabled
if (archiverEnabled) {
spawn("bun", ["run", "src/archiver/index.ts"], {
shell: true,
stdio: "inherit",
env,
});
}
}
- Dashboard, scheduler, and archiver run in a single machine
- No separate processes or services needed
- Simplified deployment and lower costs
Database Options
SQLite (Default)
By default, Pongo uses SQLite stored in the mounted volume:
# Database automatically stored at /data/pongo.db
DATABASE_URL=file:/data/pongo.db
- Zero configuration
- No external dependencies
- Fast local access
- WAL mode for concurrent reads
PostgreSQL
For production deployments or multi-region setups, use PostgreSQL:
fly secrets set DATABASE_URL="postgres://user:pass@host:5432/pongo"
- Fly.io Postgres:
fly postgres create
- External providers: Neon, Supabase, Railway
Scaling
Vertical Scaling
Increase machine resources:
# Increase memory
fly scale memory 2048
# Increase CPUs
fly scale count 2
Multi-Region Deployment
Deploy to multiple regions for geographic monitoring:
# Deploy to US East
fly regions add iad
fly scale count 2
# Set region identifier
fly secrets set PONGO_REGION=us-east
Monitoring and Logs
View logs
SSH into machine
Check machine status
View metrics
Troubleshooting
Scheduler not running
-
Verify
SCHEDULER_ENABLED=true:
-
Check logs for scheduler startup:
fly logs | grep scheduler
-
SSH into machine and check processes:
fly ssh console
ps aux | grep scheduler
Database persistence issues
-
Verify volume is mounted:
-
Check volume size:
fly ssh console
df -h /data
-
Expand volume if needed:
fly volumes extend <volume-id> --size 5
Out of memory
Increase machine memory:
Cost Optimization
Auto-stop during low usage
For development environments:
[http_service]
auto_stop_machines = 'suspend'
auto_start_machines = true
min_machines_running = 0
This will stop monitors from running when the machine suspends. Only use for development.
Use shared CPU
The default configuration uses shared CPUs, which are cost-effective for most monitoring workloads:
[[vm]]
cpu_kind = 'shared'
cpus = 1
Updating
Pull latest changes and redeploy:
git pull origin main
fly deploy