FIELD GUIDES / FIRST VPS
Learn the next
operating step.
Twenty practical guides for choosing, securing, operating and recovering a first small application.
Start with the task, not a promise.
Size the app you actually run.
Build a first VPS resource budget from memory, CPU, disk and transfer observations, with an annotated example and clear upgrade decisions.
Read the guide →FIELD GUIDEGive your first API a repeatable release.
Run one small Node.js API under systemd, verify its executable and local response, then route HTTPS through Caddy with a safe stop and rollback plan.
Read the guide →FIELD GUIDEA backup is a question. A restore answers it.
Practice a bounded PostgreSQL and upload-file restore on a disposable target, verify matching markers, and record what the exercise proves.
Read the guide →FIELD GUIDEYour first SSH session, with a way back.
Verify a server’s host key, connect with your own SSH key, and prove a separate second login before closing the first session or changing access.
Read the guide →FIELD GUIDEFollow one request from DNS to your app.
Find the failing layer between a domain and a small application: DNS records, network access, TLS, reverse proxy and the listening process.
Read the guide →FIELD GUIDEYour app stopped. Find the first useful clue.
Build a short failure timeline for a small VPS app using its service state, listening port, first relevant log entry, resources and recent release.
Read the guide →FIELD GUIDEGive a small scheduled job a clear boundary.
Design one finite Linux job with an explicit timezone, restricted permissions, useful logs and overlap protection using a systemd timer and flock.
Read the guide →FIELD GUIDERead memory and disk warnings before upgrading.
Interpret available memory, process RSS, filesystem capacity and inode use, then identify logs, uploads and release artifacts before changing VPS resources.
Read the guide →FIELD GUIDEPay for your first VPS with crypto without mixing up the network
Read the asset, network, address-only QR and exact period total as one payment instruction.
Read the guide →FIELD GUIDEA first-hour security checklist for a new Linux VPS
Plan recovery access and the smallest exposed surface before changing sign-in settings.
Read the guide →FIELD GUIDEWhy your Docker port may still be reachable
Understand the host address behind a published container port before relying on a firewall rule.
Read the guide →FIELD GUIDEChoose a Linux image for the app you want to run
Choose a supported runtime path before choosing a distribution name.
Read the guide →FIELD GUIDEMake a small systemd service restart predictably
Connect the executable, working directory and restart behavior to an observable app result.
Read the guide →FIELD GUIDERead a failed HTTPS request step by step
Follow a request from name to certificate to proxy to application without guessing the layer.
Read the guide →FIELD GUIDEKeep a database bound to the intended interface
Choose the database listener from the application’s connection path, not from a copied default.
Read the guide →FIELD GUIDETest an application backup without overwriting production
Use an explicit restore target to prove what the backup can actually recover.
Read the guide →FIELD GUIDEFind a full disk without deleting unknown data
Locate the consuming path before deleting files, logs or database data.
Read the guide →FIELD GUIDEStore app configuration separately from source code
Keep secrets and environment-specific settings out of a repository and release artifact.
Read the guide →FIELD GUIDEMeasure a small app before and after a change
Decide what one small change should improve before increasing VPS resources.
Read the guide →FIELD GUIDEMove a hostname to a replacement VPS carefully
Prepare DNS, TLS, data and a return path before changing a public name.
Read the guide →Terms used in this manual.
- VPS
- A virtual server with its own operating environment and assigned resources. For a first app, plan for the work of maintaining that environment as well as writing the application.
- vCPU
- A unit of virtual compute allocation. Two vCPUs do not promise twice as many requests: the application, database and allocation model affect the result.
- RAM
- Working memory shared by the operating system and running processes. An API, its database and a deployment can need memory at the same time.
- Available memory
- An estimate of memory Linux can make available to new work without swapping. It is usually more useful than reading the free column alone.
- Storage capacity
- Space for files, the database, uploads, logs and release artifacts. It is separate from monthly network transfer and does not establish backup coverage.
- Monthly transfer
- Data moved over the network during a billing window. Response bodies, uploads and off-server backups can all contribute; confirm the actual counting and overage rules.
- Hostname
- A name for an instance or service. Entering api.example.com in the configurator does not create a public DNS record or a TLS certificate.
- DNS
- The system that publishes records for names. A and AAAA records connect a hostname to IPv4 and IPv6 addresses; cached answers can outlive a record change.
- TLS
- The protocol used by HTTPS to protect a connection and authenticate its hostname. A valid certificate does not prove that an application is healthy.
- SSH public key
- The shareable half of an SSH key pair. The corresponding private key stays with you. OffVPS accepts a public key or lets you leave access setup for later.
- Reverse proxy
- A service that accepts an incoming request and forwards it to your application. A proxy error and an application error can require different checks.
- Service unit
- A systemd configuration describing how a service runs on a system that uses systemd. Its user, executable, working directory and restart policy need to match your actual app.
- Scheduled job
- Work started at a chosen time, such as a daily export. Decide what happens if yesterday’s run is still active or the machine was off at the scheduled time.
- Restore point
- A recoverable version of data from a particular point in time. Verify that the needed database and files can be recovered together, rather than assuming any copy is sufficient.
- Upfront period
- A full service period paid in one payment. OffVPS discounts six-month and annual periods; the payment quote shows the whole reduced total, not a monthly instalment.
- Local payment reference
- The identifier for the configuration and payment instructions saved in your browser tab. It is not a server invoice or proof that a transfer has been verified.