Manage Ubuntu services from Windows using the SSH you already have
A website incident often starts with a small question: is the application actually running? On a Windows PC, answering it can mean opening SSH, finding the right unit name, checking recent logs and remembering which local port belongs to which site. Server App Manager puts those related observations in one local dashboard.
This guide follows the current source preview from download to one registered application. It uses user@server.example.com and demo-web.service as placeholders. The walkthrough is a reproducible exercise, not a report of measurements from a production server.
Understand the three places involved
Your browser opens the dashboard on your own computer. An ASP.NET Core process on that computer serves the interface and holds the active SSH connection. Ubuntu continues running its existing applications, systemd and journal.
The public project website has a different job: it provides downloads and instructions. Opening this article does not connect to your server. Likewise, entering a public website URL in the dashboard is a health-check setting, not a replacement for the SSH address used at sign-in.
This separation explains several otherwise confusing behaviors. Closing a browser tab does not necessarily stop the local process. Quitting the local dashboard ends your monitoring session, while Ubuntu applications continue under their existing service manager. A restart button in the dashboard is a separate, explicit request concerning a registered service.
Gather the information before launching
Start with an SSH account that already works from your PC. Obtain its username, hostname or address, and any custom SSH port. Also obtain the server host-key fingerprint from a trusted console or the administrator responsible for that machine.
For your first application, identify its actual canonical systemd service name. A friendly business name or website domain is not enough. If you know only the site's public address, ask its operator which unit serves it. A server can host several applications with similar deployment names.
Check the compatibility guide for current preview requirements. The Windows source workflow needs the .NET 10 SDK. Operations such as restarting services and reading particular journal entries depend on your existing Ubuntu permissions. The dashboard does not grant those permissions during setup.
Get the current source
Use the download page to obtain the application source ZIP, then extract it into a folder on your PC. The current ZIP points to the main branch, so record the revision if you need to reproduce an evaluation later.
Inside the extracted folder, double-click Create-Desktop-Shortcut.cmd. The helper builds a local copy, checks local HTTPS certificate availability and creates the Server App Manager desktop shortcut. Read any certificate trust prompt as a local Windows setup action.
If you prefer a source command, the quickstart gives the PowerShell alternative. Keep that command window open while a direct source run is active. The shortcut instead provides a tray menu for opening and quitting the local app.
Start the dashboard on your PC
Open the new desktop shortcut. The usual browser address is https://localhost:7300. Here, localhost means the computer on which the browser is running.
A browser connection error at this address usually concerns local startup, the port or HTTPS trust. It does not establish that Ubuntu is unavailable. Check whether the local app is running before troubleshooting the remote server.
The tray icon offers Open dashboard and Quit Server App Manager. After restarting Windows, launch the shortcut again. The source preview does not promise automatic startup at Windows sign-in. Keeping this startup model clear helps avoid interpreting an absent local page as a failure of a hosted application.
Verify the SSH endpoint
Enter user@server.example.com on first run. If the existing SSH service uses a custom port, the dashboard accepts a value such as user@server.example.com:2222.
The app shows the SSH host-key fingerprint before asking you to trust the endpoint. Compare the complete fingerprint and its key type with the trusted information you collected. Do not treat a successful network connection as proof that it reached the intended server.
After confirmation, enter the Ubuntu account's SSH password. A successful first sign-in saves the endpoint and verified host identity locally. Later sign-ins verify that saved identity again. The current preview uses SSH password authentication; dashboard key authentication is separate roadmap work.
Find an existing application
Open Applications and choose Find apps. The source preview looks for recognizable .NET service launch patterns among loaded systemd units. A discovery result is a candidate to inspect.
Review the canonical service name, any detected deployment metadata and the visible listening port. Confirm that they belong to the intended application. A background worker can be a valid .NET service while exposing no HTTP listener.
Supply a descriptive display name and, for a public website, its actual HTTP or HTTPS address. The website address is not inferred solely from a listening port. Multiple public domains can share a proxy, and application ports can be private to the host.
Register only what you intend to manage
Choose Review and add, check the proposed fields and save the entry. Registration adds it to your local dashboard registry. It does not create or edit a systemd unit.
For an unrecognized launch wrapper, manual registration can still be appropriate after verifying the real service. Read the application guide rather than inventing a port or deployment path to fill a blank field.
The enabled setting and display order help organize the list. Keep environment labels explicit when names are similar. An entry called “Demo website — test” is easier to distinguish from another environment than two entries with the same friendly name.
Inspect before using the controls
Open the registered application's status and logs first. Note the service state, main PID, available memory readings and any recent failure. Missing measurements appear as N/A; that is different from a measured zero.
The Logs view loads a bounded slice of recent journal entries. Search and filter within those loaded messages. Read surrounding information entries as well as errors, because the startup sequence can explain which stage failed.
Only use a management action when it addresses a known operational need. Stop and restart require confirmation, and the service may become unavailable during the change. On a shared host, check dependencies and ownership before confirming. Practice the workflow with a disposable service first.
Read website health as an independent signal
A running service and a healthy public URL answer different questions. Open Website Health for the registered address and note the HTTP result, response time and check timestamp.
A failed public check can involve DNS, TLS, networking, a reverse proxy or the application's response. A successful check can reflect one public endpoint without proving that every feature works. For a background worker, HTTP availability may not apply at all.
When a symptom appears, write down both observations at nearby times: for example, the actual service state and the actual HTTP response. Avoid replacing this evidence with a broad conclusion such as “the whole server is down.”
Finish a session deliberately
Sign out when you finish inspecting the server. The local SSH session ends and its in-memory credential is cleared through the application's connection lifecycle. Use the tray menu if you also want to stop the local dashboard process.
Registered applications and audit history remain local for the saved server. Remove beside an application unregisters that local entry. Remove server is a separate authenticated setting that closes the connection and archives the registration's data.
Review update and removal instructions before moving or deleting local state. Treat backups as private operational data, even when they contain no saved SSH password.
Keep a useful first-session record
For an evaluation, record the source revision, Windows version, Ubuntu version, registered sample unit and which reads were available. Note permission-denied and unavailable readings separately from application failures.
A useful completed exercise has a narrow outcome: you can start the dashboard, verify the endpoint, register the correct existing service, inspect its status and logs, and distinguish its process state from website health. That is a stronger starting point than adding many unverified services at once.
For the product-specific behavior described here, consult the application README and security notes. The hosting relationship between ASP.NET Core, systemd and a proxy is documented in Microsoft's Linux hosting guide.