Remote access over SSH and VPN

Run PQEnalyzer beside the simulation files on the server and open its web interface in your desktop browser. The server reads and analyzes the original files; the interface and chart data travel over an encrypted SSH connection. No CSV conversion or manual transfer of the input files is required.

Only the server needs PQEnalyzer installed. Your desktop needs SSH and a browser. Use your own username and server hostname in place of the examples below.

From home over VPN

  1. Connect your institution’s VPN if it is required to reach the server.

  2. Confirm that normal SSH access works from your desktop:

    ssh user@login.cluster

  3. Start the app on that server, then open the tunnel from a second desktop terminal as shown below.

The same tunnel works from the institution’s network and from home over VPN. The VPN must provide a route to the SSH host and resolve its hostname, or your institution must provide a reachable address or SSH alias. The SSH server must allow TCP forwarding. Follow the cluster’s policy for login and compute nodes.

Start the app on the server

pqenalyzer web --no-open --port 8766 /path/to/simulation.en

See the startup output. Leave this terminal running; Ctrl+C stops the server.

Paths refer to files on the server. Keep each energy file’s matching .info file beside it. Supply multiple files in their dataset order when needed.

The app listens on 127.0.0.1:8766 on the server. Keep this default loopback binding; remote access is provided by SSH.

Open the tunnel on your desktop

ssh -N -o ExitOnForwardFailure=yes \
  -L 127.0.0.1:8766:127.0.0.1:8766 user@login.cluster

Open http://127.0.0.1:8766 in your desktop browser. The first address and port are on your desktop; the second 127.0.0.1:8766 is reached from the SSH server. The SSH destination must therefore be the machine running PQEnalyzer.

-N keeps SSH open for forwarding without starting a remote shell. ExitOnForwardFailure=yes reports failure to establish the forwarding listener; it does not check whether the web app is running.

Compute node behind a login node

If your site permits SSH to an allocated compute node, start PQEnalyzer there inside the allocation. Then use the login node as a jump host:

ssh -N -o ExitOnForwardFailure=yes -J user@login.cluster \
  -L 127.0.0.1:8766:127.0.0.1:8766 user@compute-node

The browser address remains http://127.0.0.1:8766. The final SSH destination is the compute node hosting the app. Keep the allocation active for the session.

Keep the session open

Keep both the app and SSH tunnel running. Ctrl+C in the tunnel terminal closes the tunnel; stop the app separately when finished.

After a VPN or SSH disconnection, reconnect the VPN and rerun the tunnel command. Restart the app too if its server session or allocation ended. The browser can then reload the same local address.

Troubleshooting

Symptom

Check

SSH times out or the hostname cannot be resolved

VPN connection, institutional DNS, hostname, and route to the SSH host. Use the same hostname or SSH alias as your normal login.

SSH login fails

Your SSH credentials and the site’s access policy; establish normal SSH access first.

Forwarding is administratively prohibited

The site’s SSH forwarding or compute-node access policy. Ask the cluster administrator which access path is supported.

Local address is already in use

Choose a different desktop port, as below.

The tunnel opens but the page does not

Check that the app is still running on the final SSH destination and that the remote port matches. From that server, try curl --fail http://127.0.0.1:8766/api/status.

For a busy desktop port, change only the local port:

ssh -N -o ExitOnForwardFailure=yes \
  -L 127.0.0.1:18766:127.0.0.1:8766 user@login.cluster

Then open http://127.0.0.1:18766. If port 8766 is busy on the server, choose another --port for PQEnalyzer and use that value for the remote port in -L.

See the OpenSSH manual for -L, -N, and -J.