FAQ & Troubleshooting

Most connection problems are one of four different failures that look alike from the outside. Identify which one you have before changing settings — otherwise you will change things that were never the cause.

Start here: which failure is it?

Work down this list in order. Each step tells you what the answer rules out.

1. Does the client say it is connected? If it never reaches a connected state, the problem is in establishing the tunnel — see Cannot connect.

2. With the tunnel up, does anything load at all? If every site fails but the client still says connected, this is almost always name resolution, not the node — see Connected but nothing loads.

3. Do some things work and others fail, or does it work and then stop? That pattern points at node selection rather than a dead tunnel — see Works intermittently.

4. Does everything load, just slowly? See Slow throughput.

Confirm traffic is actually going through the tunnel

A connected indicator only means the tunnel was established. It does not prove your traffic is using it. Confirm it directly:

1. With the tunnel off, open any page that reports your public IP address and note it.
2. Connect, then reload that page.
3. The address should change. If it does not, your traffic is bypassing the tunnel.

Common reasons traffic bypasses a running tunnel:

• The network has its own proxy configured. A proxy set on the Wi-Fi network or system takes effect for applications that honour it, sending their traffic outside the tunnel. Check the current network's proxy settings before concluding the tunnel is broken.
• Per-application rules exclude the app you are testing with, where the platform offers that feature.
• A route rule sends that destination direct — see the advanced configuration guide.

Cannot connect

The client never reaches a connected state.

1. Check the profile is still valid. Refresh the subscription. An expired or revoked profile can leave entries that no longer authenticate.
2. Check for another VPN. Most systems allow only one active VPN configuration. Turn off other VPN or filtering apps and retry.
3. Confirm the system permission. The first connection requires approving a system VPN request. If it was previously denied, the client may fail without an obvious message; re-approve it in the operating system's VPN settings.
4. Try a different network. Some networks block the transport a profile uses. Switching from Wi-Fi to mobile data — or the reverse — separates "this endpoint is down" from "this network blocks it".
5. Try one specific endpoint rather than an automatic group, so the result points at a single endpoint instead of whichever one the group happened to pick.

If a port-in-use error appears, another application holds the port the client wants. Change the local port in settings rather than stopping the other application.

Connected but nothing loads

The tunnel reports connected, but no page opens. When everything fails at once, suspect DNS before anything else — a failed lookup means no destination is ever reached, which looks identical to a dead tunnel.

1. Test with an address instead of a name. If a numeric address responds while names do not, the tunnel is fine and DNS is the problem.
2. Check which resolver the profile uses for proxied traffic, and whether that resolver is reachable from the current network.
3. Re-check after each change. DNS answers are cached; a stale cached failure can outlive the fix. Reconnect and retry in a fresh browser window.

Then check routing: if a catch-all rule sends traffic somewhere unreachable, every destination fails. Restore a known-good profile to confirm.

Works intermittently

Some connections succeed, others fail, with no pattern — or it works, then stops after a while. This usually is not a broken tunnel. Two causes account for most cases:

The group is sending traffic to endpoints that cannot be reached. A profile can contain entries that are unreachable from your network. Selection strategies that do not rank by health distribute across every member anyway, so a share of your connections goes to dead entries. Switching the group to a health-ranking strategy, or selecting one working endpoint manually, will show immediately whether this is the cause. See node selection.

The health data is stale. Latency measurements refresh on an interval and may stop refreshing while the tunnel is idle. After a long idle period, selection can be based on measurements taken much earlier, when a now-dead endpoint was the fastest. Re-run the latency test before concluding an endpoint is bad.

If it degrades only after hours of uptime, note how long it takes and whether reconnecting restores it — that distinction is useful when reporting the problem.

Slow throughput

1. Measure the baseline first. Test the same download with the tunnel off. If it is also slow, nothing in the client will fix it.
2. Do not trust latency as a speed indicator. The number next to an endpoint is a round-trip time, not bandwidth. An endpoint can answer a small test request quickly and still transfer slowly. Test the transfer you actually care about.
3. Try endpoints in different locations and of different protocol types. Throughput can differ substantially between transports on the same network.
4. Test with a large download, not a page load. Page loads are dominated by round trips and will not reveal a bandwidth ceiling.
5. Check whether the link itself is the limit. On a metered or roaming mobile connection the ceiling may be far below what the endpoint can deliver.

Battery and background behaviour on mobile

A tunnel keeps a process running, which uses power. If the connection drops whenever the screen is off, the operating system is likely suspending the app: check the system battery optimisation settings for it. Where the client offers a notification for the active connection, keeping that enabled also helps the system keep it alive.

Collecting diagnostics

Where the build provides a log view or export, capture the log covering the failure — reproduce the problem first, then export, so the relevant lines are included.

Useful details to record before asking for help:

• app version and operating system version;
• what the client displayed (the exact status or error text);
• whether the same destination worked with the tunnel off;
• whether a different network behaved differently;
• whether reconnecting or restoring a previous configuration fixed it.

Getting help

Contact support through the support page, or email service@allianceinterstellar.com. Include the details listed above, with credentials redacted.

For endpoint or subscription problems — expired plans, exhausted quota, endpoints that are down — contact the provider that issued the profile. The client cannot fix an endpoint that is not serving.