Use this guide if you’ve imported your own server details, turned on the connection in Shadowrocket, and still can’t load web pages. Turning on the connection only means the client is enabled; it doesn’t confirm that the server is reachable or that web requests are succeeding. The steps below use the iPhone / iPad interface and the same web page for comparison. Once you identify the cause, change only the relevant setting.
Step 1: Check the server and your network
Note the full domain of the page that fails, your current network, and the server selected in Home. Choose two test pages: the one that fails and a regular page known to load directly on your current network. If only the first page fails, don’t start by updating your subscription. If both fail, check the device’s basic network connection first.
-
Check direct access
If your network allows direct access, turn off the connection on
Home, then open a page known to load directly. If the switch turns back on by itself, checkSettings → On Demandfor an active automatic connection condition. Temporarily disable it only if permitted, and restore it after testing. If ordinary pages still won’t load with the connection off, check the Wi-Fi sign-in page, cellular data, or the device’s network settings. -
Test the selected server
Return to
Home, make sure the server you want to check is selected, then runConnectivity Test. Check the result for that server, not the latency shown for another item in the list. A latency reading means only that the probe received a response; it doesn’t guarantee that domain lookup, connection, and resource loading for a web page will all work. -
Check the address and port
If the selected server keeps timing out, open its configuration and compare the server address, port, protocol, and authentication fields with the details you have. HTTPS commonly uses TCP port 443 for web traffic, but the server’s listening port depends on its actual configuration. Don’t change the server port to 443 just because a web page won’t load. If the details have changed, get the correct values from your service provider rather than guessing.
-
Switch networks only
Keep the server and all other settings unchanged. If permitted, switch from Wi-Fi to cellular and run the same test again. If the same server times out on only one network, first check that network’s sign-in status, outbound restrictions, or route to the server. If it times out on both, test another server entry you already have before asking your service provider to verify the original one.
Step 2: Check Global Routing
After the server responds to a test, go to Home → Global Routing and note the selected option. The table below explains how each option affects testing. Switch options only for diagnosis; if your device must always send traffic through a specified route, don’t switch to Direct.
| Option | Behavior | When to use it |
|---|---|---|
Proxy | Routes traffic handled by the proxy through the proxy, without applying the policy rules in Config. | Check whether the same page loads through the selected server. |
Direct | Sends handled traffic directly, without using a proxy server. | Use for comparison when direct access is allowed; this does not turn off the connection. |
Config | Selects a policy based on the rules in the currently active configuration. | Check whether a domain is assigned the wrong policy or matches REJECT. |
Scene | Applies the routing behavior configured for the selected scene. | If results change after switching networks, also check whether the active scene changed. |
Takeaway: Compare routing modes with the same page
If the page loads in Proxy but fails again in Config on the same network and server, note the domain of the failed request, then check DNS and rules in the active configuration. This comparison narrows down the cause, but it doesn’t prove that REJECT matched; the request may have been routed through an unavailable outbound path instead.
Step 3: Tell DNS failures from connection timeouts
DNS resolves domain names to addresses. A successful lookup doesn’t prevent a connection from timing out. Conversely, if the server address itself is a domain name, failure to resolve it can also make Connectivity Test time out. The flow below shows the stages involved in loading a page, not a fixed execution order. Domain rules may be matched first, while whether resolution happens locally or on the outbound side depends on the configuration and connection method.
Error: ERR_NAME_NOT_RESOLVED
Cause: The browser didn’t get a usable result for the domain lookup. The domain may be misspelled, or the current resolution path may not have returned a valid address. Fix: Check the full domain in the address bar. If several known-valid domains all fail, check the DNS settings in the active configuration.
Error: ERR_CONNECTION_TIMED_OUT
Cause: The browser couldn’t establish a connection before the timeout. This error alone doesn’t indicate a DNS problem. Fix: Leave DNS unchanged and compare the server and Global Routing as described in the first two steps. If only one site times out, consider whether the site itself is unreachable.
These are errors a browser may display, not Shadowrocket alerts. If the problem started after editing a configuration, go to Config, confirm which file is active, and save a copy before checking whether dns-server in [General] was changed recently. If you entered an encrypted DNS service address, check the protocol, hostname, and path as well as the domain in the address. Standard DNS commonly uses port 53; DNS over HTTPS typically uses HTTPS on port 443. A reachable port doesn’t guarantee that the service returns a valid lookup result.
If you have a record of a change, revert only that setting to its previously working value, then open a new browser tab and retest the same domain and comparison domain. Don’t change system DNS, client DNS, and routing mode at the same time. For an internal workplace domain, a public resolver may not have the required record, so keep the resolution method required by that network. Replacing the URL with an IP address is not an equivalent test: HTTPS certificates and site selection may still depend on the original domain.
Step 4: Check for REJECT matches and rule order
Restore Global Routing to Config, then go to Config and confirm which configuration file is in use. Review its [Rule] section from top to bottom to find the first matching rule, and check the final FINAL rule. If connection logs show the actual request hostname and matched policy, use them to verify the result first; the domain in the address bar may not be the request that failed.
[Rule]
DOMAIN,portal.example.com,PROXY
DOMAIN-SUFFIX,example.com,REJECT
FINAL,DIRECT
This is a syntax example, not a complete replacement configuration. The first line matches only portal.example.com and uses PROXY. The second line matches example.com and its subdomains. If you move it above the first line, portal.example.com will match REJECT first. If the sign-in page also requests login.example.com, sign-in may fail because of the second line even if the main page loads. FINAL,DIRECT handles traffic that hasn’t matched any earlier rule; it doesn’t override rules that have already matched.
- If the main page is blank or a button doesn’t respond, first identify whether the failure is in the page itself, the sign-in endpoint, or a subresource such as a script. Don’t add a broad exception based only on the domain in the address bar.
- If you confirm that traffic was blocked by mistake, save a copy of the configuration, then adjust the rule or its order for the relevant host while preserving your existing network policy. After the change, retry the original action in
Configmode and verify that requests meant to be blocked are still handled as expected. - If you don’t have information showing which rule matched, don’t assume every timeout is caused by
REJECT. Continue by checking the actual outbound route, DNS results, and whether the page is reachable on another permitted network.
Common symptoms and what to check next
Why does the page keep loading when latency is shown?
In the same mode, try opening another page known to work. If only the original page keeps loading, note its domain and the browser error, then check the site or its subresources. If both pages fail, troubleshoot the server, routing mode, DNS, and rules in that order. Probe latency is not the time it takes to load the entire page.
Why does the connection drop after switching from Wi-Fi to cellular?
First check whether the connection switch on Home changed. Then go to Settings → On Demand and review the automatic connection conditions. If you use Scene, also check which scene became active after switching networks. Keep the same server and run Connectivity Test again to distinguish a scene change from a server that’s unreachable on one network.
Should I update my subscription or reinstall right away?
Not yet. Update a subscription you already use only after confirming that its server details are outdated or have changed. If you need to investigate further, note when the problem occurred, the network type, failed domain, current mode, and exact error message. Before sharing logs, remove subscription URLs, authentication fields, and full URLs that contain sensitive parameters.