Turning on the Shadowrocket connection only confirms that the connection process has started on your device. It does not, by itself, confirm successful server authentication, DNS resolution, or rule-based routing. This guide is for users who already have their own server details or subscription. First establish a network baseline with Shadowrocket disconnected, then check the server, Global Routing, DNS, and Config one at a time. Change only one setting per test, then restore your usual setup.
1. Check whether the internet works with Shadowrocket disconnected
Don’t start by deleting your configuration or reinstalling the app. First determine whether the issue depends on the Shadowrocket connection. If regular webpages still won’t load when disconnected, check Wi-Fi, cellular service, or network sign-in first. If browsing works while disconnected but fails when connected, continue with the proxy path. This guide refers to the iPhone / iPad interface labels Home, Config, Data, and Settings. On other platforms, check the labels shown in your app rather than assuming the buttons are in the same place.
-
Note the current settings
On Home, note the selected server and Global Routing mode. In Config, note which configuration is enabled. Keep these notes on your own device; don’t share subscription URLs, passwords, or complete connection details.
-
Pause automatic connections
Go to Settings → On Demand, note the current settings, and temporarily turn them off while troubleshooting. Return to Home, disconnect, and make sure On Demand doesn’t immediately reconnect.
-
Establish a network baseline
In your browser, open two websites you normally use that have different content. Don’t rely on old tabs that have already loaded, as cached content can make a broken connection look normal.
-
Switch networks and test again
If cellular data is available, test over both Wi-Fi and cellular. If the issue occurs only on one Wi-Fi network, first check its sign-in requirements, router, and access restrictions. Don’t change the server and DNS at the same time.
Hotel and public Wi-Fi may require you to sign in on a webpage first. Disconnect the proxy, complete the network sign-in, then reconnect from Home. Recording which network you used, when, which webpage failed, and whether Shadowrocket was connected is more useful than repeatedly toggling the connection.
2. Check the server and connection on Home
On Home, confirm that the selected server is the one you intend to use, then run Connectivity Test. A response is a useful clue, but it doesn’t guarantee the webpage you want will load. A timeout also needs to be considered alongside the test target, current network, and actual webpage request. Don’t treat a single latency result as proof of successful server authentication or support for every protocol.
Open the selected server’s edit screen and compare each field with your original saved details. Address, Port, protocol, and authentication must all match. For Trojan, check TLS and SNI; for VMess, check the relevant identifiers and transport settings; for VLESS, check the transport settings. Matching addresses alone don’t mean the configurations are equivalent. Port numbers aren’t universal: if the original details specify 443, enter 443. Don’t substitute 8443 just because it appears in another example.
| What you’re seeing | Check first | What to do next |
|---|---|---|
| The same server fails on two networks | Address, port, authentication, and remote server status | Compare with the original details and ask your service provider whether the service is still active. |
| It fails only on one Wi-Fi network | Restrictions on connections or traffic over the current network | Keep the server details unchanged and test the same webpage on another network. |
| The test responds, but webpages still fail | Authentication, TLS, DNS, and routing for the actual request | Check Global Routing and observe the same webpage request. |
| The selected server changed after a subscription update | The current Home selection versus the original server details | Confirm which entry is actually selected; don’t just check whether the subscription update completed. |
A successful subscription update only means the subscription content was retrieved and processed. It doesn’t mean every server in it is working. Don’t repeatedly import the same subscription into multiple groups while troubleshooting. First identify the selected entry, then compare its address, port, and protocol before and after the update. This site does not provide servers or subscriptions, and cannot verify the validity or access rights of a service you already have.
3. Use Global Routing to distinguish server issues from rule issues
On Home → Global Routing, check the current mode. For troubleshooting, you can temporarily compare Proxy with Config, but keep the same server, network, and test webpage throughout. Direct tests the direct connection path; it doesn’t mean every destination should be reachable directly. With Scene, check the scene settings rather than inferring traffic behavior from the name alone.
- Proxy: Shows what happens when traffic is sent through the proxy path. If the same destination still fails, check the server, authentication, transport, and name resolution before making broad changes to routing rules.
- Direct: Tests the direct connection path. A webpage failing in this mode does not, on its own, prove that the local network is generally broken.
- Config: Routes traffic according to the currently enabled configuration. Check the file selected in Config, not another file that is merely saved in the list.
- Scene: The selected scene determines how traffic is handled. Note the original setting, then temporarily switch to a clearly defined test mode to keep the comparison consistent.
If Proxy works but Config doesn’t, check the matching rule
If the same webpage works under Proxy but not Config on the same network and server, check which rule and policy the request matches under Config. Don’t immediately replace all the server details. After comparing, restore the original mode and test the actual request again.
4. Troubleshoot DNS: distinguish the server hostname from the destination hostname
DNS issues can occur in at least two different places: the hostname in the server’s Address must be resolved before connecting, and the destination hostname must be resolved when you visit a webpage. If the first lookup fails, the server connection may never be established. If the second fails, Shadowrocket may show as connected while the browser reports that the server can’t be found. These are different problems; “change DNS” doesn’t cover both.
On the server edit screen in Home, check whether Address is a hostname or an IP address. Then review the current settings in Settings → DNS and the DNS-related entries in the active Config file. If the configuration includes dns-server under [General], check it alongside the current settings. Changing a value in an inactive configuration won’t necessarily take effect. Screens and layouts vary by platform, so use the options shown in your app.
- Keep the original value: Before changing anything, note the current DNS settings and where they’re configured so you can restore them if the test fails.
- Check reachability: Traditional DNS typically uses UDP or TCP port 53; DoH uses HTTPS, commonly on port 443. An open port doesn’t guarantee correct DNS results or a successful DoH request.
- Avoid circular dependencies: If the server hostname must be resolved before the proxy connection can be established, that lookup can’t rely solely on a proxy connection that hasn’t been established yet.
- Make a fresh request: After each change, reconnect and reload the test page. Keep app, system, and browser caches in mind; an old page still being visible doesn’t show whether the connection is working.
5. Check rule order and the final policy in Config
In Config, confirm which configuration is enabled, open that file, and check [Rule]. Rules are generally matched in order: once a request matches an earlier rule, later rules won’t select a different policy for it. A rule’s policy group may also have its own current selection, so check both whether the request matches the group and whether the group uses the intended server.
[Rule]
DOMAIN-SUFFIX,example.com,DIRECT
IP-CIDR,192.168.0.0/16,DIRECT,no-resolve
GEOIP,CN,DIRECT
FINAL,PROXY
The examples below explain the syntax; they aren’t a complete configuration to replace your existing file with. DOMAIN-SUFFIX,example.com,DIRECT matches the domain and its subdomains. IP-CIDR matches based on the destination IP’s network range, while no-resolve means this rule won’t trigger an additional DNS lookup. GEOIP,CN,DIRECT relies on an IP geolocation database, not the webpage’s language. FINAL,PROXY is the fallback policy when no earlier rule matches and should appear at the end of the rules.
- Check whether an earlier
DOMAIN-SUFFIXrule or rule set matches the destination; don’t look only at the end of the file. - If the request matches DIRECT, confirm that the current network can reach the destination directly. If it can’t, review the routing decision.
- If the request matches PROXY or a policy group, check which server is actually selected in that group.
- If the request matches REJECT, first confirm that blocking it is intended. Adjust the specific rule if needed; don’t delete all blocking rules.
In Data, review information related to the request you just made. If the current screen provides request details, check the destination, policy, and failure time. The hostname may change because of redirects or page resources, so use the request that actually failed rather than searching only for the hostname first entered in the address bar.
6. Narrow down the issue using the exact error
If request details or connection logs show one of the English errors below, keep the original wording for reference. The full error can include an address, port, or call details depending on the protocol and system. These strings identify error categories; not every platform displays the exact same message. Hide sensitive addresses and authentication details before sharing logs with your service provider.
Error: i/o timeout
Cause: A network read or write didn’t finish in time. The issue may involve the connection path, remote response, or DNS; this error alone doesn’t mean the password is wrong.
What to do: Keep the server details unchanged, test on another network, and compare the destination and time of the failed request. Then ask your service provider to check the remote server status.
Error: connection refused
Cause: The destination or an intermediate device actively refused the connection. Common causes include a port with no service listening, an incorrect port, or an access restriction.
What to do: On the server edit screen in Home, verify Address and Port exactly, then confirm which port the remote service is listening on. Don’t change it to 443 without checking.
Error: no such host
Cause: The current DNS lookup couldn’t find an address for the requested hostname. First determine whether it’s the server hostname or the webpage’s destination hostname.
What to do: Check the hostname spelling and current DNS path. If it’s the server hostname, resolve it before establishing the connection, then test proxy traffic.
If the error mentions an expired TLS certificate, a hostname mismatch, or the device time, check the system date and time, the SNI in the server configuration, and the remote certificate separately. Changing rule order won’t fix a TLS failure, and skipping certificate checks should not be a routine fix.
7. Restore your settings and retest
Once you’ve identified the issue, restore any temporary changes to Global Routing, DNS, and On Demand to their intended settings, then test the complete connection again. Working under Proxy doesn’t mean your everyday Config setup is fixed. Working on cellular doesn’t mean the original Wi-Fi issue is resolved.
Why does the VPN icon stay on while webpages keep loading?
The system icon shows the connection status on your device; it doesn’t confirm that traffic is reaching its destination. Return to Home, confirm the selected server, run Connectivity Test, then compare Proxy and Config with the same webpage to distinguish server issues from routing issues.
Do I need to reinstall Shadowrocket if only one website won’t open?
Check the actual requests for the website and its resource domains, along with the matching rules and policies. If other websites work, the device isn’t completely offline. Start with the specific destination and check whether the website itself is available.
Why doesn’t it work after a successful subscription update?
Return to Home and confirm which server is selected after the update. Then check its protocol, port, and authentication details. Retrieving subscription content and successfully authenticating with a server and forwarding traffic are two separate checks.
How can I test the connection if it reconnects as soon as I disconnect?
Go to Settings → On Demand, note the current settings, and temporarily turn off automatic connections. Then return to Home and disconnect. If it still reconnects automatically, check other automatic connection settings on the device. Restore your original On Demand rules after troubleshooting.
- Save the server selection and configuration that ultimately work so you don’t keep using a temporary selection made during troubleshooting.
- Retest the webpage that failed, a regular webpage, and the connection over both Wi-Fi and cellular.
- If the issue persists, note the device platform, network type, Global Routing mode, failure time, and sanitized error text to avoid repeating the same tests.