Use this guide when Shadowrocket is on but your browser can't load pages. A connected status means the system tunnel is running; it doesn't guarantee that the proxy server, DNS resolution, or destination website is reachable. Use the same network and test domain, changing one setting at a time, to narrow the issue down to your network, proxy route, DNS, or a matched rule. Then restore your original settings.
Check Server Connectivity and Latency
First, confirm that your current network can reach a public website you normally access directly. Then check the server actually selected in Home. Don't use work portals or online banking for direct-connection tests, and don't change Wi-Fi and servers in the same comparison. The steps below use the iPhone / iPad interface.
-
Record the Current Settings
In
Home, note the current server andGlobal Routingmode. Then copy the failed hostname from your browser's address bar. Record only the server name or a personal label—never copy passwords or subscription links containing access credentials. -
Check the Direct Connection
In
Home, disconnect and use your current network to open a public page you know is reachable directly. If the connection turns back on automatically, go toSettings→On Demandand check for a setting that takes over on your current network. Note its original value before temporarily disabling it. If the public page still won't load while disconnected, check for a Wi-Fi sign-in page, cellular data permissions, or a network outage. -
Test the Current Server
Return to
Homeand run a latency test for the current server. Note whether it returns a time in milliseconds or times out. A response means the probe succeeded this time; it doesn't guarantee the original page will load. A timeout only means the probe didn't finish in time—it doesn't prove the server is unavailable. -
Check the Connection Details
If your service provider offers another server, test it separately on the same network. If only the original server repeatedly times out, check its address, port, and protocol settings. If you need to refresh your subscription, get it from your service provider. If all servers time out while direct access works, test the same server again on another network you're permitted to use.
Compare Global Routing Modes
In Home → Global Routing, check the current mode. This setting determines how requests are routed; a latency result for a server doesn't mean the current webpage is using it. Keep the connection on, retry the same public domain for each comparison, and restore the original mode when you're done.
| Mode | What It Helps Check |
|---|---|
Direct | Check whether a public test site can connect directly while Shadowrocket is on. This mode doesn't test sites that require a proxy. |
Proxy | Check the proxy route through the currently selected server. Even if a latency test gets a response, a webpage can still fail because of DNS resolution, the server's exit route, or its connection to the destination. |
Config | Route requests according to the current configuration. If a domain works under Proxy but fails in this mode, check the matched rule, the selected policy group, and DNS settings in the configuration. |
Scene | Apply the routing mode set for the current scene. If the issue appears only after switching between Wi-Fi and cellular, check which scene was activated and retest with a fixed mode. |
Diagnose the Failing Route First
If a public test page loads under Direct but not Proxy, investigate the proxy route first; don't start by reordering rules. If it fails only under Config, check how the configuration routes that request.
DNS Resolution and Browser Errors
When DNS lookup fails, the request may never reach the destination. If the server address in Home is a hostname, its lookup could fail too; if the server address is an IP, that doesn't rule out a DNS issue with the webpage's domain. Note the browser's exact error and check whether several unrelated domains fail or just one site.
In Config, check the configuration currently in use for a custom DNS server and any separate DNS settings for the affected domain. If the custom server uses DoH, verify the full HTTPS URL, including the path specified by the provider—not just the hostname and port. Keep a recoverable copy of the configuration before making changes. When retesting, don't also change the server or Global Routing mode.
Browser error: ERR_NAME_NOT_RESOLVED
Cause: The browser couldn't get a usable result for the hostname. The name may be misspelled, or the current DNS route may have a problem.
What to do: Check the address bar character by character, then try another public domain. If both fail, first check whether the custom DNS server in the current configuration is reachable.
Browser error: ERR_CONNECTION_TIMED_OUT
Cause: The connection wasn't established before the wait timed out. This error alone can't tell you whether the server, a rule, or the destination site is responsible.
What to do: Note the mode in which the error occurred. On the same network, compare the results under Direct and Proxy, then check the server or rules involved in the failing route.
Retry from the address bar instead of relying on an old tab that still shows an error. Don't replace the domain with an IP as an equivalent HTTPS test: the certificate and site selection may depend on the original hostname. Failure to open the IP doesn't prove there's a DNS problem.
Rule Order, REJECT, and Unavailable Policies
Check rules first only after confirming the issue occurs in Config mode. Open the active configuration in Config and review its rules from top to bottom. With standard sequential matching, the first matching rule determines the policy for a connection; FINAL handles requests that didn't match earlier rules. The example below illustrates match order only—don't overwrite your existing configuration.
[Rule]
DOMAIN,example.com,DIRECT
DOMAIN-SUFFIX,example.com,REJECT
FINAL,PROXY
- In this example,
example.commatchesDOMAINfirst, so it usesDIRECT.www.example.comdoesn't match the first line; it then matchesDOMAIN-SUFFIXand usesREJECT. If the homepage loads but images or sign-in fail, check the hostnames those requests use. - If the
DOMAIN-SUFFIXrule comes first, it will also rejectexample.com. If an exception for a domain doesn't take effect, check whether a broader rule above it takes precedence. Don't just check whether the exception is present. IP-CIDRmatches IP address ranges;GEOIPmatches an IP's geographic location. If no domain rule matches, check the actual destination IP and subsequent rules. Don't infer whichGEOIPentry will match from the domain suffix.- A matched rule doesn't mean the route is working. If the match points to a policy group, check which server the group actually selects. A response from a different server tested in
Homedoesn't prove that the server used by the policy group is working.
If you confirm a match to REJECT, first check whether the block is intentional. To test a suspected false match, change only the rule related to the failing hostname in a copy of the configuration, then retry the original page. Don't change FINAL to route everything directly just to fix one site.
Restore Settings and Record Your Findings
After comparing results, restore the original Global Routing mode, server selection, and any temporarily disabled On Demand settings. Keep only fixes you've verified independently. With your usual settings restored, retest the original failing page and a page that previously worked to confirm the issue is fixed—not merely bypassed by a temporary mode.
- If the public test page still won't load after disconnecting Shadowrocket, note your network type and whether another network works, then troubleshoot the network connection first.
- If multiple sites fail only through the same server, give your service provider the test time, network type, server label, and latency result.
- If the issue occurs only under
Config, record the failed hostname, matched rule, and resulting policy. Remove subscription links, passwords, and access tokens before sharing troubleshooting notes.