A connected switch does not guarantee that the selected server can forward requests or that the destination domain resolves correctly. This guide is for iPhone and iPad users who have imported their own configuration but can't access websites. Start by comparing the same URL with the connection off, then check the server, routing mode, DNS, and rules one at a time. Test after each change, then restore the original setting.
1. Server: Check the Selected Entry First
First narrow down the problem: do all websites fail, or just one? Choose a URL that fails and another that normally opens directly on your current network. Note whether you're on Wi-Fi or cellular. Keep the network, browser, and URLs unchanged during each round of testing. If the page may be cached, reload it; don't rely on a page that was already open to confirm the connection is working again.
Note the Original Settings
In
Home, note the selected server andGlobal Routingmode. InConfig, note the active configuration. If the list has multiple entries with the same name, compare their address, port, and protocol before switching; don't rely on the name alone.Test with the Connection Off
Turn off the connection switch in
Home, then load the comparison URL. If the switch turns itself back on, check whetherSettings → On Demandis enabled. If your device allows it and doing so doesn't conflict with management requirements, temporarily disable the feature for testing, then restore its original setting. If the comparison URL also fails while disconnected, first check whether the network requires sign-in and whether it has internet access.Test the Selected Server
In
Home, check the selected entry, runConnectivity Test, and review its result. A timeout means only that this test didn't finish within the time limit; one timeout alone doesn't prove the configuration is invalid. A latency result doesn't guarantee the failing website is reachable either, since the test may use a different destination.Retest with an Existing Entry
If you already have a backup entry known to work, keep the network and routing mode unchanged. Switch only the server, reconnect, and load the same URL. If switching servers resolves the problem, check the original entry's parameters and service status first. If nothing changes, select the original entry again and check the routing mode.
If the problem started right after a subscription update, compare the server that was actually selected before and after the update. Seeing entries still listed doesn't mean the configuration stayed the same. Use only service details you already have; this site does not provide, sell, or recommend nodes or subscriptions.
2. Global Routing: Compare Proxy, Direct & Config
Check the current mode under Home → Global Routing. The comparison below applies to devices where you can temporarily change the routing mode. If the device is managed, or test traffic must not be sent directly, don't switch to Direct. Test only with public pages that don't require sign-in, then restore the original mode.
- Only
Proxyfails: First check whether the comparison URL loads inDirect. If it does, recheck the selected server and its route to the destination. Don't start by changing routing rules to fix a failure inProxy. - Only
Configfails: If the same URL works in the other two modes, note the actual hostname that failed. Focus on the active configuration's rules and DNS rather than re-entering the server details. - You were using
Scene: Note the network and scene conditions when the problem occurred, then compare with a fixed mode. If the failure appears only after switching from Wi-Fi to cellular, test each network separately so a scene change isn't mistaken for a server failure.
Diagnosis: Separate Direct and Proxy Paths
If a website loads with the switch off but fails with the switch on in Direct, leave the server parameters unchanged and check DNS settings while the connection is on. The failure can't be attributed to the proxy server alone.
3. DNS: Tell Resolution Failures from Connection Timeouts
First note the URL shown in the browser's final error. The URL you entered may redirect, and a homepage, login page, or image may use a different domain. If the active configuration sets dns-server under [General], check the file currently in use in Config and note its original value. The filename in the list alone doesn't confirm that its changes are active.
Error: Could not resolve host
Cause: This common diagnostic message indicates that the tool couldn't resolve the hostname to an address. It is not specific to a message in the Shadowrocket interface.
Fix: Check the failed hostname for typos, then restore the previously working DNS setting in a copy of the configuration and retest the same URL.
Error: Connection timed out
Cause: The connection wasn't established within the time limit. The server, network path, or destination port may be involved; a timeout alone doesn't prove that DNS is the problem.
Fix: If the diagnostic log shows the destination address, leave DNS unchanged. Compare Proxy and Direct using the same URL, then check the failing path.
- Check the resolver endpoint: Traditional DNS uses port
53by default, while DNS over HTTPS typically uses443. If the configuration uses an HTTPS resolver, check the original URL's hostname, port, and path. A reachable port doesn't guarantee that the resolver will return a valid response. - Check both hostnames: If the proxy server address is a domain name, that name must also resolve. If only entries using a domain as the server address fail, check whether that address resolves; don't check only the website hostname in the browser.
- Undo ineffective changes: Change one DNS setting at a time, then reload the same URL and check whether the error changes. Don't replace an HTTPS URL with a bare IP as an equivalent test: certificate validation and site selection may change. If the change doesn't help, restore the original configuration and check the rules.
4. Rules: Check the Actual Hostname and First Match
Return to Config mode and inspect [Rule] in the active configuration. Follow the rule order from top to bottom and find the first applicable rule. Don't search only for the destination domain: a broader rule may match first, even if the configuration contains the domain rule you expected. The reserved example domain below illustrates rule precedence; it isn't a configuration for everyday browsing.
[Rule]
DOMAIN-SUFFIX,example.com,REJECT
DOMAIN,docs.example.com,DIRECT
FINAL,PROXY
docs.example.commatches the firstDOMAIN-SUFFIXrule, soREJECTapplies first and the more specificDOMAINrule below it won't change the result. To make this hostname an exception, move the exact-match rule above it in a copy of the configuration.- If the active configuration includes
IP-CIDRorGEOIP, check their position and whether the request has a destination IP available for matching. A domain missing from the configuration doesn't automatically useFINAL;FINALapplies only to requests that don't match an earlier rule. - If only part of a page loads, check the hostnames for the main page, redirects, and failed resources separately. Don't delete an entire suffix rule just to allow one subdomain. Test with the most specific exception first, and confirm that other requests still follow the original policy.