Shadowrocket DNS Troubleshooting: Websites Won’t Load Despite Normal Latency

Troubleshoot Shadowrocket DNS issues, understand DNS settings, and pinpoint why websites won’t load even when latency looks normal.

At a glance

Normal latency only shows that a test target is reachable under the test conditions. It doesn’t confirm that DNS resolution, rule routing, and web connections are all working. This guide covers troubleshooting Shadowrocket on iPhone, iPad, Mac, Windows, Android, and Linux. The interface paths refer to common iPhone and iPad layouts; look for the same option names on other platforms. First identify where the failure occurs, then check DNS, Global Routing, and IPv6 in the active Config. Change one setting at a time and retest the same domain.

1. Normal latency and a working website are separate checks

A latency result in Shadowrocket doesn’t verify that a web page loads completely. Depending on the test, it may only check whether a connection can be established or request a specific test address. A successful result doesn’t prove that DNS, TLS, and content requests for another site will all work. A web page may also load scripts, images, and sign-in requests from several domains; a failure on any one of them can leave the page blank or loading indefinitely.

Domain resolution Rule evaluation Outbound connection TLS handshake Content loading

This flow helps identify the stage where a request fails. It doesn’t mean that all DNS lookups always happen locally or strictly before every rule evaluation. Domain rules can match a hostname directly; IP rules may require a resolved address; and when using a proxy, the remote end may resolve the target domain. Use the actual logs to determine where the request failed.

2. Keep test conditions consistent and verify Global Routing

Before changing anything, note whether you’re on Wi-Fi or mobile data, the failing domain, the time, the active Config, the selected server, and Global Routing. If you already use your own provider and subscription, leave the server unchanged. Updating a subscription, switching servers, and changing DNS at the same time makes results difficult to compare.

  1. In Home, confirm that the connection toggle is in the expected state and that the system connection isn’t repeatedly disconnecting.
  2. Check Global Routing. For everyday rule-based traffic, use Config and confirm that the configuration currently enabled on the Config page is the file you just checked.
  3. Open the same full URL in the same browser, then test another domain known to work. Record whether all sites fail or only one does.
  4. Switch networks temporarily and test again. If the issue occurs only on one Wi-Fi network, check its DNS, captive portal, and outbound restrictions first.
  5. For a brief outbound comparison, switch to Proxy and restore the previous mode when you’re done. This may change the traffic path, so avoid testing with sensitive activity.
Global RoutingWhat this tells you during troubleshootingWhat you can’t conclude from it
ConfigTraffic is routed according to the rules in the currently active configuration.A configuration can exist without being active. Correct rules don’t guarantee that DNS is reachable.
ProxyA brief test in Proxy mode can help distinguish a routing issue from an outbound connection issue.It doesn’t mean every DNS request automatically uses the same proxy route.
DirectUse Direct mode to check whether the target is reachable on the current network.A failure in Direct mode alone doesn’t prove that a proxy server or subscription is unavailable.
SceneCheck the selected Scene and its conditions; its name alone doesn’t tell you which outbound route is used.Don’t assume it will always behave the same way as the active Config.

Takeaway: Keep comparison results that help isolate the cause

If the same domain works in Proxy but fails in Config, first compare the matched rules and DNS paths. Don’t delete all your rules or assume this automatically means you need a different DNS server.

3. DNS settings: what they mean and how to choose values

Open Config, edit the currently active configuration, and check the DNS-related entries under General. In a text configuration, these correspond to [General]. Note any global options in Settings too, but don’t confuse global settings with the contents of an inactive file. Options may appear differently across platforms; refer to the actual interface and the settings reference.

SettingWhat it meansHow to choose a value for troubleshooting
dns-serverSpecifies the DNS resolution source used by the configuration; the request path also depends on related settings and the outbound route.Record the current value first. Use system as a baseline for system DNS, then compare it with a resolution service confirmed to be reachable, changing only one variable at a time.
systemUses the DNS source provided by the current system network; it isn’t a fixed set of public DNS addresses.Results may vary when you switch between Wi-Fi and mobile data. If the issue goes away, that only shows that the two resolution paths differ.
DNS server addressSpecifies which DNS service receives the request.Enter the full address explicitly provided by your network administrator or service provider. Don’t enter a proxy server address, subscription link, or web URL.
DNS transport protocolTraditional DNS commonly uses UDP/TCP port 53; DoH typically uses HTTPS on port 443; DoT typically uses TLS on port 853.The protocol must match the server. A reachable port doesn’t mean that it provides the corresponding DNS service.
ipv6Relates to IPv6 use; check it alongside AAAA records returned by DNS and the actual IPv6 route.Keep the current setting initially. Only test with IPv6 temporarily disabled if you have evidence of an IPv6 failure, and record the result.

This snippet is only a baseline using system DNS, not a complete configuration file. Back up the current Config and change only the relevant line; don’t overwrite existing rules, servers, or policies. If system DNS on the current network is unavailable, this baseline will fail too.

[General]
dns-server = system

Encrypted DNS isn’t an independent channel that works without network access. If the resolver endpoint is specified by hostname, establishing a connection may require resolving that hostname first. A bootstrap lookup failure or routing loop won’t be fixed simply by switching to an encrypted protocol. Don’t replace the hostname with an IP to bypass hostname verification; this may cause TLS verification to fail.

4. Tell resolution failures, connection timeouts, and nonexistent domains apart

The messages below may appear in browser errors, system tools, or app logs; Shadowrocket may not display the same wording on every platform. When recording an error, note where it came from. A browser’s “DNS lookup failed” message and a client log showing a target connection timeout aren’t the same evidence.

Error: ERR_NAME_NOT_RESOLVED

Cause: The browser couldn’t get a usable name resolution result. The issue may involve the browser’s own Secure DNS setting, system DNS, or the domain itself.

Fix: Check the address spelling, try the same domain in another browser, and check whether the browser is configured to use its own DNS. Temporarily use a consistent resolution path, then test again.

Error: SERVFAIL

Cause: The DNS server returned a failure. Possible causes include an upstream outage, DNSSEC validation, or a problem with the domain’s authoritative DNS.

Fix: Query the same domain using another resolution path known to work. If only this domain keeps failing, contact whoever manages it. Don’t keep changing proxy protocol settings.

Error: NXDOMAIN

Cause: The response indicates that the queried domain doesn’t exist. Check for a typo, an expired domain, or an anomalous DNS response.

Fix: Verify the full hostname and test it across networks or DNS resolution paths. If the domain really doesn’t exist, changing DNS won’t bring it back.

Error: connection timed out; no servers could be reached

Cause: The query tool didn’t receive a response from the DNS server before timing out. This isn’t the same as the server explicitly reporting that the domain doesn’t exist.

Fix: Check the DNS server address, transport protocol, and network access rules, then try another network. For encrypted DNS, also check the TLS connection requirements.

On Mac, Windows, and some Linux systems, you can use the command below to check system DNS resolution. If the tool isn’t available, use the network diagnostics already built into your system. The example domain only demonstrates the query format; for troubleshooting, query the hostname from the failing page without its path or query parameters.

nslookup example.com

5. Once DNS resolves, check rules and IPv6

Getting an IP address only moves troubleshooting to the next stage. If the logs show successful resolution but the target connection fails, check the matched policy, selected outbound route, and target port. HTTPS typically uses TCP port 443, while HTTP/3 may use UDP port 443; successful DNS resolution doesn’t guarantee that either connection will work.

In the Config rules, DOMAIN-SUFFIX matches domain suffixes, while IP-CIDR and GEOIP involve the target IP. FINAL handles traffic that hasn’t matched an earlier rule. Rules are evaluated from top to bottom, and processing stops at the first match. Check for an earlier REJECT rule or whether the target is being sent to an unexpected DIRECT or PROXY policy.

  1. Check the actual failing domains: Sign-in and image domains may differ from the domain in the address bar. Check each one instead of adding a rule only for the main domain.
  2. Check whether local DNS resolution is needed: IP-based rules may trigger resolution. The no-resolve parameter changes how those rules handle resolution; it isn’t a DNS repair switch.
  3. Compare IPv4 and IPv6: Investigate IPv6 routing, outbound access, and related settings only if A lookups and IPv4 connections work while connections to the corresponding AAAA addresses keep timing out.
  4. Keep certificate verification enabled: For TLS errors, check the system date, time, and time zone, then verify the target hostname instead of disabling certificate verification.

Takeaway: Fix the layer supported by the evidence

If resolution succeeds and the correct rule is matched, investigate the target connection and outbound route. Consider changing IPv6 only if a direct comparison clearly improves the result. Don’t mask another issue by leaving a feature disabled indefinitely.

6. Verify changes and troubleshoot common issues

After each change, save the configuration, confirm it’s still active, reconnect, and test again. If needed, close and reopen the browser to reduce interference from existing connections and browser cache. DNS cache lifetimes vary, so old results may persist briefly after restoring a configuration. Don’t keep layering on new settings because of this.

Latency is consistently normal. Does that rule out a server issue?

No. A latency test and a connection to the target website have different requirements. First check how the latency test works, then inspect DNS resolution for the failing domain and the actual outbound logs. If resolution succeeds but the connection fails, check whether the selected server can reach the target.

Switching to system fixed it. Should I leave it that way?

Test it on your usual Wi-Fi and mobile network. If system DNS resolves reliably on both and meets your network requirements, you can keep it. If it works on only one network, check whether the original DNS service is reachable rather than assuming system DNS behaves identically on every network.

Only subscription updates are failing. Should I change DNS too?

First determine whether the subscription request failed during DNS resolution, timed out while connecting, or was denied access. Get the subscription link from your own service provider and confirm it’s still valid before troubleshooting based on the error. Changing DNS won’t fix denied access or an expired link.

The browser works, but other apps don’t. What should I do?

Check whether the browser uses its own Secure DNS, then identify the domains accessed by the failing app and the policies they match. A working browser doesn’t prove that system DNS works or that every resource domain used by another app is handled the same way.

A complete troubleshooting record should answer three questions: Did the name resolve? Which policy matched the request? At what stage did the connection fail? Narrowing down the issue one layer at a time is more likely to produce a reproducible fix than changing several DNS settings, protocols, and rules at once.

Download Shadowrocket