Official Manual Browse by symptom

Shadowrocket Troubleshooting Guide

First identify where the problem occurs, then change the relevant setting. Change one thing at a time and keep a configuration you can restore.

New to Shadowrocket? Start with the setup guide to import your configuration, choose a mode, connect, and verify the basics. This page is for troubleshooting after setup. If you have not established a working connection yet, you can also work through the checks in Chapter 1.

Connection switch won’t turn on: Check system permissions and connection conflicts

Tell a switch reset apart from a connection failure

Watch what happens when you tap the connection switch on Home. If it immediately turns off and the system does not show a stable VPN status, start by checking device permissions and local settings. If the switch stays on but websites will not load, the issue is elsewhere; go to the next chapter. Note whether a system prompt appears, its exact wording, and when it happens. An active switch does not prove the server has authenticated successfully: these are separate stages.

With the connection off, use your current network to visit a website that normally works without a proxy. Hotel, campus, and public Wi-Fi may require you to sign in through a web page first. Without ordinary network access, the connection is missing a necessary prerequisite. Complete network authentication in the system browser, then test Shadowrocket again. If your device has a manual network proxy configured, note its current value and check whether it is still appropriate for this network.

Check the VPN configuration and device management restrictions

The first time you connect, the system may ask permission to add a VPN configuration and request confirmation with your device passcode or biometric authentication. Approve only when you initiated the action. If you previously denied permission, try connecting again and read the system prompt. Then open system Settings, search for VPN, and check whether the configuration is present. Menu paths vary by system; using system search is more reliable than following a fixed path.

Another VPN, network filter, or security management feature on the same device may compete for the system connection. Temporarily turn off the other network feature and test Shadowrocket on its own. Change only settings you manage. If your device is managed by a school or employer, ask the administrator about restrictions; do not remove management profiles. A successful connection on another personal device helps narrow down the issue, but does not prove that system policies on this device are unrestricted.

If the system configuration exists but every launch reports that it is unavailable, first turn off On Demand and note its trigger conditions. Back up your current Config and any server details you need to keep, then try removing the corresponding VPN configuration from the system so the app can request permission again. Removing a system VPN configuration is not the same as deleting app data. Before any cleanup, make sure you have what you need to restore your setup, especially manually added entries not saved in a subscription.

Retry with a minimal setup and decide what to check next

Go to Home, select an existing server with complete settings, and temporarily avoid automatic selection or complex policies. Turn the connection off, return to the system Home screen, reopen the app, and test again. If the switch still resets immediately, restart the device and retry, noting any difference. Rapidly tapping the switch can overlap permission, startup, and shutdown processes, making the behavior harder to diagnose. Wait for the status to settle after each action before moving on.

If the error occurs only with a particular Config, keep that file and test a previously working copy. Check for recently pasted incomplete sections, deleted section headers, or references to policy names that do not exist. Do not clear all your rules. If a backup Config starts successfully, system permissions are likely working; focus on differences between the configurations. If none of your configurations start, then investigate system status or the app environment.

Connected, but no internet access: Separate network, mode, and rule issues

Start with comparable test results

Choose two websites you can usually access, one to check the device’s regular network and another to test the route through an existing proxy service. With Shadowrocket off, test both and note the results; then turn it on and repeat. Use the same network and targets for each round. Do not change Wi-Fi, server, and Config at the same time. If ordinary websites fail even with the connection off, troubleshoot the system network, sign-in page, or router before changing proxy settings.

If every website fails with the switch on, the checks differ from a case where only one site fails. For a complete failure, check the selected server, Global Routing, and DNS. If only some sites fail, focus on their rules, the target sites’ status, and the policies shown in connection records. One app failing does not mean the system connection is down: the app may have cached an old connection or use different domains and transport from your browser.

Use the three Global Routing modes to narrow down the issue

ModeWhat to checkWhat the result does and doesn’t tell you
ConfigCheck the connection policy defined by the active ConfigYou may need to check rule order and the final matched rule
ProxyTest access through the currently selected proxy routeA successful test does not prove your original routing rules are correct
DirectCheck behavior over a direct connectionDNS and the system network can still affect the result

Note the original mode, then switch briefly for comparison. If Proxy works but Config does not, check whether a rule sends the target to DIRECT, REJECT, or an unavailable policy. If Direct works but Proxy does not, focus on the server and proxy route. Restore the original mode after testing. Comparing modes helps narrow the cause but does not replace logs; DNS, cached connections, and restrictions on the target site can all affect the results.

Check the active Config and rule order

In Config, confirm which file is active, not just whether a file has been downloaded. Rules are generally matched in order, so put specific target rules before catch-all rules. A broad DOMAIN-KEYWORD or IP range rule may match first, preventing a more precise rule later from taking effect. Note the domain, matched rule, and final policy together; seeing the target name somewhere in the rule text is not enough.

[Rule]
DOMAIN-SUFFIX,example.com,PROXY
IP-CIDR,192.168.0.0/16,DIRECT,no-resolve
GEOIP,CN,DIRECT
FINAL,PROXY

This example illustrates rule order: a specific domain first, local network addresses next, and FINAL last. It is not a complete configuration for every network and should not replace other sections of your existing Config. In IP-CIDR, no-resolve prevents that address rule from triggering a domain lookup solely for its match; it does not fix DNS. GEOIP results depend on address classification data and do not guarantee a site’s location or service availability.

After making a change, close and reopen the target website. If needed, end the app’s existing connection, then check the new connection record. If the same domain resolves to multiple addresses, check whether each uses a different route. Local devices such as printers, home storage, and router admin pages may also require the right local route and permissions. Do not assume that restoring access to a public website means local network access is working; test them separately.

If the rule correctly matches PROXY but access still fails, continue with the chapters on server timeouts and DNS. For an overview of the initial settings, see the Global Routing guide. To route selected websites, read how to write rules for specific websites. Keep changes that resolved the issue and undo tests that did not help, so you can tell what is currently configured if the problem returns.

Server timeout: Check reachability, then protocol settings

First identify which step is timing out

Connectivity Test results only show whether a particular test completed under specific conditions. A timeout can occur during DNS lookup, connection setup, protocol handshake, or the target’s response; it does not necessarily mean the server is completely offline. Note the test method and target, then check a real website request. A single latency result does not represent download speed or prove that all app traffic can get through. If failures continue, use the error details to distinguish name resolution failures, refused connections, and timeouts.

Check the server address for extra spaces, line breaks, or an incorrect prefix. If the address field expects a domain, do not add a web page path. Make sure the port matches your existing service details. After pasting, look for non-English punctuation, trailing spaces, or missing digits. If the details came from a subscription, confirm that the selected entry belongs to the current subscription and is not an old entry that has been removed or replaced.

Check the required fields for the protocol

Shadowsocks requires the server’s matching encryption method and password. For VMess and VLESS, check the identity fields and relevant transport settings. For Trojan, check the password and TLS-related settings. Similar protocol names do not mean their fields are interchangeable. Do not enter arbitrary values just to make the editor save. If you are unsure about a manual entry, compare each field with your service details instead of copying settings from another entry and hoping they work.

For transports such as WebSocket, also check the transport type, Host, and path. Depending on the service configuration, the TLS server name and connection address may be entered separately; do not force them to match just because they differ. For certificate or handshake errors, first check that the system time is correct and the name is complete, then verify the server configuration. Disabling certificate verification can hide the problem and change the connection’s security boundary, so it is not a routine fix.

Connections using UDP, including Hysteria2 and WireGuard, require checking whether the current network allows that traffic. A working TCP-based entry does not prove that the UDP route works. For WireGuard, also check the peer address, key pairing, tunnel address, and allowed routes. Never share your private key to get help troubleshooting. Change a setting only when you understand what it does; ask your own service provider about missing details.

Compare networks to narrow down the cause

On the same device, with the same server and Config, compare Wi-Fi with mobile data. If timeouts occur on only one network, check its sign-in requirements, UDP restrictions, IPv6 route, or router settings. If multiple networks fail but another existing server works, focus on the original server’s status and settings. Keep mobile data usage in mind during comparisons; there is no need to transfer large files continuously just to test connectivity.

A server domain with both IPv4 and IPv6 addresses may have only one unreachable route. Check the connection record to see which address family was used, then test that path specifically. Note the original address-family setting before changing it temporarily, and restore it afterward. Replacing a domain with an IP address can break TLS naming or services that depend on the domain, so it is not a general-purpose fix.

An immediate connection refusal often points to a service that is not listening, a mismatched port, or an intermediate device actively rejecting the connection. A timeout after a long wait more often points to packet loss, a filtered port, or an unresponsive target. These are diagnostic clues, not definitive conclusions from a single error. Record the failure stage, network type, and time of day; this is more useful than repeatedly increasing the timeout. A longer wait cannot fix incorrect credentials.

Subscription update failed: Check the link, download, and content format separately

First identify the type of failure

Subscribe update failures generally fall into three categories: the URL is unreachable, the server returns an error, or the content cannot be parsed by the selected import method. Each requires a different fix. Note the exact message and whether the failure occurred while adding or updating. An update can fail even when older servers still work; the app may retain previously imported data. Do not delete the entire subscription group first, or you could lose working entries and a useful point of comparison.

Copy the complete subscription URL again from your own service provider, checking for spaces, quotation marks, or line breaks at either end. Some links include query parameters, and missing one character can change the authentication details. A short text preview in a messaging app may not show the full URL. Do not use a web admin, sign-in, or subscription information page as the subscription content URL. Check your own service instructions to confirm the supported format.

Check whether the URL returns valid content

If the message says the request timed out or the host could not be resolved, first confirm that the device’s regular network works, then check DNS. If updating depends on the current connection and the current server has stopped working, you may end up in a loop: you need to update to connect, but cannot connect to update. After noting the original settings, try another existing route that works and update again. Retrieval behavior can vary by subscription and app settings, so use the actual request record to guide your checks.

Errors such as an unauthenticated request, access denied, or missing resource point to a URL that may have been reset, expired authorization, or server-side rate limits. Error codes are clues; your own service provider can confirm the cause. Repeated update attempts may trigger further restrictions and obscure the first failure. Keep a clear record of one error, wait a reasonable interval, then retry; this is more useful than refreshing continuously.

If content is retrieved but the format is reported as unsupported, check that the import type matches the content. A server subscription, Config file, and web page are different kinds of data. Importing a rule configuration as a server subscription, or parsing a web page’s sign-in prompt as a server list, can fail. Do not judge the format by the file extension alone; check the actual server response and the import method you used.

Entries change after an update: Protect manual content

If entries disappear after a successful update, the server-side content may have changed, or filters, collapsed groups, or the current view may be hiding them. First check the subscription group name and update status, then see whether search or filtering is enabled. Entries managed by a subscription may be replaced during updates, so manual changes to their names or settings may not persist. If you need to keep custom entries long term, back them up separately using a management method supported by the app.

If entries with the same name appear after an update, do not assume they are duplicates based on the name alone. Compare their subscription group, protocol, address, and port to see whether they serve different purposes. Back up before cleaning up and keep the entries that currently work. If the list looks normal but connections fail, the issue has moved from retrieving the subscription to server availability. Repeated updates usually will not fix incorrect protocol settings.

Subscription URLs usually contain credentials. Do not post them in public forums, screenshots, or online conversion tools. If you need to show the format, include only a structure without real tokens. Exported files may also include server passwords and subscription URLs, so store them somewhere you control. Before transferring data between devices, confirm the destination device and what you are sending. After importing, delete temporary transfer copies to avoid leaving sensitive data scattered around.

Recovery means the update request succeeds, entries belong to the expected subscription, and at least one existing server has been verified by accessing a real target. Seeing “Update complete” alone does not prove the connection works. To move your data, see configuration backup and migration. For quick answers about importing and groups, visit the FAQ. This site explains how to use data you already have; it does not provide subscriptions or recommend services.

Slow connection: Check throughput, latency, and app response separately

Start by defining what “slow” means

A long wait before a page starts loading, a consistently low download rate, and frequent video buffering are different symptoms. The initial wait may involve DNS, the handshake, or the target’s response. Slow sustained transfers may involve bandwidth, packet loss, or server load. Buffering may also depend on playback behavior and the content service. Write down the specific symptom and when it occurs instead of just noting that things feel slow; otherwise, it is hard to tell whether a change helped.

Choose the same target, network, and time window for each comparison. First test regular network access with the connection off, then test one existing server, and finally compare another existing server. Do not switch routing modes while also changing the test file and Wi-Fi. Tests use bandwidth too, so pause other downloads and cloud sync. Be especially mindful of mobile data, and avoid prolonged high-volume testing when a basic connection check will do.

Understand what test results can tell you

Connectivity Test response time is not the same as real-world throughput. A small request that completes quickly may use a route with limited bandwidth, while a slightly slower route may sustain a steady download. A test target that differs from the actual website can produce different results. Check whether the page loads, whether the connection keeps retrying, and whether transfers remain steady. Do not switch servers repeatedly based on a single latency value in the list.

If things are just as slow with Shadowrocket off, check Wi-Fi signal, router load, and the access network. Test closer to the router and, if practical, compare with another network. Photo backups, system downloads, or internet sharing can also use bandwidth. Pause those tasks before testing so simultaneous transfers on the device are not mistaken for proxy performance issues. Before restarting the router, consider whether it will disrupt other people.

Check the proxy route and connection method

If only proxy traffic is slow, check the connection record to confirm which server is actually selected and whether the traffic goes through an additional forwarding route. Automatic selection can vary with test conditions, so temporarily pin an existing entry for comparison. The route from the server to the target also affects results; a nearby location does not guarantee a better connection. Location names alone cannot explain congestion on an intermediate route or restrictions on the target side.

If UDP-dependent apps behave poorly while regular websites work, check whether the current server and access network support the required traffic. No protocol is universally “faster.” Protocol settings, server implementation, packet loss, and device conditions all affect performance. Be especially cautious with MTU: test it only when you have clues such as large requests hanging while small ones work, and a clear way to compare results. Do not apply a fixed number from the internet without testing.

If only one website is slow, check whether its main domain and static assets use different policies. The page may load while images spin because their domains use another route, or the content service itself may be slow. Check the failing resources individually instead of switching the entire Config to proxy everything. A brief comparison using Proxy may help isolate the issue, but return to rules that match your actual needs once you have finished testing.

Wrap up with repeatable results, not a one-off impression

Repeat the same test at two different times and note whether the slowdown happens only during peak hours. If the same server consistently performs poorly across multiple networks and targets, share the times, symptoms, and test differences with your own service provider after redacting sensitive details. If performance drops only when the device heats up, let it cool and stop unnecessary background tasks before testing again. Sustained performance issues can involve temperature and system scheduling, not just the server.

A useful conclusion is specific: “This existing server is consistently slow for this target on this network,” not “the whole app is slow.” Keep the smallest scenario that reliably reproduces the issue and undo settings that did not help. If performance recovers, note the server, mode, and network that worked, but do not expose credentials in public posts. Speed issues are rarely fixed by a single switch; controlled comparisons are more reliable than piling on supposed optimization settings.

DNS issues: Check resolution, addresses, and routing

Separate lookup failures from connection failures

DNS resolves a domain name to an address; the device must then establish a connection and complete the application request. So a successful lookup does not mean the website is reachable, and a website that will not load does not necessarily indicate a DNS problem. Identify whether the error occurs during name lookup, connection to the address, or the TLS handshake. If the error specifically reports name resolution, check DNS settings. If an address is returned but the handshake fails, also check the server name, certificate, and connection route.

Do not rely solely on the result of visiting a website by IP address. Many HTTPS sites use the domain name to select a certificate and site content; entering an IP directly may show a certificate error or a completely different page. A better approach is to check the lookup result for the same domain, the address actually used, and the request record, then compare with another network or a previously working configuration. Account for caching so you do not mistake an old cached result for the result of a recent change.

Check the DNS route and circular dependencies

First note the current DNS-related values in Config and check whether they override the system’s resolver. If the configuration uses a resolver that must be reached over the network, make sure the route to that resolver works. An unreachable resolver can cause every domain lookup to fail. Do not keep adding unknown addresses. Instead, return to the last known working settings and change one option at a time, verifying each change with an actual lookup.

If the proxy server address is a domain, its address must be resolved before the proxy connection can be established. If DNS requests themselves depend entirely on a proxy route that is not yet available, this can create a circular dependency. Check how the initial lookup is performed instead of repeatedly switching the target website. Config files handle DNS routes differently; check the relevant sections in the file rather than assuming that one DNS address makes the entire setup work.

Check whether rules change DNS behavior

DOMAIN, DOMAIN-SUFFIX, and DOMAIN-KEYWORD match domain names, while IP-CIDR and GEOIP involve address matching. With the wrong order, address-based rules may trigger lookups unexpectedly or determine a policy too early. Check the match record to see whether the target was handled by a domain rule or an address rule. A rule specifying DIRECT does not necessarily mean the DNS lookup follows the same route; observe name resolution and data connections separately.

[Rule]
DOMAIN,example.com,PROXY
DOMAIN-SUFFIX,example.com,PROXY
IP-CIDR,192.168.0.0/16,DIRECT,no-resolve
IP-CIDR6,fc00::/7,DIRECT,no-resolve
FINAL,PROXY

This example includes an exact domain, a domain suffix, and a local address rule to illustrate match types; it does not mean you should add all of them to your current file. DOMAIN matches a specified domain, while DOMAIN-SUFFIX also matches its subdomains. The sample range in IP-CIDR6 does not cover every local IPv6 setup. Confirm the actual target before adding rules, and avoid broad address ranges that could override existing policies.

If public websites resolve but home storage or internal workplace domains do not, check whether the network requires its own internal DNS. Public resolvers usually do not know private names. An internal name may also stop working when you switch networks. First confirm that the device is on the internal network and has permission to access it, then follow the network administrator’s requirements. Do not include internal domain names or network structure in public support posts.

Verify cache and dual-stack results

After a change, stop existing connections and make a fresh request; if needed, refresh the cache using the method provided by the app or system. IPv4 and IPv6 results can both be available, and a problem with only one route may cause intermittent delays or different results across apps. Check which address family the connection actually uses, then temporarily compare one setting at a time. Permanently disabling one address family is not a universal fix for dual-stack issues. Restore unrelated settings after testing.

Recovery means DNS resolution succeeds, the returned address is expected, the connection policy is correct, and the website completes its request. If only the first two are true, return to the connection or server timeout chapters. Note the final DNS settings and the networks they apply to, so you do not carry them over incorrectly when switching to an internal network. See the FAQ for quick definitions and common symptoms. For complex issues, keep separate evidence for the lookup and connection stages.

Unusual battery drain: Check reconnects, background traffic, and test tasks

Start with total battery use and usage context

An app’s percentage on the system battery screen is relative; it does not mean the app used that same percentage of the battery’s total capacity. When overall device use is low, a network feature running continuously may account for a larger share. Check the time range, screen use, background activity, and battery change before deciding whether the drain is truly unusual. Do not diagnose a problem from one screenshot, and do not compare charging data directly with battery-only use.

Note whether you were on mobile data, the signal was weak, the device was warm, or video playback, downloads, or cloud sync were running. Shadowrocket handles network connections from other apps, so system usage figures need to be considered alongside actual traffic. Higher battery use during heavy transfers does not automatically indicate a configuration problem. More concerning is a device that gets warm or loses significant charge while idle with very little traffic.

Stop unnecessary continuous tasks first

Turn off manually started repeated tests and check for overly frequent automatic connectivity tests or subscription updates. Restore detailed logging used during troubleshooting to your everyday setting once you have the evidence you need. Continuous tests increase network wake-ups and requests, especially when the signal fluctuates. Do not disable everything at once. Pause one type of task, compare over the same time period, and then decide whether further changes are needed.

If connection records show repeated failures followed by immediate retries, identify the domain, server, or app triggering them. A background app may keep retrying after network failures, causing the device to communicate while it appears idle. Temporarily close the app and see whether retries stop. If the same request keeps switching between two routes, check automatic selection, rule conflicts, and network changes. Shortening the timeout can make retries more frequent, not solve the problem.

Check On Demand and network switching

On Demand starts connections when its conditions are met. If the device frequently switches between Wi-Fi and mobile data, or the trigger conditions do not match the current environment, the connection may repeatedly start and stop. Note the original conditions, briefly turn off On Demand, and test by connecting manually. If reconnecting and battery drain improve, restore conditions one at a time to find the combination causing the cycle. Avoid keeping a complex setup that you cannot explain.

System sleep, network address changes, and access point changes can also trigger a reconnect. First distinguish a normal reconnection after network recovery from repeated failures in a short period. Do not casually shorten keepalive intervals for protocols that need to maintain communication; more frequent traffic can increase battery use. For settings such as WireGuard, follow your own service requirements and change values only when you understand their purpose and can compare before and after. Fixed values are not a universal battery-saving solution.

Compare battery use under similar conditions

With the same network, similar signal strength, and similar background app activity, compare three scenarios: disconnected, connected but idle, and normal use. Keep the screen state as consistent as possible and observe over a long enough stable period. Do not browse with the screen on in one round and leave it locked in another, then attribute the entire difference to the app. Keep the server and Config fixed during comparisons to avoid automatic changes, and avoid leaving the log screen open throughout the test.

If only one existing server causes repeated reconnects, check its service and network route in the server timeout chapter. If the same issue occurs with every server and regular network access is also unstable, focus on the device’s network environment. If the device is too hot, stop demanding tasks and let it cool naturally before testing again. Background activity after a system update may affect short-term results, but confirm persistent issues with repeated tests under the same conditions.

Issues after an update: Back up first, then compare system and configuration changes

Identify what changed

First determine whether the issue followed a Shadowrocket update, a system update, a Config update, or new subscription content. If several things changed on the same day, do not assume the last update you noticed caused the problem. Note the last working network, mode, and server, then list changes in order. If you need to report version information, check the version currently installed on your device; do not infer current behavior from old version descriptions online.

Before restoring anything, export the Config and manually added server details you need to keep, and make sure the backup file can be opened. Subscription URLs, rule files, and app settings are separate kinds of data; saving one does not mean the others are backed up. Before migrating or resetting, note your current Global Routing, On Demand, and DNS settings. For iCloud, confirm syncing is complete and keep a separate copy so an accidental deletion does not propagate to other devices.

Start with steps that do not risk your data

Turn the connection off, reopen the app, and check system permissions and the active Config. If needed, restart the device, then test again on a network that previously worked. Individual apps may retain old connections after an update; close those sessions and make new requests. Check the proxy route only after confirming the basic network works. Do not start by deleting the app; reinstalling can erase the very state and error clues you need for comparison.

If a backup Config works but the latest file fails, compare recently changed sections, rule order, and policy references. Check whether a server subscription was mistakenly imported as a configuration, or whether updating a remote file overwrote local customizations. Restore one known change at a time and test again. Find out what an unfamiliar setting does before changing it. Do not switch everything on just because the issue appeared after an update; that can mix the original problem with new ones.

Limits of app and installation recovery

If you need to get the app again, choose the correct device option on the download page. The iPhone and iPad app is a one-time $2.99 purchase on the App Store; the Mac, Windows, Android, and Linux versions are free. If you already bought the iPhone or iPad app, check your purchased items using the original purchase account. See device replacement and redownloading for details. Getting the app again will not fix an expired subscription or invalid server credentials.

Desktop and Android installers are linked from the site’s download page and served from the site’s /dl/ paths. If Mac blocks the first launch, check the app under Privacy & Security and choose “Open Anyway” if appropriate. If macOS says the app is damaged, first confirm the file’s source, then follow the download page’s instructions for handling the local file. For Windows SmartScreen, verify the source and choose “Run anyway” under “More info.” On managed devices, follow your administrator’s requirements rather than changing security policies yourself.

On Android, allow the browser or file manager that will install the app to install unknown apps; you can revoke this permission afterward. On Linux, AppImage files need executable permission, while deb and rpm packages should use the package tools for your distribution to handle dependencies. Find commands and system permission steps on the download page. Do not blindly run long commands from unknown sources. Record installation issues separately from connection problems after launch; installation permissions are not fixed through network settings.

Verify the complete workflow after recovery

After recovery, check permissions, subscription, selected server, Global Routing, DNS, and actual access—in that order. Confirm the basic connection with a known-working minimal configuration, then restore On Demand, custom rules, and automated tasks one at a time. Run the same test after each change to find where the issue returns. Importing all the old settings at once only leaves you with the same symptom and no way to identify which change caused it.

If the issue still reproduces reliably in the minimal setup, note the system type, network type, steps, exact error, and expected result. Include only necessary fields from a redacted configuration snippet, and hide account details and subscription URLs in screenshots. A clear description of what happened after a specific step is more useful than simply saying “it stopped working after an update.” Once the issue is fixed, save a new working backup and keep the old one for now so you do not overwrite your only recovery point.

iPad-specific checks: Sync, permissions, and local network access

The same subscription does not mean the same setup

If an iPhone works but an iPad does not, first confirm that both devices actually use the same server, Config, and Global Routing mode—not just a subscription with the same name. Each device may have different selections, DNS settings, or On Demand conditions. Note the differences side by side, then test on the same network and target. Do not assume every setting on the phone has automatically synced to the tablet before confirming what is included in sync.

On iPad, a split view or sidebar may show a list and details at the same time. Viewing a Config’s contents does not necessarily make it the active file, and viewing a server’s settings does not mean it is selected. Check the active selection in Home and Config before reconnecting. If rotating between landscape and portrait only moves a control, find the same function in its new location rather than importing all your data again.

Keep separate backups before syncing or migrating

Before migrating through iCloud, check the accounts, app sync permissions, available storage, and network status on both devices. Sync can take time; content that has not appeared yet is not necessarily lost. Save important settings from each device separately, then wait for sync to settle. If you need to restore immediately, manually import a file you exported yourself. After importing, check the rules, servers, and active mode rather than relying on matching file names.

For duplicate entries after migration, check their subscription group and specific settings first. Do not delete entries in bulk just to tidy the list; deletion may sync to other devices. For testing, select an existing server you know works and wait for a stable connection before cleaning up. See configuration backup and migration for the full process. Treat export and sync as separate recovery methods; one does not replace verification of the other.

Check local network permissions and access requirements

On cellular iPad models, check the app’s mobile data permission. On Wi-Fi models, focus on the current access point and any sign-in page. Wireless networks with the same name may still place different devices on isolated networks. If websites work but printers, home storage, or shared computers do not, confirm that the device is on a network that allows local access and check the app’s local network permission in system settings. Proxy rules cannot by themselves remove device isolation on a public network.

When local network resources are unreachable, test the device name and actual local address separately. If the name fails but the address works, focus on internal DNS or name discovery. If both fail, check network isolation, routing, and the target device’s status. Features such as screen casting or service discovery may also rely on multicast; adding a DIRECT rule for one address does not guarantee that every discovery mechanism works. Describe “the device is not discovered” separately from “the device was discovered but will not connect.”

Compare behavior after locking, using split view, and switching networks

If the issue occurs only after waking from the lock screen, note whether Wi-Fi disconnected, the address changed, or On Demand triggered again. Compare with a manual connection first, then restore automatic conditions. An app’s old session may no longer work after system sleep, so make a fresh request. Do not keep the screen on all the time to hide the problem; that changes battery and network behavior and does not show where the wake-up process failed.

When using split view or switching between apps, one app may display cached content while another starts a new connection. That alone does not prove that one app is supported and the other is not. Refresh the same webpage and check whether both requests match the same policy. Private browser sessions, other system network privacy features, and app-specific proxy settings can also change test conditions. Note them separately and compare them temporarily if needed.

Finally, verify the iPad as a standalone device: system permission is granted, an existing subscription updates successfully, server settings are complete, the Config is active, DNS works, and the target resource is reachable. An iPhone working is a useful comparison, but cannot replace these checks. If the issue remains, return to the relevant symptom chapter on this page. For a missed initial setup step, see the setup guide. Keep a redacted list of differences and a working backup so you have a clear starting point next time you migrate, update the system, or switch networks.

Download Shadowrocket