01 / Permissions
The connection switch won’t turn on or turns off immediately
If the switch won’t stay on, first check whether the system successfully created the network extension or VPN configuration. Don’t start by changing the server password. The Home connection switch asks the client to handle relevant network traffic; system approval, extension startup, and establishing a server connection are separate stages. If it immediately returns to Not Connected, determine whether it stopped at the permissions stage or the connection stage. Note any system authorization prompt, authentication prompt, or error text—the exact wording is more useful than simply saying it “won’t turn on.”
Check system authorization and configuration first
On the first connection from an iPhone or iPad, allow the system to add a VPN configuration and complete any required authentication. If you canceled the prompt, return to Shadowrocket and try connecting again; don’t rapidly tap the switch repeatedly. Then check the corresponding entry in the VPN section of system Settings. The menu name can vary by OS version, so you can also search Settings for VPN. A configuration entry doesn’t guarantee that the server is reachable, but a latency test can’t replace a permissions check if no configuration was created.
On a school- or organization-managed device, check whether management policies restrict adding VPN configurations, changing network settings, or running network extensions. A disabled button on a managed device is not the same issue as a temporary error on a personal device. Save any restriction message and contact the device administrator; don’t repeatedly remove management profiles. If the system requires the device to be unlocked before settings can be changed, unlock it and retry in the foreground rather than mistaking a locked-state failure for a damaged configuration.
Rule out connection conflicts, then distinguish startup from handshake failures
Disconnect other active VPN sessions and disable features that may be taking over the same system network channel, then test Shadowrocket. The goal is to reduce the number of factors changing routes at once—not to uninstall every networking tool. If the switch stays on after you disconnect another session, restore your previous settings one at a time to identify the conflict. A VPN indicator in the status bar only means that a system connection exists; check the name and status in the system VPN list to confirm it’s the session you expect.
If the switch stays on for a few seconds and then disconnects, check whether the error mentions the server connection, authentication, a certificate, or configuration parsing. A wrong server address and a system authorization failure require different fixes: check the server and protocol settings for the former, and system permissions for the latter. Don’t clear all your servers just because the switch turns off. If the switch stays on but websites won’t load, go to the next chapter rather than repeating checks for a permissions issue that has already been resolved.
- Disconnect and keep your current configuration and subscription records.
- Keep Shadowrocket in the foreground until you’ve responded to any system prompts, then check the switch again.
- Record the exact error. If the switch stays on, test one specific network request.
- Remove and reauthorize the VPN configuration only after confirming that the system configuration is faulty.
When to reauthorize and how to recover
Before removing a system VPN configuration, make sure you’ve kept your server details and a copy of Config. A system VPN configuration and an in-app rules file are different things. Check which screen you’re on before deleting anything; don’t mistake your subscription list for a system configuration. Reopen Shadowrocket and connect to add the configuration again when prompted. If it still fails, restart the device and test once more on the same network. A restart can clear a temporarily stalled system extension; it doesn’t prove the original configuration was correct.
On Mac, Windows, Android, and Linux, first confirm that the app launches normally, then check the relevant network permissions. If the problem occurs before the app opens for the first time, follow the installation guidance on the download page: on Windows, verify the source, then select “More info → Run anyway” in SmartScreen; on Mac, look for “Open Anyway” under “Privacy & Security”; on Android, check “Install unknown apps” permission; on Linux, check the package format and execute permissions. Don’t confuse an installation block with a connection failure, and don’t allow files from unknown sources to run.
02 / Routing
Connected, but websites and apps can’t access the internet
A connection switch that stays on means the system connection is established; it doesn’t prove that every request is taking the intended route. Record whether nothing is accessible, only some domains fail, or a specific app fails while the browser works. These can point to different causes, including the underlying network, server route, rule matching, DNS, or the app itself. Choose one reliable target with a clear purpose and repeat the same test. Don’t compare results from a different website each time and draw a general conclusion.
Check the underlying network while disconnected
Disconnect Shadowrocket and use your current Wi-Fi or cellular network to visit a page that normally works without a proxy. If that fails too, check Wi-Fi sign-in, cellular data permissions, your router, or the system’s network status first. Public Wi-Fi may require web-based sign-in; seeing a connection icon doesn’t mean you’ve completed authentication. Sign in while disconnected from the VPN, then start Shadowrocket. If the sign-in page itself won’t load, don’t keep changing rules to look for the cause.
Once the underlying network works, restore the original server and configuration, then confirm that the intended SERVER is selected in Home. If the list contains similar names, check the address or subscription rather than relying on the display name. A successful server test doesn’t automatically mean the next connection will use that server. If you’re using a selection policy, confirm which available entry it resolves to. Treating a policy name as a fixed server makes before-and-after tests hard to compare.
Compare the three Global Routing modes
Note the current Global Routing mode, then switch modes briefly for diagnosis. Config applies the active configuration; Proxy routes applicable traffic through the selected proxy path; Direct provides a direct-connection comparison. This is a diagnostic test, not a permanent recommendation. Restore the original mode when you’re done, especially if you temporarily selected Proxy just for testing.
| Mode | What it helps compare | What to check based on the result |
|---|---|---|
| Config | Observe how the current rules handle traffic | If only this mode fails, check rule order and policies |
| Proxy | Reduce differences caused by rule routing | If it fails too, check the server, authentication, and DNS |
| Direct | Check the target’s direct network path | If it fails too, check the underlying network and target service |
If Proxy works but Config doesn’t, check which rule matched the target and whether its policy exists. Rules are usually evaluated in order, so a broad entry near the top can override a more specific one later. The snippet below illustrates rule order; it isn’t a complete importable configuration. Make sure the policy referenced by PROXY is defined in your active configuration—pasting a rule alone doesn’t provide a working route.
[Rule]
DOMAIN-SUFFIX,example.com,PROXY
GEOIP,CN,DIRECT
FINAL,PROXY
Narrow it down to a single request
If only one app fails, visit the same service’s public page in a browser, then compare app sign-in, content loading, and uploads. If the page loads but sign-in fails, the cause could be authentication, the device’s time, or server-side fraud prevention. Images may load from a different domain, so adding a rule for the home page alone may not help. Check the request log for the actual failing domain and its handling policy; a successful request isn’t evidence of a failure.
If switching networks fixes the issue, save the results from both networks with the same configuration, then check local network restrictions and DNS differences. If it fails on every network, focus next on the server or configuration. After each change, send a new request; if needed, close the old connection first, since existing long-lived connections may not use updated rules immediately. For the full import and connection sequence, see the Config setup guide. For protocol fields, see the protocol FAQ.
03 / Server and transport
Server timeouts, intermittent connections, and slow speeds
A timeout means a step didn’t receive the expected response within the wait period. It doesn’t necessarily mean the password is wrong or that the server is down. Latency tests, protocol handshakes, and real website requests may use different methods, so record each result separately. First identify whether one server times out, every server in the same subscription times out, or timeouts only happen on a particular network. The more precisely you define the scope, the easier it is to decide whether to check your local network, subscription settings, or remote service.
Distinguish a failed test from a failed real-world request
Before running Connectivity Test, check the test target and method, then make a real request in a browser. Some servers may not respond to a particular probe but can still handle normal proxy traffic. Conversely, a successful probe doesn’t guarantee that the route to your destination works. Don’t treat a latency reading as a download-speed measurement or judge a server’s long-term reliability from one test. Several consistent results under the same conditions are more useful than a single reading.
Keep the configuration unchanged and test the same server on Wi-Fi and cellular. If it fails only on Wi-Fi, check the router, public network restrictions, and DNS. If it fails only on cellular, check the app’s cellular permission and the current mobile data connection. Tests use data, so check your allowance before transferring large files. If you already have another server you know works, use it as a comparison; don’t import entries from unknown sources just to troubleshoot.
Check protocol settings one at a time
Start with the server address, port, and authentication details; then check TLS, SNI, transport type, and path. A successful import only means the client can read the content—it doesn’t mean every field matches the server. Don’t use the server address for every name field or assume that the same port means the transport settings are identical. Compare each value with the original details from your own service provider, and keep the existing entry before making changes.
| Protocol or feature | Check these details | Common misdiagnosis |
|---|---|---|
| Shadowsocks | Address, port, encryption method, and password | Changing only the password and overlooking the encryption method |
| VMess / VLESS | Identity fields, transport, TLS, and related settings | Assuming all other fields are interchangeable when the protocol name matches |
| Trojan | Authentication details, TLS, and SNI | Confusing the connection address with the certificate name |
| Hysteria2 / WireGuard | UDP reachability and the relevant authentication and key settings | Assuming UDP works just because ordinary websites load |
If you see a certificate error, first check that the device’s date, time, and time zone are correct, then verify the server certificate name and SNI. Don’t treat disabling certificate verification as a general fix; it can hide a name mismatch, an invalid certificate, or a connection to the wrong host. If your service provider has changed the connection details, update your own subscription and compare the new and old entries instead of guessing at field values.
Break speed problems down by stage
A slow connection setup, slow first page load, and slow sustained download point to different bottlenecks. Setup delays may occur during DNS resolution or the handshake; a slow first page may involve the destination and several supporting domains; sustained download speed also depends on local bandwidth, server load, and the remote route. Compare the same target and file at similar times, changing only the server or network in each test. Also account for background system updates and cloud sync.
If small pages load normally but large downloads keep stalling, note whether this happens only on one network or transport, then ask your own service provider to check the server. Don’t change MTU or concurrency settings without evidence; the wrong values can make the problem worse. To confirm that a setting is responsible, check whether restoring its original value makes the issue reproducible again. When reporting a problem, include the time, network type, exact error, and redacted settings—not full passwords or private keys.
04 / Subscription data
Subscription import or update fails, or the list doesn’t change
A subscription update has two stages: retrieving the content and parsing it. Retrieval depends on whether the URL is reachable, authentication, and the server response. Parsing depends on the format, protocol fields, and whether the client can recognize the content. First identify whether the error happened during download or import. Handle timeouts, authentication errors, empty responses, and format errors separately. Your subscription URL may contain access credentials; don’t include the full URL in public screenshots, chat messages, or synced browser notes.
Make sure you received subscription data, not a web page
Copy the subscription URL again from your own service provider’s account page. Make sure you haven’t copied a help page, management page, or sharing page instead. Check that there are no spaces, line breaks, or punctuation accidentally added at either end, especially if a messaging app wrapped the text. URL parameters may provide authentication, so don’t delete them to make the address look shorter. Successfully scanning a QR code only means the content was read; it doesn’t prove that it’s in the subscription format the client expects.
If the response says authentication failed or the link is invalid, check your service provider account to see whether the subscription was reset, expired, or restricted by access conditions. A saved URL may stop working as soon as it’s reset; repeated update attempts can’t restore expired credentials. If the response is a sign-in or verification page, you’ve downloaded a web page, not a server list. Get the correct subscription URL instead. Don’t turn the page contents into a few lines of addresses and treat them as a complete subscription.
Check the retrieval path and preserve the existing list
Some subscription URLs are unreachable on your current direct connection but work over an existing connection that you know is usable; the reverse can also happen. Record the original connection state, then test updates both while disconnected and while using a confirmed working connection. Don’t use a server that depends on the subscription update you’re trying to recover as the retrieval route—that creates a troubleshooting loop. If every server is unavailable, ask your own service provider to confirm the subscription URL and server status.
If an update fails, keep the existing list instead of deleting the whole subscription. If an old entry still connects, it can help determine whether the problem is limited to the subscription endpoint or also affects the servers. After an update, check the server count and entry fields, but don’t assume that an unchanged count means nothing changed. Names may stay the same while addresses, authentication, or transport settings change. Likewise, a longer list doesn’t mean every new entry will connect.
Identify parsing errors and update behavior
If the content downloads but parsing fails, confirm that the format provided by your service provider matches the import method you selected. Add Server, Subscribe, Scan QR Code, and Import from Cloud JSON don’t accept exactly the same content. A full configuration file, a single connection entry, and a subscription URL are not interchangeable. Use the matching import option to add a copy and check the result before replacing the old entry. See the subscription import guide for details.
If you manually edit an individual subscription entry, a later update may overwrite your changes with the service provider’s values. If you need to keep a temporary fix, save it as a separate, clearly named entry and check the original details with your provider. Don’t rely on reapplying the same manual edit after every update. If entries disappear, check filters and group display settings, then confirm that the subscription response actually included them. Don’t mistake a display filter for data loss.
Leave a pause between retries. Repeated taps can’t fix authentication or format errors and may trigger access limits. When documenting the issue, note which stage failed, the exact error, the network, and whether the connection was active; share only a redacted version of the URL structure. If you’re asked for logs, check for subscription credentials, server passwords, and private domains first. Subscription availability depends on the relevant service; the app’s purchase status doesn’t determine the status of your subscription account.
05 / DNS resolution
DNS failures, unreachable domains, and unexpected rule matches
DNS translates domain names into addresses used to connect. If the server connects but some domains don’t load, DNS resolution may have failed—or resolution may have succeeded while the connection to the destination failed. A browser error alone can’t distinguish the two. Record the failing domain, time, and Global Routing mode, then check the request or diagnostic log for resolution errors. A blank page alone isn’t a reason to add more DNS servers; changing too many settings at once makes the results hard to interpret.
Separate DNS resolution failures from connection refusals
If the log shows that no address was obtained, check whether the DNS server is reachable, whether its transport depends on the current network, and whether the relevant settings are active. If an address was returned but the connection then times out, check routing, the exit path, and the destination service. Don’t use a destination IP in place of its HTTPS URL as your only test: HTTPS also relies on the domain, certificate, and server name. Failure to access an IP directly doesn’t prove that DNS is working or failing.
If the same domain behaves differently on Wi-Fi and cellular, keep the server and configuration the same and compare both networks. If only internal LAN names fail, check whether they need to be resolved by the current router or your organization’s internal DNS. Public DNS services usually don’t know the names of devices on a private LAN. Sending every request to the same remote resolver may fix external access while breaking internal services. Handle internal domains according to the network you’re on rather than applying a one-size-fits-all address.
Check how DNS settings interact with rule order
In Config, confirm which file is active before reviewing its DNS settings; otherwise you may edit an unused copy. Record system DNS, DNS settings in the configuration, and any encrypted DNS settings, then determine which one handles the current request. Don’t assume every address shown in a settings field is active at the same time, or confuse system-level resolution with resolution on the proxy server. Different paths can return different addresses, so check the actual request log.
Check rules starting with the earliest possible match. DOMAIN matches an exact domain; DOMAIN-SUFFIX matches a domain suffix; IP-CIDR and GEOIP match address conditions; FINAL handles requests that haven’t matched earlier rules. Address-based conditions may depend on DNS results, so a change in DNS can also change routing. The example below shows rule types only; it doesn’t include complete policy definitions and can’t replace your active configuration as-is.
[Rule]
DOMAIN,example.com,PROXY
DOMAIN-SUFFIX,example.org,DIRECT
IP-CIDR,192.168.0.0/16,DIRECT
GEOIP,CN,DIRECT
FINAL,PROXY
If a domain that should use PROXY is matched earlier by a broad DIRECT rule, adjust the rule’s position or condition instead of repeatedly appending the same domain at the end of the file. Also check whether the app’s actual request uses the domain you expected: sign-in, images, APIs, and push notifications may use separate domains. The address bar won’t show every background request. Find the failing request in the log to identify what needs checking.
Cache, IPv6, and retesting
After changing DNS, old connections and cached results may still affect what you see. Save the configuration, disconnect and reconnect, then make a new request to the target. If needed, close and reopen the target app. Don’t decide the setting had no effect by refreshing a diagnostics page without waiting for a new request. Restarting the device can clear some cache and session issues, but record the original results first so you don’t lose the conditions needed to reproduce the problem.
If IPv6-related connection failures occur only on some networks, compare the IPv4 and IPv6 addresses in the logs and check how the current configuration handles each address type. Don’t permanently disable IPv6 at the first timeout, or assume that enabling it will always make things faster. First confirm that failures consistently correlate with one address type, then run a brief comparison and restore the original setting when done. For DNS policies managed by a router or organization, check with the network administrator rather than overriding managed settings.
06 / Background activity and connection policy
Background battery drain, heat, and repeated On Demand connections
To investigate battery drain, check the system battery usage report first, then decide whether Shadowrocket is a significant factor. Percentages on the Battery page show each app’s share of usage during the selected period; they don’t mean the battery’s total capacity dropped by the same percentage. When the device has been used for only a short time, background activity can account for a deceptively large share. Choose a representative period and record screen time, network type, signal strength, and background downloads. Don’t draw conclusions from a brief reading taken just after opening the app.
Compare results under consistent conditions
In system Settings, check Shadowrocket’s foreground and background activity on the Battery page, then compare it with the browser, video playback, cloud sync, and system tasks during the same period. Traffic through the system VPN channel isn’t necessarily generated by the client itself; other apps transferring data in the background can keep the network extension active. Traffic records on the Data page can help identify what’s active, but traffic volume doesn’t translate to battery use at a fixed rate. Also consider connection frequency, wireless signal, and device temperature.
Pause any clearly identifiable large background transfers, then see whether the issue persists. If the device heats up only on a weak cellular signal, mobile network retries and radio use may be contributing factors. If it continues on stable Wi-Fi, check the client’s connection policy. Keep screen brightness, usage, and network as consistent as possible. Don’t compare standby battery use with a test that involves charging and downloading large files.
On Demand isn’t a dedicated battery-saving switch
On Demand starts connections when configured conditions are met. Turning it on doesn’t guarantee better battery life, and turning it off doesn’t guarantee a more stable connection. If its conditions don’t match your usual network changes, the device may repeatedly try to connect over Wi-Fi, cellular, or when waking from sleep. Note when connections are triggered: after locking the screen, leaving Wi-Fi, joining a network, or opening an app. Compare those events with the configured conditions to identify repeated triggers.
Before disconnecting to observe what happens, check whether On Demand will immediately reconnect. Tapping the Home switch without addressing automatic connection conditions can make it look as though the switch isn’t working. Temporarily disable On Demand, then observe under the same network and similar usage conditions before restoring it and comparing results. This isolates the trigger; it doesn’t mean everyone should leave it off. If you need automatic connections, adjust the rules to match the networks you actually use.
- Record the existing On Demand conditions and connection status.
- Pause automatic triggers and establish one stable connection manually.
- Check for continued reconnects, network interruptions, or unusual background activity.
- Restore conditions one at a time to find which one coincides with the issue.
Trace repeated reconnects to the server and network
If the manual connection still drops frequently after automatic triggers are disabled, go back to the server, network, and DNS chapters rather than continuing to adjust battery settings. An unstable server connection, failed authentication, or DNS timeout can all make the client retry. Background battery drain may be a result of connection failures, not a separate performance issue. Restore a stable connection first, then assess standby battery use; this is usually more informative than simply restricting background activity.
If reconnects happen only on one Wi-Fi network, check whether it requires periodic sign-in, disconnects when idle, or causes the device to switch access points frequently. Low Power Mode, app background behavior, and network policies may also affect the results, so record them too. Don’t disable system protections that aren’t relevant just to lower a reading. If your use case requires a persistent connection, note whether related background tasks remained active during testing.
When you finish, turn off detailed logging and stop diagnostic tests that generate a lot of activity so the test itself doesn’t keep the device busy. If the issue is reproducible, describe a simple scenario with a fixed network, one server, a clear On Demand state, and one repeatable action. See the Data page traffic statistics guide for traffic details, or troubleshoot On Demand and background battery drain for connection policy guidance.
07 / Updates and recovery
Configuration issues after an app or system update
An issue happening “after an update” doesn’t prove that the update caused it. An app update, system update, subscription refresh, rule change, or server-side setting may have happened around the same time. List the changes that actually occurred, then retest with a saved copy of the configuration. Don’t reinstall, reimport the subscription, and change DNS all at once. Even if the issue goes away, you won’t know which change helped—and you’ll have no reliable reference if it happens again.
Save the current state, then verify what’s actually in use
Keep the active Config, server details, and exact error text. Confirm that the same server and configuration file are still selected after the update. Changes to list order, groups, or names can lead you to select a different entry, making it look like an old configuration stopped working. If you use an automatic selection policy, check which entry it actually selected. If you made local changes, see whether a subscription update replaced them. Verifying the active items is usually more useful than immediately resetting everything to defaults.
Separate an app that won’t launch, a connection switch that won’t stay on, a failed server handshake, and rules that only partially work. If the app opens but asks for VPN permission again, follow the system prompt and return to Chapter 1. If the switch stays on but access fails, follow the routing and server checks. If a system update temporarily disrupted the network extension, restart the device and retest under the same conditions—but record the current prompts before restarting so you don’t lose the only error message.
Use a minimal configuration to check compatibility and references
Make a copy and remove nonessential rules and policies, keeping one server whose settings you’ve verified. Test the basic connection first. A minimal configuration can show whether the issue comes from complex policy references or another configuration item; it doesn’t mean you should permanently delete all routing rules. If the basic connection works, restore rules in groups, saving and testing after each one. If the issue returns after restoring a group, check its conditions and policy names.
Fix configuration parsing errors by checking the specific line and field. A renamed policy, a reference to a missing group, a misplaced section, or an invisible character added during pasting can all prevent a file from loading as expected. Confirm that you’re editing the active file and that it saved successfully. Don’t overwrite a new environment with an entire old file and immediately delete the backup. Keep both versions so you can distinguish an issue in the original content from a change made during migration.
Where to get the app again and how to allow it to install
Update or redownload the iPhone and iPad app through the App Store. Check the App Store listing for system requirements. Verify the developer, Shadow Launch Technology Limited, the official icon, and app ID 932747118. See the all-platform download page for Mac, Windows, Android, and Linux updates. Installers are available only from this site under /dl/shadowrocket/. Don’t overwrite your current app with a same-named program from an unfamiliar source.
On desktop and Android, system warnings may appear again the first time you launch or replace an installation. On Windows, verify the source, then select “More info → Run anyway.” On Mac, select “Privacy & Security → Open Anyway”; if you see a specific warning that the app is damaged, follow the download page’s instructions to check the path before handling extended attributes. On Android, allow the app you’re using to open the installer to install apps. On Linux, choose a package format supported by your distribution; AppImage files need execute permission. Ask an administrator about restrictions on managed devices.
Before redownloading, make sure your backup can actually be used. A screenshot showing only server names may not preserve authentication fields, and a subscription name alone can’t replace the full URL. Keep backups in a location you control, not a public space. If the same configuration behaves consistently differently before and after an update, report the steps to reproduce, exact error, system environment, and redacted configuration. Avoid including your full browsing history or unrelated personal data.
08 / iPad
iPad troubleshooting for landscape, device migration, and network changes
The troubleshooting sequence on iPad is the same as elsewhere: check system permissions, the underlying network, the server, rules, and DNS in that order. Also account for landscape and multitasking layouts, Wi-Fi-only versus cellular models, keyboard input, and whether configuration data migrated correctly from another device. A subscription that works on iPhone doesn’t mean you can skip checks of the iPad’s current network and selected server; the two devices may have different connection conditions.
Confirm what you’re selecting in landscape and multitasking views
In landscape, the list and details may appear side by side. Opening an entry to edit it doesn’t necessarily select it as the server for the current connection. After editing, return to Home and confirm the selected entry and actual connection target before testing. If a multitasking or split-screen layout hides a button, expand Shadowrocket to a wider window or full screen before importing and saving. This helps distinguish limited screen space from a missing feature.
When pasting a subscription with an external keyboard, make sure the input field has focus, the entire URL was pasted, and Return wasn’t pressed accidentally. A long URL that doesn’t fit in a narrow field isn’t necessarily truncated; check both ends and its parameters rather than judging by what’s visible. When copying a configuration snippet, preserve English punctuation and actual line breaks. Don’t let a text editor automatically replace commas, quotation marks, or hyphens. Reopen the relevant entry after saving to verify the change was applied.
For device comparisons, control the network and configuration
To compare an iPhone and an iPad, connect both to the same Wi-Fi, use the same server and Global Routing mode, and test the same target. If only the iPad fails, compare its system VPN permissions, active Config, DNS, and On Demand settings. Don’t use cellular on one device and public Wi-Fi on the other, then attribute the difference to the device model. A comparison is useful only when the conditions match.
A Wi-Fi-only iPad can use your own available hotspot as a second network for comparison. Note the hotspot’s upstream connection and data limits so the test path is clear. On a cellular iPad, check app cellular permissions, data connection status, and the results after switching from Wi-Fi. If the connection drops after leaving Wi-Fi, first confirm that the underlying network switch has completed, then see whether Shadowrocket reconnects. Don’t mistake a brief failure while the network is still coming up for an ongoing issue.
A brief loading delay when reopening the app after locking the screen may be related to an old session reconnecting, a network change, or the target app resuming. Note whether it recovers and sends another request on its own, and whether the Home connection status actually changes. If only one app needs refreshing, check its session first. If the system connection keeps dropping, return to the connection switch and On Demand chapters. For public Wi-Fi, check whether you need to sign in again after the device wakes.
Verify previous purchases and configuration migration separately
The iPhone and iPad app costs $2.99 as a one-time App Store purchase; the Mac, Windows, Android, and Linux versions are free. On iPad, sign in with the same Apple ID used for the purchase and look for Shadowrocket in your purchased items. Check the App Store listing for system requirements. If you still see a price, check the signed-in account and purchase history before paying again. Getting the app and migrating your connection settings are separate steps; completing the first doesn’t confirm that the second is done.
After migrating from another device, check the subscription URL, server settings, Config, DNS, and automatic connection conditions separately. The same server name doesn’t prove that all the details match, and not all local settings are guaranteed to migrate. Import your own subscription and verify one connection first, restore routing rules next, then enable On Demand. This order makes it easier to identify which step introduced a problem.
If you need to sync files, use storage you control and verify that you’re not using an old copy. Don’t send configuration files containing credentials through public links; remove credentials before asking for help. If the issue occurs only with a specific iPad layout, network switch, or locked-screen scenario, record the steps and conditions needed to reproduce it. See iPad purchase, download, and landscape-use tips for related guidance, or return to the Getting Started guide to check the basics.