Official manual · Troubleshooting
Shadowrocket Troubleshooting Guide
Identify whether the issue involves the system connection, server, subscription, rules, or DNS. Preserve the current setup, then narrow it down by changing one thing at a time.
For initial setup, start with the user guide to import your configuration, choose Global Routing, and make your first connection. This page is for users who already have their own subscription or server details but encounter a problem at a particular step. Change only one setting at a time and test the result before continuing. Don’t start by reimporting everything or reinstalling the app.
01 / System connection
The connection won’t start, turns off immediately, or keeps reconnecting
First determine whether the connection fails to start or connects but doesn’t work
After tapping the connection toggle on Home, check whether it stays on, whether the system shows a VPN status, and whether the app immediately displays an error. If the toggle turns off, check system authorization, the connection configuration, and whether the extension starts. If it stays on but webpages fail, continue to the next section. A VPN status only means the system has established a network tunnel. It doesn’t confirm that the remote server is reachable or that rules, DNS, and the destination service are working. Treating these as the same issue can lead to repeatedly changing server settings without addressing a system-level problem.
The first time you connect, your iPhone or iPad may ask for permission to add a VPN configuration. The device owner must approve the request and complete any required system authentication. If you dismissed it, try connecting again in the app and watch for the system prompt in case another action interrupts it. On managed devices, organization policies may restrict adding or changing VPN configurations. Check for a management notice in system settings and ask your device administrator to confirm your permissions. Don’t delete a management profile as a test or overwrite a connection that must remain in place for your organization.
Check configuration and connection conflicts with as few variables as possible
Disconnect, then note the selected server, Config file, and Global Routing setting. Exit any other VPN sessions that are controlling system traffic. Check system VPN settings for an older connection that is still active; disconnect only entries you recognize, and don’t delete them in bulk. Return to Shadowrocket, select an existing server entry with complete settings, and connect. If one configuration fails to start but another saved configuration works, check the failing file’s format, references, and recent changes first.
If the toggle briefly turns on and then switches off, check any error details or logs available in the app. Note when it happened and the last few relevant messages. A configuration parsing error, an authentication failure, and a reconnect triggered by a network change require different fixes. For a parsing error, undo the most recent addition and check separators, blank lines, and truncated text. For an authentication error, verify the server settings. An older error in the log may not relate to the current problem. Stop the connection, narrow the log to the relevant period, and reproduce the issue before drawing conclusions.
Check for automatic actions triggered by On Demand
If the connection turns back on after you manually switch it off, check whether On Demand is enabled and whether its triggers match the current network. To establish a baseline, temporarily disable On Demand, then manually connect and disconnect on the same network. If manual control works, investigate the trigger conditions instead of changing the server password. Restore the conditions you need after troubleshooting, then test what happens when joining Wi-Fi, leaving Wi-Fi, and waking the device. Don’t mistake an On Demand reconnect for a faulty toggle, or force-quit the app instead of managing the system connection.
Only restart the device after checking authorization, configuration, and trigger conditions. Save any error details first. After restarting, keep the same network, server, and configuration so you can compare results; changing networks at the same time makes diagnosis harder. If the system VPN configuration is genuinely corrupted, back up your app configuration, remove only the entry you can confirm belongs to Shadowrocket, and let the app request a new one. You’ll need to authorize it again. Recreating the configuration won’t remove device management restrictions, and you shouldn’t rely on the system to restore unsaved connection details.
02 / Access path
Connected, but no internet access or some apps don’t work
Establish a baseline with the connection off
Disconnect Shadowrocket and confirm that your device can open sites you normally use on the current network. If basic access also fails, check whether Wi-Fi requires a captive portal, cellular data is available, and system settings allow apps to use the network. Complete any public Wi-Fi sign-in with the connection off, then test again. Don’t start by changing DNS or replacing the entire ruleset. If the basic network isn’t ready, every higher-level failure may have the same cause. The more settings you change, the harder it will be to know what’s still needed once the network works.
Once the basic network is working, choose one existing server and two test pages you can normally access for different purposes. Record the result for each page with the connection on and off. Don’t test only a service’s homepage and image domain, since they may share a backend. A VPN indicator isn’t proof that the connection works. If all apps fail, focus on server reachability and DNS. If only one app fails, check its associated domains, UDP use, system permissions, and rule matches. Pause large downloads during testing to keep the results clear.
Use Global Routing for a controlled comparison
With Config, Global Routing applies the rules in your configuration to select a policy. Proxy routes relevant traffic through the selected proxy path, while Direct provides a direct-connection comparison. You can briefly switch modes for testing, but note the original setting and restore it afterward. Proxy isn’t an automatic fix: incorrect remote settings will still fail. Direct access working doesn’t prove the proxy path is broken; the destination service may treat traffic differently depending on its source network. Compare the same destination over different paths rather than treating one mode as a permanent solution for every case.
| Observed result | Check first | Next step |
|---|---|---|
| Direct works, Proxy fails | Server, transport, and remote DNS | Keep the server unchanged; check its settings and logs |
| Proxy works, Config fails | Rule order and policy references | Find the rule that actually matches the destination |
| Both fail | Basic network, destination service, and DNS | Try a separate destination and check the resolution stage |
Review the rules and find the first match
If Proxy works but Config doesn’t, check whether an earlier DIRECT or REJECT rule is catching the destination. Rules are evaluated from top to bottom, and processing stops at the first match. FINAL only handles requests that haven’t matched an earlier rule. Don’t just check whether your new rule exists; look for a broader DOMAIN-SUFFIX, DOMAIN-KEYWORD, or IP rule above it. If you use a custom policy name, make sure the reference is valid. A spelling mismatch doesn’t mean the server is unavailable.
[Rule]
DOMAIN-SUFFIX,ad.example.com,REJECT
DOMAIN-SUFFIX,example.com,PROXY
IP-CIDR,192.168.0.0/16,DIRECT
GEOIP,CN,DIRECT
FINAL,PROXY
This example illustrates rule order only; it isn’t a complete configuration for your network. Put a more specific ad subdomain rule before a broader domain suffix rule, route local network ranges directly, and place FINAL at the end. A real app may also use separate domains for authentication, media, and APIs, so a working homepage doesn’t mean every dependency is restored. Check failed requests individually and make the narrowest necessary rule change. Don’t permanently switch the entire configuration to another mode to fix one page. See custom rules and FINAL for details.
03 / Servers and transport
Server timeouts, authentication failures, and TLS errors
First identify which layer the test reached
Interpret Connectivity Test results in the context of the selected test method. A successful TCP connection only shows that the address and port were reachable at that time; it doesn’t confirm authentication, TLS, higher-level transport, or the destination request. An application-level test can fail even when the port is reachable. Note the test method, current network, and specific server, then compare with an actual webpage request. Don’t rank results from different test methods as if they were equivalent, or delete untested entries from an entire subscription because one test timed out.
Choose one server and check each field against the information from your existing service provider, in this order: protocol, address, port, authentication fields, transport, and TLS. When copying, watch for full-width punctuation, leading or trailing spaces, and text broken across lines. The address field usually shouldn’t include a subscription URL’s path or query parameters, and you can’t infer the port from the server name. If several entries in the same subscription share an endpoint and all fail, check the shared endpoint and network first instead of changing each password. If only one entry fails, keep the working ones as a comparison.
| Field group | What to check | Common assumption to avoid |
|---|---|---|
| Shadowsocks | Encryption method, password, and plugin settings | Assuming the address and port are all that matter |
| VMess / VLESS | User ID, transport, and server requirements | Using settings from one protocol with another protocol’s entry |
| Trojan / TLS | Authentication details, Server Name, and certificate validity | Assuming the connection address and certificate name are the same field |
| Hysteria2 | Authentication, TLS, and UDP network conditions | Assuming TCP reachability guarantees QUIC will work |
Distinguish the certificate name, connection address, and system time
With TLS, the server connection address and Server Name may differ. Use the values from your existing server configuration; don’t assume they should be the same domain. With transports such as WebSocket, Host and Path may also serve separate purposes. Match the path’s capitalization, leading slash, and value required by the server. For certificate errors, first check whether system time is set automatically, whether the certificate has expired, whether its name matches, and whether the network requires sign-in. Don’t disable certificate verification to hide a name mismatch: that weakens connection identity checks without fixing the cause.
If a server times out on one Wi-Fi network but works on cellular, leave its settings unchanged and compare how the two networks handle the port, UDP, and long-lived connections. Hysteria2 uses QUIC and depends on UDP; network restrictions on UDP may cause it to fail even if ordinary webpages work on the same network. Check the transport requirements for that endpoint with your existing service provider, or compare it with another transport already configured in your own entries. Changing the protocol name in a setting won’t make an incompatible configuration work.
Separate remote service problems from local errors
If the same entry fails on different networks while other existing entries work on the same device, check whether that server is under maintenance, its credentials have expired, or its endpoint has changed. The client can’t fix a server that isn’t listening, a misconfigured certificate, or changed account permissions. When contacting your existing provider, include the protocol, time of failure, network type, and a redacted error. Don’t send a complete subscription URL, password, or user ID. “It won’t connect” doesn’t distinguish authentication from network problems; identifying the failure stage is usually more useful than repeatedly importing the same link.
After changing one field, reconnect and test the same destination again. Don’t stop just because the test indicator changes color. If the error changes from a timeout to an authentication failure, the request has reached a different stage; continue based on the new evidence rather than assuming the change had no effect. For a more detailed sequence, see troubleshooting server timeouts, protocol settings, and system time.
04 / Import and updates
Can’t import a subscription, refresh fails, or content is incomplete
Separate failure to retrieve content from failure to parse it
Subscription import has two stages: retrieving a response and recognizing its contents. Request timeouts, domain resolution failures, and certificate errors occur during retrieval. Check the response format only if content was received but the app doesn’t recognize it, shows no entries, or reports missing fields. First note where the issue occurs: Subscribe in Config, clipboard import, or Scan QR Code. These entry points don’t all accept the same content. A server share, server-list subscription, and complete configuration file aren’t interchangeable just because they look like links.
Copy the original subscription address again from your own service provider. Check that the scheme, hostname, path, and query parameters are complete, and look for line breaks added by a messaging app, a trailing period, or characters lost during copying. A subscription token is a credential: don’t paste the complete link into a public post, screenshot, or search field. The example below is for format illustration only; it can’t be imported and doesn’t point to a working service. If a QR code contains an account page rather than subscription content, scanning it only means the text was recognized. Check with the source to confirm what it’s for.
https://example.com/sub?token=xxxx
Check the network path and returned content
If a manual refresh times out, verify the basic network first, then check whether the subscription domain resolves on the current network. Determine the path used by subscription updates from the app’s relevant settings and actual logs; don’t infer it from the name Global Routing. If you’re stuck because the connection needs an update but updating requires a connection, keep an existing working entry and use it for a comparison. Don’t clear the list first. If your provider gives specific update instructions, use an existing valid entry as directed. Don’t rewrite the domain or remove query parameters, which may break authentication.
If a request gets a response that the app can’t recognize, it may have received a login page, expiration notice, verification page, or error text instead of the expected configuration. A link opening in a browser doesn’t mean its contents are suitable for import. If you need to inspect the response, do so only in an environment you trust, and don’t upload content containing credentials to online conversion or inspection tools. If an address is explicitly identified as a complete configuration file, handle it through Config’s configuration management workflow. Don’t try to silence an error by manually turning unfamiliar content into a server list.
If entries disappear or custom content is lost after a refresh
Subscription refreshes generally use the remote content as the source of truth. Entries removed by the provider may disappear, and refreshes may overwrite fields you edited manually. Before refreshing, note the selected entry, groups, and custom fields. Keep manually managed entries separate from remotely managed content if you need them long term. If the selected entry disappears after a refresh, choose another valid entry and test the connection again. Don’t assume a failed connection caused by a missing selection means the entire subscription is broken. Entries with the same name can still have different addresses, ports, or transport settings; compare the key fields if needed.
If automatic updates fail but manual updates work, check whether the network was the same during both attempts, especially if one happened just after app launch, device wake, or before Wi-Fi sign-in was complete. Automatic updates reduce repetitive work; they shouldn’t need to run at very short intervals to keep a connection working. If the remote service limits request frequency, stop retrying and wait as instructed. Repeatedly importing the same link adds duplicate groups and troubleshooting noise. It doesn’t extend credential validity or fix an error in the remote response.
05 / Performance and network path
Connected, but speeds are slow, webpages stall, or video buffers
Look separately at startup delay, transfer speed, and stability
“Slow” can describe at least three different issues: a long delay before loading starts, slow transfer after it begins, or repeated stalls during transfer. The first can involve DNS, connection setup, or TLS. The second calls for checking available bandwidth and server load. The third may involve packet loss, wireless conditions, or network switching. Describe the symptom precisely and choose a page or file you can test repeatedly. A latency test isn’t a throughput test, and a quick response to a small request doesn’t prove that large transfers will be stable.
Before comparing results, pause system updates, cloud sync, and large downloads on other devices. Keep the device location and network unchanged. Test the basic network with the connection off, then use the same server and Global Routing setting to access the same destination. If the service changes video quality or download sources dynamically, note that too. Repeated speed tests in quick succession use data and may affect later results through congestion control, caching, and remote rate limits. Leave a reasonable gap and look at the trend instead of selecting the best or worst result.
Find the bottleneck by changing one variable at a time
If things are slow even with the connection off, check local Wi-Fi signal, router load, cellular signal, and the destination service. If moving closer to the access point helps, investigate the wireless link first. If other devices are also slow, don’t focus only on this device’s app settings. If the basic network works but one existing server is slow, compare it with another entry you already have on the same network, keeping the destination unchanged. A comparison can show a difference between paths or servers; it can’t establish the overall performance of a protocol from one test.
If the same server behaves differently across networks, note the network type, time of day, and any UDP restrictions. Transports such as Hysteria2 have their own network requirements; correct settings don’t mean they’ll work well on every network. TCP-based transports are also affected by packet loss, round-trip time, and congestion. Don’t set connection counts, buffers, or bandwidth values to extreme levels based on guesswork. Poor estimates can increase queuing and instability. Follow the server provider’s guidance, and record the original values so you can restore them if there’s no improvement.
Change settings only when the evidence points to a specific issue
If a webpage pauses before loading but then runs at normal speed, see the DNS section and check for retries after resolution failures or delays between resolution paths. If only some resources are slow, check which policies their requests actually match. The main page may use Proxy while image or API requests go Direct, or the other way around. If you recently changed rules, undo the latest change and test again before changing DNS or servers as well. A rule that sends traffic over the wrong path is more important to check than the number of rules; use the actual match records to find out.
If a large request stalls repeatedly despite a successful handshake, MTU or path fragmentation can be considered as a later possibility, but a “stuck webpage” alone doesn’t identify the cause. First check whether the issue reproduces across networks and destinations, then make a small adjustment using options the app actually provides and the server’s requirements. Don’t repeatedly try fixed values copied from online posts: overhead differs across tunnels, transports, and access networks. If an adjustment happens to help one destination but makes others worse, restore the original value, keep the test notes, and continue troubleshooting rather than treating the temporary result as a permanent setting.
If evidence points to remote server load or a rate limit on your existing service, repeatedly reconnecting locally is unlikely to help. Give your provider the time of the issue, endpoint name, network type, and redacted comparison results. There’s no need to send your full configuration. For protocol background, see Hysteria2 and QUIC explained. Use results from your current network to guide changes, even after reading about the underlying technology.
06 / Name resolution
DNS failures, domains that won’t open, and unexpected rule matches
Identify the resolution stage instead of relying on the browser error
DNS translates domain names into addresses, but connection, TLS, and application requests still have to happen afterward. A domain that won’t open doesn’t necessarily indicate a DNS problem. A nonexistent name, a query timeout, an unreachable returned address, and a certificate name mismatch can all result in a failed page. Check requests or logs for the domain to see whether a result was returned before changing DNS settings. Don’t blame every access problem on DNS based on a generic browser message, and don’t confuse a subscription domain with the domain you’re trying to visit.
Choose one known-good domain and one failing domain. Test them on the same network and server, and record the time and any message about the resolution stage. Entering an IP address in a browser isn’t a full substitute for visiting the domain: HTTPS certificates, Server Name, Host, and content delivery may all depend on the domain. A response from an IP only shows that part of the network path is reachable; it doesn’t prove the original domain’s DNS settings are wrong. For services that require sign-in, distinguish a failure to resolve the login endpoint from invalid login credentials.
Understand the resolution paths used by the device, configuration, and remote service
Check the DNS settings in the current Config first, and note whether Wi-Fi on the system has a manually specified resolver. If the device also uses organization-provided DNS or another network extension, the actual path may depend on more than one source. Don’t delete every system setting at once. Confirm what each setting belongs to and use a test network you control for comparison. After changing DNS in the app, make sure the setting belongs to the active configuration, not a file that isn’t selected.
If an encrypted DNS service address is a domain name, reaching the resolver may itself require an initial lookup. This dependency is easy to overlook during troubleshooting. If the bootstrap lookup or access path is unavailable, switching to a longer-looking address won’t automatically solve the problem. Follow the configuration provider’s guidance for the bootstrap path, endpoint, and routing requirements. Use logs to determine whether the failure happens before or after connecting to the resolver. Addresses shown in examples or online articles explain the fields; they aren’t a substitute for a working configuration suited to your network.
Check domain and IP rules together
Domain rules can match a domain included in or identified from a request. Some IP rules need a destination address first. Before adding or changing IP-CIDR, understand whether it triggers additional resolution and what the related options in your configuration do. Local network services may use internal domains that public resolvers can’t resolve correctly. If a printer or internal site stops working while connected, check the local DNS and access path it originally relied on. Adding a public-domain rule for the destination won’t necessarily fix it.
[Rule]
DOMAIN,printer.example.com,DIRECT
DOMAIN-SUFFIX,example.com,PROXY
IP-CIDR,192.168.0.0/16,DIRECT
IP-CIDR6,fc00::/7,DIRECT
FINAL,PROXY
printer.example.com in this example only illustrates that a more specific DOMAIN rule should come before a broader suffix rule. It isn’t a usable address for a local device. IP-CIDR6 is for IPv6 address ranges; don’t substitute an IPv4 range. If the network provides both IPv4 and IPv6, one address family may have a problem, causing some requests to wait before falling back. Check the address family and order of failures in the logs, then test in a controlled environment. Don’t permanently disable IPv6 as a universal fix for DNS issues.
After making a change, reconnect and send the request again. App and system caches may briefly retain older results. Test the failing domain, a known-good comparison domain, and local network services. Keep the change only if all three behave as expected. If resolution works but access still fails, return to the server or rules sections rather than piling on more DNS settings without evidence. Look up common terms in the glossary.
07 / Background activity
Battery drain, device heat, and frequent wakeups
Check actual battery change before interpreting usage percentages
An app’s share of system battery usage is its relative share for the selected period, not the same percentage of the entire battery consumed by that app alone. A high share can appear even when the device used very little power overall. Note the observation period, starting and ending battery levels, screen use, network type, and whether the device was charging. Distinguish foreground use from idle time. When Shadowrocket is part of the system’s network path, ongoing traffic from other apps can keep related network processing active. Its name appearing in the list alone doesn’t identify the source of unusual battery use.
Start with a comparable idle test. Keep the location, signal strength, and background tasks similar, then record actual battery change with the connection on and off. Don’t compare a night on weak cellular reception with a short idle period on stable Wi-Fi. Cloud backups, photo sync, system updates, and media uploads can all change network activity. If those tasks run in only one test, the results aren’t enough to determine whether the client is responsible. The test doesn’t need to run for a very long time; consistent conditions and a repeatable comparison matter most.
Check for repeated reconnects and frequent background tasks
Look in app logs or connection records for rapid disconnects, new handshakes, and repeated DNS lookups. Weak signal, switching between Wi-Fi and cellular, a remote server closing connections, or On Demand conditions repeatedly triggering can all increase wakeups. First test a manual connection on a stable network. If reconnects decrease, investigate the network or trigger conditions instead of weakening security checks across the board. Also check whether cellular data is allowed when the device leaves Wi-Fi; otherwise it may keep retrying over an unavailable path.
Review settings for automatic subscription updates, automatic tests, and other scheduled tasks. Avoid repeatedly contacting the same remote service at very short intervals. Testing every entry automatically costs more than testing only the current one, especially when there are many entries and some endpoints are unreachable. Keep the checks you actually need for everyday use; save broad tests for active troubleshooting. If your provider requires periodic updates, follow its reasonable instructions. Frequent refreshes won’t make an existing connection faster and shouldn’t be used to keep it online.
Check logs, protocols, and background traffic separately
Detailed logs are useful for briefly reproducing a problem, not for keeping the highest logging level enabled after you’ve diagnosed it. Restore the normal level when you’re done and clear old logs using the app’s available options. Don’t delete all evidence before troubleshooting; save the small amount relevant to the current issue first. If you suspect an app is constantly transferring data, temporarily pause its sync task for comparison instead of changing proxy protocols. This helps distinguish the cost of processing traffic from activity generated by the traffic source.
If only one server causes heat or increased battery use, compare connection stability, retransmissions, and transport requirements on the same network. Some networks may cause more retries when handling UDP, but that doesn’t mean a particular protocol always uses more power. Test another existing, fully configured entry while keeping other activity the same. Don’t set keep-alive values to extremes: overly frequent keep-alives can increase activity, while overly sparse ones may let network equipment drop the connection and trigger more reconnections.
If the device is noticeably hot, pause unnecessary high-volume tasks and let it return to a normal temperature before comparing results. Charging, downloading, and high screen brightness add other variables. If unusual battery drain continues with the connection off, check system battery usage for other activity. If it reliably happens only when connected, note the minimal configuration, network type, observation period, and redacted logs, then follow the connection troubleshooting steps. Reproducible conditions are more useful for diagnosis than a screenshot of usage percentages, and make it easier to tell whether a setting change actually helped.
08 / Recovery and device differences
Issues after updates, restoring on a new device, and iPad-specific checks
Trace app updates separately from configuration changes
An issue appearing “after an update” doesn’t prove the app update caused it. First identify everything that changed around the same time: app update, system update, subscription refresh, Config replacement, or network change. Save your current configuration and compare it with an earlier saved version. If the old configuration works but the new one doesn’t, look for differences in rules, DNS, and policy references. If both fail, compare the network and server. Don’t start by deleting the app; that removes local state and turns a problem you could compare step by step into a full environment restore.
Manage app updates for iPhone / iPad through the App Store; check the App Store listing for system requirements. Mac, Windows, Android, and Linux users can get the latest installer from the download page under /dl/. On Mac, check About This Mac first to confirm whether you have Apple Silicon or Intel. If the system blocks the first launch, verify the prompt before proceeding: on Windows, select “More info → Run anyway” in SmartScreen; on Mac, choose “Open Anyway” under Privacy & Security; on Android, confirm the source before allowing that app to install; on Linux, follow the permissions for your package format. See installation prompts and permissions for full steps.
Restore backups without spreading credentials or the problem
Before reinstalling or moving to a new device, use the export options actually provided by the app to save Config, custom rules, and any server entries you need. Confirm that the backup can be opened rather than keeping only a screenshot of the list. Backups may contain passwords, user IDs, and subscription tokens. Store them somewhere protected that you control, and don’t share them publicly. When restoring, import only what you need to start. Connect using one existing server, then restore rules, DNS, and On Demand in stages. If you restore everything at once and the problem returns, it’s much harder to identify what caused it.
iPhone / iPad users who already purchased the app can look for it under Purchased using the account that made the purchase. Check the account region and purchase record as shown in the App Store. Identify the genuine app by its official icon, developer Shadow Launch Technology Limited, and app ID 932747118—not by a similar name. The iPhone / iPad App Store app is a one-time $2.99 purchase; the Mac / Windows / Android / Linux versions are free. Buying the client doesn’t include a connection service from an existing provider, and restoring the app won’t restore expired subscription credentials.
iPad: Re-establish the network baseline, even with the same configuration
If Shadowrocket works on iPhone but fails on iPad, don’t assume the tablet version is different. Confirm that both devices really use the same Wi-Fi, server, Config, and Global Routing setting. Entries with the same name may still have different settings after an update. Some iPads don’t have cellular connectivity, so you can’t directly repeat the phone’s network-switch test; use another network you control or a personal hotspot instead. On public Wi-Fi, each device may need to authenticate separately. Signing in on the phone doesn’t mean the iPad has completed sign-in.
On iPad, judge the connection by system status and actual request results after using split view, switching apps, or locking the screen. Whether Shadowrocket remains in the foreground doesn’t tell you whether the connection works. If the problem happens only after waking the device, note the sequence from wake to network recovery. Check whether Wi-Fi reconnects, On Demand triggers, and the server needs a new handshake. First disconnect and reconnect manually. If that restores access, investigate On Demand conditions and network switching. Don’t use a permanently lit screen or leaving the app in the foreground as a workaround.
After troubleshooting, restore the Global Routing, On Demand, and logging settings you need for everyday use, and keep a working configuration backup. If the same issue returns, start with the comparison conditions you’ve already verified before testing new variables.
Back to the top of this page ↑