127.0.0.1 / FAQ index
Clash FAQ
If a configuration won’t import, start with the subscription and YAML. If the proxy is on but nothing changes, check whether connections are reaching the core. Choose a topic below to expand the relevant answers.
127.0.0.1
Basics
Start by understanding the app, configuration, and how traffic gets routed.
What do a Clash client, the Mihomo core, and a subscription each do?
The client imports configurations, displays logs, and lets you change settings. Clients that use Mihomo rely on the core to handle listening, DNS resolution, and rule matching. A subscription is one way to obtain a configuration; it is not activated automatically when you install the client. If the app opens but connections fail, first check that the core is running, then confirm the current configuration is enabled, and finally test a node. If the log reports an error while loading the configuration, fix that error first—switching nodes will not resolve a startup problem.
What’s the difference between 127.0.0.1, 192.168.1.1, and 0.0.0.0?
127.0.0.1 is the local IPv4 loopback address. If you enter it on a phone, it points to the phone itself, not your computer. 192.168.1.1 is a private address often used as a router gateway; its actual purpose depends on your network settings. When used as a listen address, 0.0.0.0 binds to all local IPv4 interfaces—it is not a proxy server address for other devices. To use a proxy only on your computer, use the loopback address and make sure LAN sharing is off.
Can I use the mixed-port as the dashboard port?
mixed-port is the proxy listener that accepts both HTTP and SOCKS connections. external-controller sets the listening address for the core’s management API, which compatible dashboards use to read status or change policies; it cannot forward browser traffic. When configuring an app to use the proxy, use the listening address and port currently shown in your client, not example numbers from a guide. Bind a management API intended for local use to the loopback address. Don’t expose it to your LAN while troubleshooting connection issues.
What traffic do Rule, Global, and Direct mode handle?
All three modes only handle connections routed to the core. Rule mode checks rules in order and uses the target specified by the first match. Global mode sends connections through the outbound selected by the global policy. Direct mode connects without using a proxy node. Direct mode does not turn off the client or undo proxy settings already configured in apps. Temporarily switching modes can help distinguish rule issues from outbound issues; switch back to your usual mode when testing is done.
mixed-port
Installation & Setup
Check each step, from importing a file to starting the core.
After installing for the first time, how can I confirm an imported configuration is active?
Import a local YAML file or subscription URL from the client’s Profiles or Subscriptions page. Once the configuration appears in the list, select it and apply it. A successful import only means the client received the file; it does not mean the file replaced the active configuration. Return to the connection page and confirm the newly enabled configuration is shown. Then select an available outbound in a proxy group, visit a known-working site in your browser, and check the connection log for the destination domain and outbound used.
Clash subscription update failed: how can I tell whether the link is invalid or the fetch failed?
Keep the old configuration if it still works, then check the update log. For a 401, 403, or 404, visit the subscription provider’s site to check your account, whether the link has changed, and whether requests are restricted by their source. A status code alone does not prove the subscription has expired. For a timeout or DNS resolution error, first check whether your current network can reach the subscription service, then see how the client is trying to update it. If the response is a login page instead of a configuration file, get a fresh subscription URL. Don’t include the full URL in a public screenshot.
YAML import fails with a parse error. What should I check first?
Start with the line number and field name in the log, and check the line above the reported error too: a missing colon, quote, or indentation on the previous line can cause the problem. Indent with spaces, not tabs, and align items at the same list level. If the error mentions an unknown field or unsupported proxy type, check the documentation for your current core; don’t mistake a compatibility issue for an indentation error. Make changes to a copy of the configuration, save and reload it, and replace your everyday configuration only after the core starts successfully.
TUN reports insufficient permissions or fails to create a virtual network adapter. What should I do?
Check whether the log shows a failure while creating the virtual adapter or while changing routes. On desktop, install the service component as documented by your client, or run it as an administrator if you’re authorized to do so; also check that the client service is running. On mobile, if traffic is routed through the system VPN interface, make sure the required system permission is granted. If another VPN or virtual adapter is active, check for route conflicts. Don’t loosen permissions on a device managed by your organization; try the proxy settings supported by the app instead.
DIRECT
Usage Tips
Use new connections to check rules, outbounds, and DNS results.
Why do connection logs still show the old node after I change a proxy group?
Check the matched rule and proxy group name in the connection details. A rule may reference a different group or point directly to DIRECT, so changing the group you’re looking at may not affect that connection. URL-test and fallback groups also choose an outbound based on their own test results and ordering. Existing long-lived connections may keep using the previous outbound. Close the test page or restart the app, make a new request, and check the new log entry rather than relying on historical records.
Only some domains fail to load. How can I tell if DNS is the problem?
Distinguish a DNS resolution failure in the log from a connection timeout after successful resolution; the latter is not necessarily a DNS issue. If only your browser is affected, check whether it uses its own Secure DNS setting. For work or intranet domains, confirm whether your organization requires a specific DNS service before switching to a public DNS provider. Note the original settings before making changes, change one DNS setting at a time, and retry the same domain in the same app to see what helped.
How can my phone use a proxy port on a computer on the same LAN?
Enable LAN access only on a trusted network. On your phone, enter the computer’s private IPv4 address on that LAN and the actual HTTP or mixed proxy port. Make sure the computer is listening on that interface and its firewall allows connections from your phone. If the address and port are correct but the connection still fails, check whether Wi-Fi client isolation is enabled. Not every app uses the phone’s Wi-Fi proxy settings. Turn off sharing when you’re done, and don’t expose the listening port to the public internet.
How can I change the rule for one domain without editing the whole subscription?
If your client supports overrides or merging, put custom rules in a separate override configuration and check their order after merging. Rule mode uses the first matching rule, so place a specific domain rule before any broader rule that could match first, and keep MATCH as the final fallback. Confirm the target proxy group exists, then make a new connection and check the log. If the client has no override feature, you can test with a local copy of the subscription configuration, but that copy won’t receive future updates automatically.
127.0.0.1 / Listening
Troubleshooting
First check whether the request reached the core, then trace what happens next.
Does a node failing a latency test mean the node is broken?
A timeout only means the node did not complete a request to the configured test URL in time. Compare it with another node using the same test URL. If only one node fails, check its connection log. If several fail, check your network, whether the test URL is reachable, and whether the subscription needs updating. If the test succeeds but browsing still fails, confirm that the actual connection uses the same outbound. The test and the app may connect to different destinations, so their results are not interchangeable.
Why aren’t my browser or terminal using the proxy even though the system proxy is on?
Make a new request and check the client log. If there’s no entry, verify that the system proxy address and port match the core’s listener. For browsers, check extensions, separate proxy settings, and bypass lists. Terminal tools may not use the system proxy, so check the tool’s documentation to confirm its proxy options or environment variables are set correctly. If the connection appears in the log, the request reached the core; check the matched rule and outbound instead of repeatedly toggling the system proxy.
On Windows, what should I do if only UWP apps can’t connect to the local proxy?
First, use a regular desktop browser to confirm the same local proxy address and port work. Then check whether the affected app actually runs in an app container subject to loopback restrictions. Installing an app from the Store does not necessarily mean it is subject to this restriction. If the container cannot connect to the loopback address, and your device management policy allows it, use the client’s loopback settings to add an exception for that app only. Fully exit the app before testing again. The exception only allows loopback access; it won’t make an app that doesn’t read proxy settings start using a proxy.
The startup log says a port is in use. Which port should I change?
Check the address and port that failed to bind in the log, then identify whether they’re in use by an old client process, another proxy, or a different service. Don’t terminate processes you don’t recognize. If the conflict is on the mixed proxy port, choose an unused port in the client, restart the core, and update the system proxy or app settings to match. If the management API port is in use, change the port in its address and update the dashboard connection settings too. Finally, confirm in the log that the listener started.