Clash FAQs
and Troubleshooting

Mihomo is the core responsible for proxy protocols, rule matching, DNS, and connection forwarding. Clients such as Clash Plus, Clash Verge Rev, and FlClash provide interfaces for importing configurations, switching policies, and controlling the system proxy. When troubleshooting, distinguish interface issues, configuration errors, and core connection errors; only the latter two usually appear in runtime logs.

A configuration file is the YAML content the client actually loads, typically containing proxy nodes, policy groups, rules, DNS, and port fields. A subscription URL is the address used to fetch that file; updating a subscription downloads and replaces the associated Profile. Before editing a local copy generated from a subscription, check whether updates will overwrite your changes. Rules that must persist are better kept in an override or merged configuration.

Rule mode matches traffic from top to bottom against the rules in the configuration and is best for everyday use. Global mode sends traffic through one proxy policy and is useful for quickly checking whether a node works. Direct mode bypasses the proxy and helps determine whether Clash is causing the problem. During troubleshooting, switch between Global and Direct mode to identify whether the issue is with the node, rules, or local network.

The system proxy only handles apps that follow the operating system’s proxy settings. Browsers and most desktop apps usually work, but some games, command-line tools, and store apps may bypass it. TUN mode uses a virtual network interface to capture a broader range of traffic and requires additional system permissions. Start with the system proxy; enable TUN only after confirming that an app is bypassing it, then check routing and DNS settings.

Fake-IP returns reserved addresses for domains while Clash maintains the domain mapping, offering efficient rule matching for most standard configurations. Redir-Host returns real DNS results and is more compatible with apps that rely on LAN discovery, special domain validation, or fixed resolutions. If printers, casting, corporate intranets, or some games behave unexpectedly, first add the relevant domains to fake-ip-filter instead of changing the entire DNS mode.

Open the client’s connection records or log page, revisit the target website, and check the domain, destination address, rule type, and final policy group. Rules are matched from top to bottom in configuration order, so broad DOMAIN-SUFFIX or GEO rules may intercept a request first. If the result is unexpected, move the more specific DOMAIN rule above the broad rule and verify the override order again after updating the subscription.

A client latency test usually measures the time for one HTTP request to a specified test URL. It does not fully reflect packet loss, peak-hour congestion, line bandwidth, the target website’s location, or connection reuse. Evaluate node quality using repeated-test stability, actual download speed, and the target site’s response as well. A node with consistently higher latency may feel faster than one with a lower number but frequent timeouts.

mixed-port accepts both HTTP and SOCKS5 connections and is useful when you want to maintain only one local port. port provides HTTP proxy access only, while socks-port provides SOCKS5 access only. When an app uses a manually configured proxy, its protocol and port must match. If the client is already using mixed-port 7890, no other program can listen on the same address and port. After changing it, update the system proxy or the app’s proxy settings as well.

First confirm that dns.enable is enabled, then check whether fields such as nameserver, proxy-server-nameserver, and fallback match the syntax supported by the current core. Query the same domain in both Direct and proxied states, comparing the system result, DNS requests in the Clash logs, and the actual connection target. An app’s own encrypted DNS may bypass both the system and Clash DNS; disable it in the app or explicitly capture it with TUN and rules.

Give each Profile a name that identifies its source and purpose, and keep its update schedule and override configuration separate. Subscription content is best updated by the server, while local rules belong in a separate file or the client’s Merge or Override feature. After switching Profiles, review the policy groups again because different configurations may use the same policy names with different nodes. Before deleting an old configuration, confirm that no automation script or system service still references it.