· Troubleshooting · 8 min read
Clash proxy timeouts: troubleshooting subscriptions, DNS, and system time
When all Clash proxies time out or a single proxy won’t connect, check your subscription, local port, DNS, system time, and rules to find and fix the cause.
1. Check that your subscription is up to date and active
First, tell apart a failed subscription update, a timed-out latency test, and an app that can’t connect. An update request goes to the subscription server; a latency test reaches its test URL; and the app connects to its actual destination. These requests may take different routes, so one failure doesn’t tell you whether the others succeeded. Stay on the same network, test two proxies, and note which step fails along with the first related error in the client logs.
- Check the subscription update result. Manually update it from the client’s subscription or profile page, then check the update time and error details.
401or403means the server denied the request; first check whether the link was revoked or your account permissions changed. For404, check the subscription URL. For429, wait as instructed by the server before trying again. If the request times out, check whether the subscription domain resolves and is reachable before assuming the subscription has expired. - Check the active configuration. A successful update doesn’t mean the core has loaded the new configuration. Check the selected profile, proxy list, and load logs. If you use
proxy-providers, check each provider’s update status separately; a provider may still be using an old cache even when the main configuration loads successfully. - Look for a single failure or a shared cause. If only one proxy fails in the same group and against the same URL, check that proxy’s server address, port, protocol, and transport settings first. If several proxies fail, check whether they share an entry domain or latency-test URL. If the configuration load log says a protocol or field is unsupported, check which core and version are actually running—not just the client version.
2. Check that the local proxy port is listening
Test whether the app can reach the core with an explicit proxy request; there’s no need to switch on the system proxy and TUN at the same time. Check the active configuration for mixed-port or port: the former accepts HTTP and SOCKS connections, while the latter is an HTTP proxy port. The examples below assume an HTTP proxy listening on 127.0.0.1:7890. 7890 is just an example, not a universal default. In PowerShell, use curl.exe; on macOS and Linux, use curl.
Windows: check the listener, then send a request
Get-NetTCPConnection -LocalPort 7890 -State Listen
curl.exe --noproxy "" -v --proxy http://127.0.0.1:7890 --connect-timeout 5 --max-time 15 https://example.com/
- No listener found, or the connection is refused immediately: Check that the core is running, the port number matches, and it’s listening on the expected address. If the log shows
address already in use, check the listener’sOwningProcessto identify the process using the port. If you change the port, update any apps still configured to use the old one. - Connected to the local port, but the request later times out: The local listener is unlikely to be the main issue. Check the core logs for the same time and identify whether the failure occurs during proxy-domain resolution, the remote connection, the handshake, or the connection to the destination site. The 5- and 15-second values in the command are diagnostic thresholds, not actual latency or measures of proxy quality.
- The destination site returned an HTTP response: This request passed through the local proxy entry point, but it may not have used the selected proxy. A rule may route
example.comtoDIRECT; check the connection log to see which outbound was actually used. If the explicit request works but your usual app doesn’t, check whether the app reads the system proxy settings or its traffic is routed through TUN.
3. Check DNS separately for the proxy server and destination
The proxy server address is the server the core connects to for outbound traffic; the destination domain is the site the app wants to reach. They may be resolved by different resolvers. If the system can resolve the destination site, that doesn’t mean the core can resolve the proxy server. And using an IP address for the proxy server doesn’t eliminate the need to resolve destination domains. Check the logs first to see which hostname is failing.
Proxy server address won’t resolve
On Windows, run Resolve-DnsName -Name example.net -Type A and Resolve-DnsName -Name example.net -Type AAAA separately, replacing example.net with the proxy server’s actual hostname. This lets you compare system DNS results; it doesn’t prove that mihomo uses the same resolver. Having no AAAA record isn’t a problem by itself. If the logs repeatedly show timeouts when connecting to an IPv6 address, check whether your network has a working IPv6 route.
If you use mihomo with built-in DNS enabled and your configuration explicitly sets dns.proxy-server-nameserver, check whether the upstream resolver for proxy server domains is reachable. If that resolver itself must be accessed through a proxy that hasn’t connected yet, you may have a dependency loop. Choose an upstream resolver that’s independently reachable on your network, then retest with the original proxy. Don’t replace the proxy hostname with an IP from a single DNS lookup: the address may rotate, and doing so may break handshake settings that depend on the hostname.
Destination DNS issues or unexpected Fake-IP addresses
In fake-ip mode, if your configuration uses 198.18.0.1/16, it’s normal for the client to receive a synthetic address from that range. What matters is whether the subsequent connection returns to the same core that maintains the domain mapping—not whether the address is directly reachable. If only one app fails, check whether it uses its own DNS or bypasses the expected proxy entry point. Start by comparing with an explicit proxy request, then check TUN routing and DNS interception. Don’t switch to a different enhanced mode based on a single synthetic address.
4. Check system time and TLS errors
An incorrect device clock can affect certificate validity checks. Subscription updates, HTTPS DNS queries, and TLS-enabled proxy handshakes may all be affected. The client may show only a timeout even when the logs report x509: certificate has expired or is not yet valid. First identify whether the error comes from the subscription server, DNS upstream, or proxy server. If the first error is a connection timeout, correcting the clock usually won’t fix a routing or port issue.
- In Windows 11, open Settings → Time & language → Date & time, turn on Set time automatically, then select Sync now under Additional settings. Check the date and time zone as well. Certificate validation uses the actual system clock; changing the displayed time zone alone won’t correct the clock.
- On Linux with systemd, run
timedatectl statusand checkSystem clock synchronizedalong with the current UTC time. Check again after resuming a virtual machine, especially if the issue started after it came back from a paused state. - After correcting the clock, retry the same proxy. If you still see an expired certificate, hostname mismatch, or untrusted certificate error, check the relevant server certificate’s validity period, the server name in the configuration, and the certificate chain. Don’t hide the error by disabling certificate verification.
5. Check the matched rule and actual outbound
A successful latency test only shows that the proxy reached its test URL during that test; it doesn’t mean your app’s request used that proxy. Keep Rule mode enabled and retry the failing destination. In the connections view or logs, check the destination domain, matched rule, proxy group, and final outbound. Rules are generally matched in order: once an earlier rule matches, later exceptions won’t apply. The final MATCH rule acts as a fallback.
- Matched
DIRECTorREJECT: Check whether this is intentional or a broader rule appears before the exception for the destination domain. If you use a remote ruleset throughRULE-SET, also check whether its provider loaded successfully and whether it’s using an outdated cache. - The right proxy group matched, but a different proxy was used: Check the group’s current selection.
selectuses the currently selected option;url-testchooses automatically based on test results, so the proxy you see in the interface may not be the one handling the request. For nested groups, follow the connection log to the final group. - No corresponding connection appears in the core logs: First confirm the log is for this request, not a reused connection. Close and reopen the app, then test again. If an explicit proxy request appears in the logs but the app’s request doesn’t, focus on system proxy settings, the app’s own proxy settings, and TUN interception—not on rewriting domain rules.