If Discord will not load, messages keep spinning, or voice chat stays on “RTC Connecting” while Clash is enabled, the cause is usually one of four things: the selected proxy group is not actually proxying Discord, the chosen node is unhealthy, a rule sends part of Discord traffic DIRECT, or DNS/TUN handling differs between web and UDP traffic. The quickest way to diagnose it is to change one variable at a time and test text, media, and voice separately. This guide shows how to check proxy mode, rules, DNS, and TUN without replacing your entire profile or adding broad rules you do not need.

Identify What Is Failing First

Discord uses multiple services and connection types. The app can load its interface while failing to fetch images, or text chat can work while voice remains stuck. Treat those as separate symptoms rather than assuming that “Discord is broken” means every Discord connection follows the same path.

SymptomLikely area to checkUseful first test
Discord website or app does not loadProxy mode, node health, or domain rulesSwitch to a known-working node and test another site through the same group
Text works, but images or attachments do notCDN domains, DNS, or a rule that only covers the main siteCheck whether media domains are routed through the intended group
Messages work, but voice stays on connectingUDP support, TUN capture, or voice traffic routingTest another UDP-capable node and inspect the connection log during a voice attempt
Only one device or client has the problemSystem proxy, permissions, TUN, or app-specific settingsCompare the same profile and node on another device or client

Before changing configuration, note the client and core in use. Clash Verge, Clash Verge Rev, Clash for Android, and other front ends can expose different controls, while the core may be Mihomo or another Clash-compatible implementation. Labels and feature support therefore vary. Also check Discord’s service status and your local network: a service outage, captive portal, restrictive workplace network, or unstable Wi-Fi can look like a proxy-rule failure.

For a clean baseline, choose a node that already works for ordinary HTTPS browsing. Confirm that the active proxy group points to that node, then test Discord again. A successful latency check alone is not proof that a node can reach every destination, and it does not necessarily test UDP voice traffic.

Check Proxy Mode and Node Health

In Rule mode, Clash evaluates rules from top to bottom and uses the first matching rule. In Global mode, traffic is sent through the selected global proxy group, which is useful as a temporary diagnostic because it reduces the number of domain-specific rules involved. In Direct mode, traffic bypasses the proxy. A common first check is to switch temporarily from Rule to Global, select a known-working proxy, and test Discord. If Discord begins working, the profile’s routing rules are a stronger suspect than the Discord installation itself.

Do not leave Global mode enabled as a permanent fix unless that is what you intend. It can send unrelated traffic through the proxy, and it may conceal a rule-order problem. Return to Rule mode after the comparison and inspect the actual matched rule in the client’s connection or log panel. If the log shows Discord traffic using DIRECT, the route is not what you expected. If it shows the intended proxy but the connection still fails, investigate node reachability, DNS, or protocol support next.

  • Test a second node. Keep the same mode and rules, changing only the selected node. If the result changes, the original node or its route to Discord may be the issue.
  • Check the proxy group selection. A rule can point to a group whose current selection is DIRECT, a failed node, or a different group than the one you are monitoring.
  • Compare another HTTPS destination. If several unrelated sites fail through the same node, troubleshoot the node or client connection before editing Discord rules.
  • Review recent changes. A subscription update, profile merge, rule-provider update, or client upgrade can change behavior even when the visible profile name stays the same.
Keep tests controlled: Change one setting at a time and record the outcome. Switching mode, node, DNS, and TUN simultaneously may restore service, but it will not tell you which change fixed the fault.

Inspect Discord Rules and Their Order

A rule for discord.com alone may not cover every host used by the app. Discord can use separate domains for the API, invites, media, attachments, and content delivery. If the main interface loads but media does not, check the connection log for the specific failing hostname rather than guessing. Then add only the necessary domain rules, directing them to a proxy group that actually contains a working proxy.

The following example illustrates the placement and format. Replace Proxy with a proxy group name that exists in your configuration. Put service-specific rules above broad rules such as a catch-all DIRECT rule, and check that no earlier rule-provider entry matches the same traffic first.

Example Discord Domain Rules
rules:
  - DOMAIN-SUFFIX,discord.com,Proxy
  - DOMAIN-SUFFIX,discord.gg,Proxy
  - DOMAIN-SUFFIX,discordapp.com,Proxy
  - DOMAIN-SUFFIX,discordapp.net,Proxy
  - DOMAIN-SUFFIX,discord.media,Proxy
  - DOMAIN-SUFFIX,discordcdn.com,Proxy
  # Keep these above broader matching rules, then retain your existing rules below.

This is a starting point, not a guarantee that every Discord connection will be matched by a domain rule. Some connections may be represented by an IP address in the log, especially when DNS mode, fake-IP mapping, or TUN capture affects what the core sees. If you find an IP-based match, do not automatically add a wide IP range to the profile: service address ranges can change, and an overly broad CIDR rule may route unrelated traffic. First confirm the hostname, the matched rule, and whether the connection is actually failing.

Rule order matters because Clash normally stops at the first match. For example, if a broad rule-provider entry sends a domain to DIRECT before your Discord-specific entries are reached, adding another Discord rule at the bottom will not override it. Move the specific rule above the conflicting entry, reload the configuration, and verify the matched rule in the log. If the group name is invalid or the profile has not been reloaded, the expected route may never take effect.

Run a Controlled Test in the Client

Use this sequence in the middle of troubleshooting, after noting the original settings. It works with most Clash-compatible clients, although menu names and log details differ. Keep Discord open during the test so connection attempts appear in the client’s activity view.

  1. Confirm the baseline. Check that Clash is running, the profile is active, and the system proxy or TUN feature you normally rely on is enabled. Verify that a separate, known-working website loads through the currently selected proxy.
  2. Test Global mode briefly. Select Global mode and choose a working proxy group or node. Reopen Discord or reload the page. If it works only in Global mode, return to Rule mode and focus on rule matching and order.
  3. Test a second node. Keep Global mode and all other settings unchanged. If one node works and another fails, the problem is likely node-specific, route-specific, or related to that node’s UDP support rather than a general Discord rule issue.
  4. Return to Rule mode and inspect the log. Start Discord again and look for requests involving Discord domains. Record the destination, matched rule, selected group, and result. A DIRECT match points toward rule ordering; a proxy match followed by a timeout points toward connectivity or the selected node.
  5. Test functions separately. Check sign-in and text, then load an image or attachment, and finally join a voice channel. This separates HTTPS and media problems from the UDP path used by voice.
  6. Change DNS or TUN only if the evidence points there. Make one adjustment, reload or restart the affected app if needed, and repeat the same test. Restore the prior setting if there is no measurable improvement.
Do not treat a successful ping as a complete test. A node’s latency probe may use an HTTP URL and TCP connection. Discord voice can need UDP, so a node may appear healthy while voice connections still fail.

Check DNS and TUN for App and Voice Traffic

DNS can affect Discord in several ways. A domain may resolve through a local resolver while the connection itself is expected to use the proxy, or a cached result may remain after changing DNS settings. Fake-IP mode can make domain-based matching more reliable in some setups, but applications that depend on address behavior may need additional care. The important diagnostic question is not simply “which DNS mode is best?” but “what hostname or address does Clash see, and which rule does it apply?”

Open the connection log while Discord is attempting the failing action. If the request is shown under a hostname and matches the intended proxy rule, DNS is less likely to be the first problem. If it appears only as an IP, or the log shows a different destination than expected, check whether the client uses fake-IP, whether DNS hijacking is enabled, and whether another DNS service or local security tool is intercepting queries. After a DNS change, restart Discord or clear its network state so it does not keep using stale connections or cached answers.

TUN mode captures traffic at the network interface level, which can include applications that do not obey the operating system’s HTTP proxy setting. This is often relevant to desktop voice traffic. If Discord’s web interface works through the system proxy but voice does not, check whether TUN is enabled, whether the client has permission to create its virtual interface, and whether another VPN or firewall is competing for traffic capture. On mobile, inspect the operating system’s VPN permission and battery or background restrictions as well.

Voice traffic is commonly UDP-based, and not every proxy protocol, node, or client configuration handles UDP in the same way. Confirm that the selected node and proxy group support UDP and that the relevant client/core options are enabled. A domain rule may correctly route Discord’s HTTPS requests while voice packets fail because UDP is unsupported or not captured. If switching to another UDP-capable node fixes voice without changing the rules, that is strong evidence against adding more domain entries.

ObservationWhat it suggestsNext action
Global mode fixes all Discord functionsRule selection or rule order is likely involvedInspect the matched rule in Rule mode and move specific entries above conflicting rules
Text works but voice fails on one nodeUDP support or node-specific routing is a likely causeTest another node known to support UDP and review TUN capture
Only media or attachments failA related CDN or media hostname may be routed differentlyUse the log to identify the failing hostname and add a narrow rule if needed
Discord works in a browser but not the desktop appSystem-proxy handling, app permissions, or UDP capture may differCompare TUN and system proxy behavior, then test the app again
All applications fail after enabling TUNInterface permissions, routing conflicts, or DNS interception may be involvedDisable competing VPNs, check permissions, and temporarily compare without TUN

Frequently Asked Questions

Why does Discord work in Global mode but not Rule mode?

Global mode bypasses most domain-specific routing decisions by sending traffic through the selected global proxy. If that fixes Discord, Rule mode may be matching a DIRECT rule, selecting an unintended group, or encountering a broad rule-provider entry before the Discord-specific rule. Inspect the connection log in Rule mode and correct the first matching rule instead of leaving all traffic in Global mode by default.

Why do messages work while voice chat stays on connecting?

Text and voice do not necessarily use the same transport. Text and media commonly use HTTPS, while voice relies on UDP connections that may not be supported by every proxy node or captured by every setup. Test a UDP-capable node, confirm TUN or the client’s network capture is working, and check whether the failure follows the node or the device.

Should I add Discord IP ranges to my rules?

Usually, start with the hostname and the connection log. Service IPs can change, and broad IP rules may catch unrelated traffic or become stale. Add an IP-based rule only when the log shows that IP matching is genuinely required, you have a reliable source for the range, and you understand how the rule interacts with your existing routing policy.

Will changing DNS always fix Discord?

No. DNS changes can help when a hostname resolves incorrectly or domain-based routing is not working as intended, but they cannot make an unhealthy node reachable or add UDP support to a proxy that lacks it. Use the matched-rule and connection evidence first, then change DNS only when the observed behavior points to name resolution or DNS interception.

Once Discord works in Rule mode for text, media, and voice, keep the smallest effective change: a corrected group selection, a narrowly placed domain rule, a working UDP-capable node, or a repaired DNS/TUN setting. Re-test after updating the profile or switching nodes, since rule providers and server routes can change over time. A controlled configuration is easier to maintain—and much easier to troubleshoot the next time one part of Discord stops connecting.

Get Started

Take Full Control of Your Traffic with Clash

Available on Windows, macOS, Linux, Android, and iOS. Flexible rules, simple setup, ready to use.

Download Free View Setup Guide →