Connectivity Troubleshooting · Clash Technical Blog

Fix Clash Verge Rev GLOBAL node selection

No GLOBAL group in Clash Verge Rev v2.5.9, and changing nodes has no effect? Update the app to v2.5.10, choose the exit in GLOBAL, verify it with a new connection, and keep your configuration available for rollback.

Read the articleBack to blog
  • Clash Verge Rev
  • v2.5.10
  • GLOBAL
  • Global mode
  • Node selection
On this page
Published
Estimated reading time
About 8 minutes

The Proxies page already shows Global, but the list still contains your subscription groups, and picking another node does not change the global exit. First check whether GLOBAL is present. Avoid starting with a subscription reset, DNS changes, or a service reinstall.

Why does changing nodes have no effect after switching to Global?

The user in official issue #8516 upgraded from v2.5.8 to v2.5.9 while already using Global mode. Reopening the proxy list showed the Rule-mode groups, and changing their selections did not affect the global settings. The report was from Windows 11; it does not establish testing on every platform.

The shared proxy-list code in v2.5.9 omitted the separate GLOBAL group in Global mode. The official stable v2.5.10 release, published on 11/3/2026, includes the fix. This guide addresses that specific regression, rather than treating every Global-mode connection failure as the same fault.

Identify the case from what the page actually shows

What you seeWhat to check next
In v2.5.9, Global mode shows subscription groups but no GLOBALCheck the app update and GLOBAL selection using this guide
GLOBAL is present, but DIRECT or an unavailable node is selectedCheck the actual group selection and whether the node works
A message says no subscription is imported or activatedImport and activate a valid subscription first
The core is unavailable, or the Proxies page is emptySave the first core error, then investigate communication or startup

GLOBAL controls the global exit; an arbitrary group does not

The official Mihomo mode documentation states that global mode requires selecting a proxy or policy in GLOBAL. Groups familiar from Rule mode, such as Node selection, may remain visible. This does not mean that clicking an arbitrary node in one of them changes the current global exit.

If GLOBAL points to another policy group, the final exit also depends on that group's selection. If it points directly to a node, changes to other groups will usually leave that path unchanged. For the test, select a known working node directly in GLOBAL to remove the extra layer of ambiguity.

Global changes the outbound choice for connections that have already entered the core. It does not automatically capture every application on the device. System proxy, TUN, or an application's own proxy settings bring traffic into the core; there is no need to enable every capture method just to restore GLOBAL.

Official fix commit 3027e2b restores GLOBAL in the Global-mode list while retaining subscription groups. Seeing other groups in v2.5.10 is therefore part of the retained behavior. A list that still resembles Rule mode does not, by itself, mean the fix failed.

Record both modes' selections and save your configuration before updating

Record the full app version, the current mode, and the selected members of the existing policy groups and GLOBAL. If GLOBAL is absent, record that fact. Keep a recoverable external copy of the current Profile, overrides, and important settings. Do not paste a whole subscription or credential-bearing configuration into a public report.

If Rule mode already provides a working route, you can temporarily use it and record its original selections. Avoid repeatedly changing shared policy groups to compensate for the missing group: that can also change everyday routing when you return to Rule and lose the previously working state.

Keep these details before upgrading

  • The operating system, full app version, mode before and after the problem, and what the page shows.
  • An independent backup of the current Profile and custom overrides, plus the original group selections.
  • A known working node or policy for a real-request check after the upgrade.
  • The first redacted error or necessary screenshot; do not upload subscription links, passwords, or complete log bundles.

Update to v2.5.10, then select the actual exit in GLOBAL

The fix is in the Clash Verge Rev app's list-rendering logic. Updating Mihomo, refreshing the subscription, or restarting the core does not replace this app update. The verified stable fix is v2.5.10; a later stable release with evidence of the same fix does not need to be downgraded.

Choose the correct system and architecture on the official v2.5.10 Release page in the references. This maintenance run did not update the installers hosted by this site. The old v2.5.8 mirror is not the fix package for this issue. Keep your normal installation method and existing configuration directory.

Restore independent global selection

  1. Exit normally, then update the app

    Save your work and configuration first. When a brief interruption is acceptable, turn off the capture method you use as needed, fully exit the app from the tray, and install through the supported official process. Do not merely close the window, replace the core by hand, or delete app data.

  2. Reopen and check the full app version

    Confirm that v2.5.10, or a later stable release with evidence of this fix, is actually running and that your subscription and overrides are intact. Open Proxies and confirm that a valid subscription is active and the core can supply proxy information.

  3. Switch to Global and find GLOBAL

    Choose Global on the Proxies page, find the restored GLOBAL group, and expand it. Other subscription groups may also appear. Select the exit in GLOBAL; you do not need to change every group.

  4. First select a known working node directly

    For this check, choose a node you already know works. Make sure you have not accidentally selected DIRECT, REJECT, or another unavailable route. If you normally point to a policy group, restore it after the test and check its final selection.

Check the exit with a new connection, then return to your usual mode

Use your normal system proxy or TUN capture method and make a fresh real request to a site you are allowed to test. An existing long-lived connection may keep its previous route. Close the test connection or test application before trying again; an old page still loading does not prove the new selection took effect.

On the app's Connections page, check that the new entry's proxy chain corresponds to the selected node or the policy's final exit, and confirm the request succeeds. A latency result, the Global label, or the reappearance of GLOBAL alone does not prove that this request used the intended route.

After testing, restore Rule or your usual mode and check that the recorded group selections have not been disturbed. If you intend to keep using Global, verify both the policy selected in GLOBAL and that policy's downstream choice.

Checklist for this fix

  • The running app includes the fix corresponding to 3027e2b, the current subscription is valid, and the core is available.
  • GLOBAL is visible in Global mode, and you can choose the exit in that group.
  • A new test connection shows the chain matching the actual selection, and the request succeeds.
  • Other groups remaining visible are not mistaken for GLOBAL disappearing again.
  • Your normal connections still work after restoring the usual mode and original selections.

This site has not reproduced this issue or tested the upgrade in a real client. These steps are based on the official user report, stable Release, fixed-version code, and mode documentation. Website builds, checks in five languages, and browser acceptance do not verify the proxy route on your device.

If problems remain, keep a working mode and recover by symptom

Already on v2.5.10, but GLOBAL is still missing

Check the version of the app you actually reopened and the mode on the normal Proxies page. Record whether a valid subscription and core information are available, and capture the list. Preserve logs for a report rather than clearing the configuration directory.

GLOBAL can be selected, but new requests still fail

First check whether the connection enters the core, what GLOBAL currently selects, and whether the target node works. Then investigate network or DNS errors from the first error message instead of treating the missing-group regression as the cause of everything.

The upgraded app or original configuration no longer works

Stop adding changes and save the current state and error separately. Use the project's supported recovery process with the previously working, safe app version and a matching backup. Do not let old and new apps share the only copy of the data directory.

If you cannot update yet, you can continue working in Rule mode once you have confirmed it works. That does not repair global selection in v2.5.9. Avoid downgrading merely to restore the old page layout; older versions may have other known issues.

When reporting upstream, include the system, full app version, upgrade source, current mode, whether GLOBAL exists, the actual selection, and the first error plus a redacted chain for a new connection. Do not publish full subscriptions, credentials, or connection records containing sensitive destinations.

References