openclaw - 💡(How to fix) Fix 2026.4.29 TUI stays on connecting and spins CPU while gateway is healthy [1 comments, 2 participants]

Official PRs (…)
ON THIS PAGE

Recommended Tools

×6

Utilities matched from this issue’s tags and category — try them while you read without losing context.

GitHub issue graph ai analysis

Paste a GitHub issue URL. We fetch that issue, discover linked issues from bodies/comments/timeline, collect linked pull requests, and produce a structured English report.

The report is written in English Markdown for sharing and archival.

Helpful · Quick feedback

Loading…
GitHub stats
openclaw/openclaw#75800Fetched 2026-05-02 05:29:52
View on GitHub
Comments
1
Participants
2
Timeline
3
Reactions
2
Timeline (top)
closed ×1commented ×1cross-referenced ×1

After upgrading OpenClaw to 2026.4.29 on macOS, openclaw tui --session main stayed visually stuck on connecting | idle even though the gateway was healthy and the TUI process had an established WebSocket connection. The openclaw-tui child process burned about 100% CPU until killed. Rolling back to 2026.4.27 restored normal TUI startup.

Root Cause

Rolling back to 2026.4.27 fixed the TUI. After rollback, I also had to quarantine the stale generated plugin registry because it referenced a 2026.4.29 bundled plugin manifest that no longer existed:

Fix Action

Workaround

Rolling back to 2026.4.27 fixed the TUI. After rollback, I also had to quarantine the stale generated plugin registry because it referenced a 2026.4.29 bundled plugin manifest that no longer existed:

~/.openclaw/plugins/installs.json

Then openclaw doctor --fix rebuilt the plugin registry for 2026.4.27. A fresh openclaw tui --session main then reached:

gateway connected | idle

and CPU returned to idle.

Code Example

connecting | idle
agent main | session main | unknown | tokens ?

---

~/.openclaw/plugin-runtime-deps/openclaw-2026.4.27-...   ~926M
~/.openclaw/plugin-runtime-deps/openclaw-2026.4.29-...   ~3.9G after rebuild

---

~/.openclaw/plugins/installs.json

---

gateway connected | idle
RAW_BUFFERClick to expand / collapse

Summary

After upgrading OpenClaw to 2026.4.29 on macOS, openclaw tui --session main stayed visually stuck on connecting | idle even though the gateway was healthy and the TUI process had an established WebSocket connection. The openclaw-tui child process burned about 100% CPU until killed. Rolling back to 2026.4.27 restored normal TUI startup.

Environment

  • macOS on Apple Silicon
  • Shell: fish/zsh
  • Node used by OpenClaw wrapper: /opt/homebrew/opt/node@24/bin/node (v24.15.0)
  • Broken version: OpenClaw 2026.4.29 (a448042)
  • Working rollback version: OpenClaw 2026.4.27 (cbc2ba0)
  • Gateway bind/port: 127.0.0.1:18789

Symptoms

  • openclaw tui --session main rendered the initial TUI frame, then stayed at:
connecting | idle
agent main | session main | unknown | tokens ?
  • openclaw-tui sat around ~100% CPU.
  • lsof showed an established TCP/WebSocket connection between the TUI process and the gateway on 127.0.0.1:18789.
  • Gateway health/status calls succeeded while the TUI was stuck.
  • Non-TUI agent calls worked, for example openclaw agent --session-id main --message ... --json returned successfully.

Diagnostics

Gateway was not the blocker:

  • openclaw gateway call health --timeout 10000 --json returned "ok": true.
  • openclaw gateway status --deep showed the LaunchAgent loaded, runtime running, connectivity probe OK, and admin-capable.
  • Dashboard returned HTTP 200.

Profiling the stuck openclaw-tui process on 2026.4.29 pointed at plugin/runtime dependency path resolution, with hot frames around functions such as:

  • normalizePluginLoaderAliasMapForJiti
  • collectRuntimePackageWildcardImportTargets
  • isPathInside
  • resolve from node:path

and bundled files similar to:

  • boundary-path-*.js
  • sdk-alias-*.js
  • loader-*.js

The generated runtime dependency directory for 2026.4.29 had also grown very large compared to prior versions:

~/.openclaw/plugin-runtime-deps/openclaw-2026.4.27-...   ~926M
~/.openclaw/plugin-runtime-deps/openclaw-2026.4.29-...   ~3.9G after rebuild

Before rebuild, the 2026.4.29 runtime dependency directory was about 9.4G. Quarantining and rebuilding it with openclaw doctor --fix reduced the size but did not fix the TUI spin.

Things Tried

  • Fixed config permissions from 644 to 600.
  • Removed stale disabled plugin config stubs for acpx and memory-core from plugins.allow / plugins.entries. This removed config warnings but did not fix the TUI hang.
  • Restarted gateway and verified gateway health.
  • Rebuilt plugin runtime deps with openclaw doctor --fix.
  • Tried openclaw tui --local --session main; it reached local ready | idle but still spun CPU.
  • Tried --history-limit 0; still spun.
  • Tried updating nested @mariozechner/pi-tui from 0.70.6 to 0.71.1; still spun.

Workaround

Rolling back to 2026.4.27 fixed the TUI. After rollback, I also had to quarantine the stale generated plugin registry because it referenced a 2026.4.29 bundled plugin manifest that no longer existed:

~/.openclaw/plugins/installs.json

Then openclaw doctor --fix rebuilt the plugin registry for 2026.4.27. A fresh openclaw tui --session main then reached:

gateway connected | idle

and CPU returned to idle.

Expected

openclaw tui --session main should connect promptly when the gateway is healthy and the WebSocket is established, without pegging CPU in plugin runtime path resolution.

Actual

On 2026.4.29, the TUI appeared stuck at connecting | idle and burned CPU even though the gateway and WebSocket connection were healthy.

extent analysis

TL;DR

The most likely fix is to roll back to OpenClaw version 2026.4.27 due to issues with plugin runtime path resolution in version 2026.4.29.

Guidance

  • Verify that the gateway is healthy and the WebSocket connection is established before attempting to connect to the TUI.
  • Check the size of the generated runtime dependency directory for excessive growth, which may indicate a problem with plugin runtime path resolution.
  • Try rebuilding the plugin runtime dependencies with openclaw doctor --fix to see if it resolves the issue.
  • If the issue persists, consider rolling back to a previous version of OpenClaw, such as 2026.4.27, which has been reported to work correctly.

Example

No code snippet is provided as the issue appears to be related to a specific version of OpenClaw and its interaction with plugin runtime dependencies.

Notes

The exact cause of the issue is unclear, but it appears to be related to changes in plugin runtime path resolution in OpenClaw version 2026.4.29. Rolling back to a previous version may be the most reliable workaround until the issue is resolved.

Recommendation

Apply the workaround of rolling back to OpenClaw version 2026.4.27, as it has been reported to resolve the issue. This is recommended because the problem appears to be specific to version 2026.4.29 and rolling back to a previous version has been shown to work correctly.

Vote matrix · Quick signals

Works
Did the solution work? Tap to confirm.
Easy Fix
Was it a quick fix?
Time Saver
Did it save you time?
Blocking
Was it severely blocking?
Common Issue
Are others likely hitting this too?
Flaky / Intermittent
Is it intermittent?
Verified / Reproducible
Can you reproduce it reliably?
Loading…

Still need to ship something?

×6

Another batch ranked right after the header list — different links, same matching logic.

Back to top recommendations

TRENDING

openclaw - 💡(How to fix) Fix 2026.4.29 TUI stays on connecting and spins CPU while gateway is healthy [1 comments, 2 participants]