openclaw - 💡(How to fix) Fix [Bug]: auth phase takes ~8.5s on Windows vs ~1-2s on Mac (same version, same model)

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…

The auth phase of embedded-run takes consistently ~8.5 seconds on Windows, while the same setup on macOS takes only ~1-2 seconds. This adds significant latency to every message response.

Root Cause

The auth phase of embedded-run takes consistently ~8.5 seconds on Windows, while the same setup on macOS takes only ~1-2 seconds. This adds significant latency to every message response.

Fix Action

Fix / Workaround

liveness warning: reasons=event_loop_delay eventLoopDelayP99Ms=1961.9 eventLoopDelayMaxMs=11307.8 eventLoopUtilization=0.785 cpuCoreRatio=0.779 Event loop is blocked for up to 11 seconds during runs.

Full startup stages (typical run)

workspace: 0ms@0ms runtime-plugins: 112ms@112ms hooks: 0ms@112ms model-resolution: 112ms@224ms auth: 8522ms@8746ms ← bottleneck context-engine: 0ms@8746ms attempt-workspace: 4ms@8750ms attempt-prompt: 0ms@8750ms attempt-runtime-plan: 2ms@8752ms attempt-dispatch: 0ms@8752ms

RAW_BUFFERClick to expand / collapse

Bug type

Regression (worked before, now fails)

Beta release blocker

No

Summary

Description

The auth phase of embedded-run takes consistently ~8.5 seconds on Windows, while the same setup on macOS takes only ~1-2 seconds. This adds significant latency to every message response.

Environment

  • OS: Windows 10 x64 (10.0.19045)
  • Node.js: v24.14.1
  • OpenClaw: 2026.5.20 (e510042)
  • Provider: zai/glm-5-turbo (via open.bigmodel.cn, custom baseUrl)
  • Same config works fast on macOS with identical OpenClaw version and model

Steps to reproduce

Evidence

Trace logs (Windows)

auth:8522ms@8746ms totalMs=8752 auth:8549ms@8781ms totalMs=9378 auth:8718ms@8950ms totalMs=9615 auth:9211ms@9576ms totalMs=9615 auth:9580ms@8781ms totalMs=9850 auth:8897ms@8781ms totalMs=9150 auth:9580ms@8781ms totalMs=9850

Every single run shows auth in the 8-10 second range, consistently.

Liveness warning

liveness warning: reasons=event_loop_delay eventLoopDelayP99Ms=1961.9 eventLoopDelayMaxMs=11307.8 eventLoopUtilization=0.785 cpuCoreRatio=0.779 Event loop is blocked for up to 11 seconds during runs.

Full startup stages (typical run)

workspace: 0ms@0ms runtime-plugins: 112ms@112ms hooks: 0ms@112ms model-resolution: 112ms@224ms auth: 8522ms@8746ms ← bottleneck context-engine: 0ms@8746ms attempt-workspace: 4ms@8750ms attempt-prompt: 0ms@8750ms attempt-runtime-plan: 2ms@8752ms attempt-dispatch: 0ms@8752ms

Network is fast (not a network issue)

Direct API calls to open.bigmodel.cn take ~200-300ms. All DNS resolution is fast (~18ms). Ping to bigmodel CN is 12ms. The delay is entirely within OpenClaw's internal auth phase.

model-prewarm sidecar

sidecars.model-prewarm: 3730ms

The model-prewarm sidecar also takes ~3.7s, which may be contributing to event loop pressure.

Things already tried

  • ✅ Cleared task history (236 → 0 records, eliminated ~12s session-resource-loader overhead)
  • ✅ Added Windows Defender exclusions for .openclaw, openclaw node_modules, and node.exe (no improvement)
  • ✅ Tested with VPN on/off (no difference)
  • ✅ Verified DNS, TLS, and HTTP connectivity are all fast
  • ✅ Disabled IPv6 (not the issue)

Expected behavior

Expected behavior

auth phase should complete in ~1-2 seconds on Windows, similar to macOS.

Actual behavior

Event loop is blocked for up to 11 seconds during runs.

Full startup stages (typical run)

OpenClaw version

2026.5.20

Operating system

windows 10

Install method

npm

Model

zai/glm-5-turbo

Provider / routing chain

gateway → direct → zai (open.bigmodel.cn) → glm-5-turbo

Additional provider/model setup details

No response

Logs, screenshots, and evidence

Impact and severity

No response

Additional information

No response

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…

FAQ

Expected behavior

auth phase should complete in ~1-2 seconds on Windows, similar to macOS.

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 [Bug]: auth phase takes ~8.5s on Windows vs ~1-2s on Mac (same version, same model)