openclaw - 💡(How to fix) Fix [Bug]: v2026.5.24-beta.2: Gateway extremely slow still — RPCs take 30-67s, provider auth prewarm blocks event loop 83s (Windows 11)

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…

Root Cause

  1. Provider auth pre-warm blocks the event loop for 83 seconds: provider auth state pre-warmed in 83108ms eventLoopMax=46942.7ms — the event loop is blocked for up to 47 seconds during prewarm.
  2. RPCs are consistently slow on every connection: agents.list ~33s, models.list ~31-35s, sessions.list ~47-67s. These never return quickly.
  3. The problem persists across reconnections: refreshing the page doesn't help, the slowness is server-side.
  4. Webchat disconnects due to timeout: code=1006 disconnects happen because the gateway is too slow to respond.
  5. Downgrading to v2026.5.20 resolves everything immediately (confirmed in #85877).

Fix Action

Fix / Workaround

Workaround / 临时方案

Downgrade to v2026.5.20:

npm install -g openclaw@2026.5.20 --force
openclaw gateway install --force

Code Example

15:51:44 [ws] webchat connected
15:51:44 [ws] ⇄ res ✓ health 278ms cached=true
15:52:18 [ws] ⇄ res ✓ agents.list 34087ms       ← 34 seconds!
15:52:18 [ws] ⇄ res ✓ models.authStatus 95ms
15:52:18 [ws] ⇄ res ✓ commands.list 286ms
15:52:53 [ws] ⇄ res ✓ models.list 34571ms        ← 35 seconds!
15:53:05 [ws] ⇄ res ✓ sessions.list 47013ms      ← 47 seconds!
15:53:05 [ws] ⇄ res ✓ chat.history 47181ms       ← 47 seconds!
15:53:06 [ws] webchat disconnected code=1006
15:53:07 [gateway] provider auth state pre-warmed in 83108ms eventLoopMax=46942.7ms  ← 83s prewarm, 47s event loop block!

---

15:53:07 [ws] webchat connected
15:53:40 [ws] ⇄ res ✓ agents.list 32939ms
15:54:11 [ws] ⇄ res ✓ models.list 31140ms
15:54:47 [ws] ⇄ res ✓ sessions.list 67034ms      ← 67 seconds!
15:54:48 [ws] ⇄ res ✓ chat.history 67204ms       ← 67 seconds!

---

15:55:53 [ws] webchat connected
15:56:26 [ws] ⇄ res ✓ agents.list 32694ms
15:56:57 [ws] ⇄ res ✓ models.list 30981ms
15:57:31 [ws] ⇄ res ✓ sessions.list 65394ms      ← 65 seconds!
15:57:31 [ws] ⇄ res ✓ chat.history 65566ms       ← 66 seconds!

---

15:58:34 [ws] ⇄ res ✓ models.list 31353ms
15:59:08 [ws] ⇄ res ✓ sessions.list 66378ms      ← 66 seconds!

---

npm install -g openclaw@2026.5.20 --force
openclaw gateway install --force
RAW_BUFFERClick to expand / collapse

Environment / 环境

  • OS: Windows 11 x64 (Build 26200)
  • Node.js: v24.15.0
  • Shell: PowerShell
  • OpenClaw: v2026.5.24-beta.2
  • Previous working version: v2026.5.20
  • Related issue: #85877 (same problem on v2026.5.22-beta.1, still not fixed)

Summary / 概要

Gateway startup and Control UI are extremely slow on v2026.5.24-beta.2. Simple WebSocket RPCs like agents.list, models.list, sessions.list take 30-67 seconds each. Provider auth pre-warm takes over 83 seconds. The gateway is essentially unusable after startup.

This same type of slow startup issue appeared in some April versions, was fixed back then, but has now regressed again starting from v2026.5.22 and persists through v2026.5.24-beta.2. This is a recurring regression.

网关启动和 Control UI 在 v2026.5.24-beta.2 上极其缓慢。简单的 WebSocket RPC 如 agents.listmodels.listsessions.list 每个都需要 30-67 秒。Provider auth 预热超过 83 秒。网关启动后基本不可用。

这种启动缓慢的问题曾在 4 月份的部分版本中出现过,后来修复了,但从 v2026.5.22 开始又再次回归,v2026.5.24-beta.2 仍然存在。这是一个反复出现的回退。

Steps to Reproduce / 复现步骤

  1. Update to v2026.5.24-beta.2: npm install -g [email protected] --force
  2. Run openclaw gateway start or openclaw doctor — both are extremely slow
  3. Open Control UI in browser, observe blank/slow loading

Observed Gateway Logs / 网关日志

After simply loading the gateway and refreshing the web page once:

15:51:44 [ws] webchat connected
15:51:44 [ws] ⇄ res ✓ health 278ms cached=true
15:52:18 [ws] ⇄ res ✓ agents.list 34087ms       ← 34 seconds!
15:52:18 [ws] ⇄ res ✓ models.authStatus 95ms
15:52:18 [ws] ⇄ res ✓ commands.list 286ms
15:52:53 [ws] ⇄ res ✓ models.list 34571ms        ← 35 seconds!
15:53:05 [ws] ⇄ res ✓ sessions.list 47013ms      ← 47 seconds!
15:53:05 [ws] ⇄ res ✓ chat.history 47181ms       ← 47 seconds!
15:53:06 [ws] webchat disconnected code=1006
15:53:07 [gateway] provider auth state pre-warmed in 83108ms eventLoopMax=46942.7ms  ← 83s prewarm, 47s event loop block!

Second connection after auto-reconnect — even worse:

15:53:07 [ws] webchat connected
15:53:40 [ws] ⇄ res ✓ agents.list 32939ms
15:54:11 [ws] ⇄ res ✓ models.list 31140ms
15:54:47 [ws] ⇄ res ✓ sessions.list 67034ms      ← 67 seconds!
15:54:48 [ws] ⇄ res ✓ chat.history 67204ms       ← 67 seconds!

Third connection — same pattern:

15:55:53 [ws] webchat connected
15:56:26 [ws] ⇄ res ✓ agents.list 32694ms
15:56:57 [ws] ⇄ res ✓ models.list 30981ms
15:57:31 [ws] ⇄ res ✓ sessions.list 65394ms      ← 65 seconds!
15:57:31 [ws] ⇄ res ✓ chat.history 65566ms       ← 66 seconds!

And after some time, repeating the same slow pattern:

15:58:34 [ws] ⇄ res ✓ models.list 31353ms
15:59:08 [ws] ⇄ res ✓ sessions.list 66378ms      ← 66 seconds!

Key Observations / 关键发现

  1. Provider auth pre-warm blocks the event loop for 83 seconds: provider auth state pre-warmed in 83108ms eventLoopMax=46942.7ms — the event loop is blocked for up to 47 seconds during prewarm.

  2. RPCs are consistently slow on every connection: agents.list ~33s, models.list ~31-35s, sessions.list ~47-67s. These never return quickly.

  3. The problem persists across reconnections: refreshing the page doesn't help, the slowness is server-side.

  4. Webchat disconnects due to timeout: code=1006 disconnects happen because the gateway is too slow to respond.

  5. Downgrading to v2026.5.20 resolves everything immediately (confirmed in #85877).

  6. Provider auth 预热阻塞事件循环 83 秒eventLoopMax=46942.7ms,事件循环被阻塞长达 47 秒

  7. 每次连接的 RPC 都很慢agents.list ~33s, models.list ~31-35s, sessions.list ~47-67s

  8. 重连后依然缓慢:刷新页面无法改善,问题在服务端

  9. Webchat 因超时断开code=1006 断连是因为网关响应太慢

  10. 降级到 v2026.5.20 立即恢复正常(已在 #85877 确认)

Workaround / 临时方案

Downgrade to v2026.5.20:

npm install -g openclaw@2026.5.20 --force
openclaw gateway install --force

Related Issues / 相关 Issue

  • #85877 — Our original report on v2026.5.22-beta.1 (same problem, still open)
  • #83963 — Control UI startup RPCs take 5-10s on v2026.5.18 (macOS)
  • #86599 — Local model calls block gateway event loop on v2026.5.24-beta.1 (Windows)
  • #86752 — Docker/WSL2 gateway event-loop starvation, 284s provider-auth prewarm
  • #76884 — OpenClaw on native Windows getting slower with each new version
  • #63257 — Windows Gateway API timeout 30s at startup (April version, was fixed)

Expected Behavior / 期望行为

Gateway RPCs should respond in under 1-2 seconds. Provider auth prewarm should not block the event loop. v2026.5.20 works perfectly, so this is clearly a regression introduced after that version.

网关 RPC 应在 1-2 秒内响应。Provider auth 预热不应阻塞事件循环。v2026.5.20 完全正常,所以这是之后版本引入的回退。

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 [Bug]: v2026.5.24-beta.2: Gateway extremely slow still — RPCs take 30-67s, provider auth prewarm blocks event loop 83s (Windows 11)