openclaw - 💡(How to fix) Fix [Bug]: Performance regression in v2026.4.29: embedded-run prep stalls and gateway event-loop blockage [1 comments, 2 participants]

Bug type

Regression (worked before, now fails)

Beta release blocker

No

Summary

After upgrading from v2026.4.27 to v2026.4.29 in a Docker-based NAS deployment, gateway responsiveness became much worse, with embedded-run startup/prep delays and liveness warnings showing event-loop blockage.

Steps to reproduce

1. Start OpenClaw v2026.4.29 in the attached Docker/NAS gateway environment.
2. Connect through active channels/surfaces used in the log sample (webchat, Slack, Telegram).
3. Trigger normal gateway activity such as session listing, model listing, node listing, or an embedded agent turn.
4. Observe slow RPCs and liveness warnings in gateway logs, including long embedded-run startup/prep timings, websocket/Slack timeouts, and event-loop blockage indicators.

Expected behavior

In v2026.4.27 under the same deployment context, comparable slowdown was not observed; the gateway was expected to remain responsive without long per-turn prep delays or event-loop blockage warnings.

Actual behavior

After upgrading from v2026.4.27 to v2026.4.29 in a Docker-based NAS deployment, gateway responsiveness became much worse, with embedded-run startup/prep delays and liveness warnings showing event-loop blockage.

OpenClaw version

2026.4.29

Operating system

Linux 5.10.55+ (x64), Docker-based NAS deployment

Install method

docker

Model

openai-codex/gpt-5.4

Provider / routing chain

openclaw -> openai-codex

Additional provider/model setup details

Gateway log reports agent model: openai-codex/gpt-5.4. Channels active in the same runtime include Slack, Telegram, and webchat. Browser control sidecar is also enabled in this deployment.

Logs, screenshots, and evidence

Relevant log excerpts from v2026.4.29:

[agent/embedded] [trace:embedded-run] startup stages: totalMs=20529
- model-resolution: 10431ms
- auth: 5905ms
- attempt-dispatch: 4188ms

[agent/embedded] [trace:embedded-run] prep stages: totalMs=28532
- core-plugin-tools: 13506ms
- system-prompt: 5521ms
- stream-setup: 6521ms

[diagnostic] liveness warning:
- eventLoopDelayP99Ms=13522.4
- eventLoopDelayMaxMs=13522.4
- eventLoopUtilization=1
- cpuCoreRatio=1.01

Slow RPCs observed:
- [ws] models.list 23441ms
- [ws] models.list 39782ms
- [ws] node.list 15009ms
- [ws] node.list 17388ms
- [ws] device.pair.list 16331ms

Related symptoms:
- [slack] failed to start native approval handler: Error: gateway event loop readiness timeout
- [ws] handshake timeout
- [WARN] socket-mode:SlackWebSocket: A pong wasn't received from the server before the timeout of 15000ms
- [slack] socket disconnected (disconnect). retry 1/12 in 2s

Version comparison:
- Last known good: v2026.4.27
- First known bad: v2026.4.29

Impact and severity

Affected: This Docker/NAS gateway deployment and the connected channels/surfaces shown in the logs (webchat, Slack, Telegram) Severity: High (significant responsiveness degradation and timeout symptoms during normal gateway use) Frequency: Observed repeatedly in the attached log sample during the same runtime session Consequence: Slow or stalled agent turns, delayed RPC responses, and channel instability/timeouts

Additional information

Last known good version: 2026.4.27 First known bad version: 2026.4.29

This appears to be a regression observed immediately after upgrading. The timeout symptoms in Slack/websocket logs appear in the same window as the embedded-run timing spikes and event-loop blockage warnings.