Symptom
On OpenClaw 2026.9.4, a stdio MCP server that does not finish initialize within 30 seconds takes the whole gateway down. Every channel drops with the main process.
Evidence what you'll see
- The log line is
MCP server "zai-mcp-server" timed out: did not complete initialize within 30s. - Six milliseconds later:
Unhandled promise rejection: Error: service child cleanup identity lost: anchor channel closed without a matching closing receipt. - systemd recorded
Main process exited, code=exited, status=1/FAILURE.
Why it happens
The timeout aborts the child while cleanup still expects a closing receipt on the anchor channel. That rejection is unhandled, so the gateway process exits 1. A cold npx -y start is what pushed the report past 30 seconds.
Affected versions
Ubuntu, npm-global OpenClaw 2026.9.4, a systemd user service, one main agent. The server command in the report is npx with args -y and @z_ai/mcp-server@latest. The linked issue does not name a Node version.
Fix steps
0 of 3 done
Your check marks stay in this browser only.
How to check it worked
- A server that logs
did not complete initialize within 30sdoes not get followed byservice child cleanup identity lostandstatus=1/FAILURE.
If it still fails
- The stability bundle path in the log is
~/.openclaw/logs/stability/with the suffixunhandled_rejection.json. The linked issue does not describe this exit on a build that includes pull request 144941.
Sources and credit
Summarized in our own words from these public reports. Credit goes to their authors.
Provided as is, without warranty. Try changes in a non-production environment and back up first. Security-related steps: verify against the official docs. Terms · Request removal or attribution change
Hit an error we don't cover? Submit it