Bug type

Regression (worked before, now fails)

Summary

Intermittent local gateway websocket handshake failures on loopback (ws://127.0.0.1:18789) cause frequent CLI failures (openclaw cron list) and break Antfarm cron setup (workflow ensure-crons).

This reproduces on latest available build in this environment:

OpenClaw: 2026.3.12 (6472949)
Host: Ubuntu 24.04 VPS
Gateway bind: loopback (127.0.0.1:18789)

Steps to reproduce

1) Confirm gateway appears healthy

openclaw gateway status

Observed: running, probe ok, listening on 127.0.0.1:18789.

2) Repeated CLI calls to gateway

for i in {1..50}; do
openclaw cron list >/dev/null && echo "ok $i" || echo "fail $i"
sleep 0.5
done

Observed in this run:

• ok: 16
• fail: 34
• ~68% failure rate

Typical failure:

gateway connect failed: Error: gateway closed (1000):
Error: gateway closed (1000 normal closure): no close reason
Gateway target: ws://127.0.0.1:18789
Source: local loopback
Config: /home/sai/.openclaw/openclaw.json
Bind: loopback

3) Antfarm failure path

antfarm workflow ensure-crons healthcare-ai-research

Observed error:

Failed to create cron job for agent "scoper": CLI fallback failed: Error: gateway connect failed: Error: gateway client stopped
Error: gateway timeout after 120ms
Gateway target: ws://127.0.0.1:18789
...

Gateway logs (journalctl)

Repeated patterns during failures:

• handshake timeout ... remote=127.0.0.1
• closed before connect ... host=127.0.0.1:18789 ... code=1000
• occasional code=1008 reason=connect challenge timeout

Representative excerpts:

[ws] handshake timeout conn=... remote=127.0.0.1
[ws] closed before connect conn=... remote=127.0.0.1 ... host=127.0.0.1:18789 ... code=1000 reason=n/a
[ws] closed before connect conn=... remote=127.0.0.1 ... code=1008 reason=connect challenge timeout

Expected behavior

Expected behavior

• Local loopback websocket connections for CLI commands should be stable.
• openclaw cron list should not fail intermittently under normal local load.
• Connect challenge should not time out this frequently on localhost.

Actual behavior

Actual behavior

• Frequent handshake/connect challenge failures on loopback.
• CLI commands that need gateway connectivity fail intermittently.
• Higher-level workflows (Antfarm cron setup) become unreliable and can leave partial state.

OpenClaw version

2026.3.12

Operating system

Ubuntu 24.04 VPS

Install method

npm global

Model

anthropic/opus-4.6

Provider / routing chain

openclaw

Config file / key location

No response

Additional provider/model setup details

No response

Logs, screenshots, and evidence

Repeated CLI calls to gateway

for i in {1..50}; do
  openclaw cron list >/dev/null && echo "ok $i" || echo "fail $i"
  sleep 0.5
done

Observed in this run:

• ok: 16
• fail: 34
• ~68% failure rate

Impact and severity

Unable to manage things through CLI (i.e crons). Breaks antfarm workflows

Additional information

Notes

• This is not an auth misconfiguration: same config/environment succeeds intermittently.
• Failures occur while other websocket clients can still be active.
• Workaround used locally: make Antfarm workflow run tolerant when crons already exist, but root issue appears to be gateway WS handshake stability.