pidIsAlive misidentifies own PID as duplicate runtime when lock file is stale

[34262315716]

Bug Description

When the OpenClaw gateway exits abnormally (e.g. TerminateProcess, power failure, OOM), the openclaw-runtime.lock/ directory and its owner.json remain on disk. On restart, if Windows assigns the same PID to the new process (common due to PID recycling), pidIsAlive() checks the stale owner PID which happens to be the current process's own PID and incorrectly reports that another OpenClaw instance is running.

Reproduction Steps

1. Start openclaw gateway (gets PID e.g. 12345)
2. Kill the process forcefully (not clean shutdown)
3. Start openclaw gateway again
4. If Windows assigns PID 12345 to the new process, the lock check fails

Root Cause

Missing self-PID check in pidIsAlive() at apps/memos-local-plugin/adapters/openclaw/runtime-lock.ts:104

The function process.kill(pid, 0) when pid === process.pid always succeeds (you can always signal yourself), so the function returns true even though the lock is from a previous instance.

Fix

Add one line in pidIsAlive() to exclude the current process PID:

if (pid === process.pid) return false;

Environment

• memos-local-plugin: 2.0.5
• OpenClaw: 2026.6.10
• OS: Windows 11
• Node: 22.x