fix: clean stale gateway PIDs before triggerOpenClawRestart calls launchctl/systemctl

SidQin-cyber · steipete · commit 63c6080d5000 · 2026-02-26T15:22:35.000Z
When the /restart command runs inside an embedded agent process (no SIGUSR1 listener), it falls through to triggerOpenClawRestart() which calls launchctl kickstart -k directly — bypassing the pre-restart port cleanup added in #27013. If the gateway was started via TUI/CLI, the orphaned process still holds the port and the new launchd instance crash-loops. Add synchronous stale-PID detection (lsof) and termination (SIGTERM→SIGKILL) inside triggerOpenClawRestart() itself, so every caller — including the embedded agent /restart path — gets port cleanup before the service manager restart command fires. Closes #26736 Made-with: Cursor
diff --git a/src/infra/restart.test.ts b/src/infra/restart.test.ts
@@ -0,0 +1,19 @@
+import { describe, expect, it } from "vitest";
+import { findGatewayPidsOnPortSync } from "./restart.js";
+
+describe("findGatewayPidsOnPortSync", () => {
+  it("returns an empty array for a port with no listeners", () => {
+    const pids = findGatewayPidsOnPortSync(19999);
+    expect(pids).toEqual([]);
+  });
+
+  it("never includes the current process PID", () => {
+    const pids = findGatewayPidsOnPortSync(18789);
+    expect(pids).not.toContain(process.pid);
+  });
+
+  it("returns an array (not undefined or null) on any port", () => {
+    const pids = findGatewayPidsOnPortSync(0);
+    expect(Array.isArray(pids)).toBe(true);
+  });
+});
diff --git a/src/infra/restart.ts b/src/infra/restart.ts
@@ -1,9 +1,11 @@
 import { spawnSync } from "node:child_process";
+import { resolveGatewayPort } from "../config/paths.js";
 import {
   resolveGatewayLaunchAgentLabel,
   resolveGatewaySystemdServiceName,
 } from "../daemon/constants.js";
 import { createSubsystemLogger } from "../logging/subsystem.js";
+import { resolveLsofCommandSync } from "./ports-lsof.js";
 
 export type RestartAttempt = {
   ok: boolean;
@@ -283,10 +285,106 @@ function normalizeSystemdUnit(raw?: string, profile?: string): string {
   return unit.endsWith(".service") ? unit : `${unit}.service`;
 }
 
+/**
+ * Find PIDs of gateway processes listening on the given port using synchronous lsof.
+ * Returns only PIDs that belong to openclaw gateway processes (not the current process).
+ */
+export function findGatewayPidsOnPortSync(port: number): number[] {
+  if (process.platform === "win32") {
+    return [];
+  }
+  const lsof = resolveLsofCommandSync();
+  const res = spawnSync(lsof, ["-nP", `-iTCP:${port}`, "-sTCP:LISTEN", "-Fpc"], {
+    encoding: "utf8",
+    timeout: SPAWN_TIMEOUT_MS,
+  });
+  if (res.error || res.status !== 0) {
+    return [];
+  }
+  const pids: number[] = [];
+  let currentPid: number | undefined;
+  let currentCmd: string | undefined;
+  for (const line of res.stdout.split(/\r?\n/).filter(Boolean)) {
+    if (line.startsWith("p")) {
+      if (currentPid != null && currentCmd && currentCmd.toLowerCase().includes("openclaw")) {
+        pids.push(currentPid);
+      }
+      const parsed = Number.parseInt(line.slice(1), 10);
+      currentPid = Number.isFinite(parsed) && parsed > 0 ? parsed : undefined;
+      currentCmd = undefined;
+    } else if (line.startsWith("c")) {
+      currentCmd = line.slice(1);
+    }
+  }
+  if (currentPid != null && currentCmd && currentCmd.toLowerCase().includes("openclaw")) {
+    pids.push(currentPid);
+  }
+  return pids.filter((pid) => pid !== process.pid);
+}
+
+const STALE_SIGTERM_WAIT_MS = 300;
+const STALE_SIGKILL_WAIT_MS = 200;
+
+/**
+ * Synchronously terminate stale gateway processes.
+ * Sends SIGTERM, waits briefly, then SIGKILL for survivors.
+ */
+function terminateStaleProcessesSync(pids: number[]): number[] {
+  if (pids.length === 0) {
+    return [];
+  }
+  const killed: number[] = [];
+  for (const pid of pids) {
+    try {
+      process.kill(pid, "SIGTERM");
+      killed.push(pid);
+    } catch {
+      // ESRCH — already gone
+    }
+  }
+  if (killed.length === 0) {
+    return killed;
+  }
+  spawnSync("sleep", [String(STALE_SIGTERM_WAIT_MS / 1000)], { timeout: 2000 });
+  for (const pid of killed) {
+    try {
+      process.kill(pid, 0);
+      process.kill(pid, "SIGKILL");
+    } catch {
+      // already gone
+    }
+  }
+  spawnSync("sleep", [String(STALE_SIGKILL_WAIT_MS / 1000)], { timeout: 2000 });
+  return killed;
+}
+
+/**
+ * Inspect the gateway port and kill any stale gateway processes holding it.
+ * Called before service restart commands to prevent port conflicts.
+ */
+function cleanStaleGatewayProcessesSync(): number[] {
+  try {
+    const port = resolveGatewayPort(undefined, process.env);
+    const stalePids = findGatewayPidsOnPortSync(port);
+    if (stalePids.length === 0) {
+      return [];
+    }
+    restartLog.warn(
+      `killing ${stalePids.length} stale gateway process(es) before restart: ${stalePids.join(", ")}`,
+    );
+    return terminateStaleProcessesSync(stalePids);
+  } catch {
+    return [];
+  }
+}
+
 export function triggerOpenClawRestart(): RestartAttempt {
   if (process.env.VITEST || process.env.NODE_ENV === "test") {
     return { ok: true, method: "supervisor", detail: "test mode" };
   }
+
+  cleanStaleGatewayProcessesSync();
+
   const tried: string[] = [];
   if (process.platform !== "darwin") {
     if (process.platform === "linux") {
