Start the daemon after a reboot hands its old PID to another process (#5277)

Closes #5257

packages/server/src/server/daemon-instance.test.ts

@@ -0,0 +1,96 @@
+import { spawn, type ChildProcess } from "node:child_process";
+import { existsSync } from "node:fs";
+import { mkdtemp, readFile, rm, writeFile } from "node:fs/promises";
+import { tmpdir, uptime } from "node:os";
+import { join } from "node:path";
+import { afterEach, beforeEach, describe, expect, test } from "vitest";
+
+import { readDaemonInstance, stopDaemonInstance } from "./daemon-instance.js";
+import { acquirePidLock, getPidLockInfo, isLocked, type PidLockInfo } from "./pid-lock.js";
+
+// A process cannot have started before the machine booted, so a lock stamped
+// before this boot names a PID that some unrelated process now holds.
+function bootedAt(): number {
+  return Date.now() - uptime() * 1000;
+}
+
+async function writeLock(paseoHome: string, lock: PidLockInfo): Promise<void> {
+  await writeFile(join(paseoHome, "paseo.pid"), JSON.stringify(lock));
+}
+
+function lockFor(pid: number, startedAt: Date): PidLockInfo {
+  return {
+    pid,
+    startedAt: startedAt.toISOString(),
+    hostname: "old-host",
+    uid: process.getuid?.() ?? 0,
+    listen: "127.0.0.1:6767",
+    desktopManaged: true,
+    heartbeat: true,
+  };
+}
+
+describe("daemon instance identity across a reboot", () => {
+  let paseoHome: string;
+  let bystander: ChildProcess | undefined;
+
+  beforeEach(async () => {
+    paseoHome = await mkdtemp(join(tmpdir(), "paseo-daemon-instance-"));
+  });
+
+  afterEach(async () => {
+    bystander?.kill("SIGKILL");
+    bystander = undefined;
+    await rm(paseoHome, { recursive: true, force: true });
+  });
+
+  test("a lock stamped before this boot has no running owner", async () => {
+    await writeLock(paseoHome, lockFor(process.pid, new Date(bootedAt() - 60 * 60_000)));
+
+    expect(await readDaemonInstance(paseoHome)).toBeNull();
+    expect(await isLocked(paseoHome)).toMatchObject({ locked: false });
+  });
+
+  test("a supervisor started during this boot still holds the lock", async () => {
+    await writeLock(paseoHome, lockFor(process.pid, new Date()));
+
+    expect(await readDaemonInstance(paseoHome)).toMatchObject({ pid: process.pid });
+    expect(await isLocked(paseoHome)).toMatchObject({ locked: true });
+  });
+
+  test("a new supervisor takes over a lock stamped before this boot", async () => {
+    await writeLock(paseoHome, lockFor(process.pid, new Date(bootedAt() - 60 * 60_000)));
+
+    await acquirePidLock(paseoHome, null, { ownerPid: process.pid + 10_000 });
+
+    expect(await getPidLockInfo(paseoHome)).toMatchObject({ pid: process.pid + 10_000 });
+  });
+
+  test("stopping a lock stamped before this boot leaves the process holding that pid alone", async () => {
+    // Records delivery rather than dying of it, so a signal cannot be missed by arriving late.
+    const signalMarker = join(paseoHome, "bystander-signalled");
+    bystander = spawn(
+      process.execPath,
+      [
+        "-e",
+        `process.on("SIGTERM", () => require("node:fs").writeFileSync(${JSON.stringify(signalMarker)}, "SIGTERM"));` +
+          `setTimeout(() => {}, 120_000);`,
+      ],
+      { stdio: "ignore" },
+    );
+    const bystanderPid = bystander.pid;
+    if (bystanderPid === undefined) throw new Error("bystander process did not start");
+    let exited = false;
+    bystander.once("exit", () => {
+      exited = true;
+    });
+
+    await writeLock(paseoHome, lockFor(bystanderPid, new Date(bootedAt() - 60 * 60_000)));
+
+    expect(await stopDaemonInstance(paseoHome)).toMatchObject({ action: "not_running" });
+
+    expect(existsSync(signalMarker)).toBe(false);
+    expect(exited).toBe(false);
+    await expect(readFile(join(paseoHome, "paseo.pid"), "utf-8")).rejects.toThrow(/ENOENT/);
+  });
+});

packages/server/src/server/daemon-instance.ts

@@ -4,7 +4,7 @@ import path from "node:path";
 import { setTimeout as delay } from "node:timers/promises";
 import {
   getPidLockInfo,
-  isPidRunning,
+  isPidLockOwnerRunning,
   isSamePidLock,
   releasePidLock,
   type PidLockInfo,
@@ -44,7 +44,7 @@ export class DaemonInstanceError extends Error {
 
 export async function readDaemonInstance(home: string): Promise<PidLockInfo | null> {
   const lock = await getPidLockInfo(home);
-  return lock && isPidRunning(lock.pid) ? lock : null;
+  return lock && isPidLockOwnerRunning(lock) ? lock : null;
 }
 
 export function daemonLogPath(home: string): string {
@@ -160,7 +160,7 @@ export async function stopDaemonInstance(
       `Supervisor changed for ${home}; refusing to stop PID ${instance.pid}.`,
     );
   }
-  if (!instance || !isPidRunning(instance.pid)) {
+  if (!instance || !isPidLockOwnerRunning(instance)) {
     if (instance)
       await releasePidLock(home, { ownerPid: instance.pid, startedAt: instance.startedAt });
     return {
@@ -175,7 +175,7 @@ export async function stopDaemonInstance(
   let { forced, usedLifecycleRpc } = await requestInstanceStop(home, instance, options);
   const waitForExit = async (waitMs: number) => {
     const exitDeadline = Date.now() + waitMs;
-    while (isPidRunning(instance.pid)) {
+    while (isPidLockOwnerRunning(instance)) {
       const current = await getPidLockInfo(home);
       if (current && !isSamePidLock(instance, current))
         throw new DaemonInstanceError(

packages/server/src/server/pid-lock.test.ts

@@ -73,7 +73,7 @@ describe("pid-lock ownership", () => {
         pidPath,
         JSON.stringify({
           pid: process.pid,
-          startedAt: "2026-01-01T00:00:00.000Z",
+          startedAt: new Date().toISOString(),
           hostname: "old-host",
           uid: process.getuid?.() ?? 0,
           listen: "127.0.0.1:6767",
@@ -106,7 +106,7 @@ describe("pid-lock ownership", () => {
         pidPath,
         JSON.stringify({
           pid: process.pid,
-          startedAt: "2026-01-01T00:00:00.000Z",
+          startedAt: new Date().toISOString(),
           hostname: "old-host",
           uid: process.getuid?.() ?? 0,
           listen: "127.0.0.1:6767",
@@ -138,7 +138,7 @@ describe("pid-lock ownership", () => {
         pidPath,
         JSON.stringify({
           pid: process.pid,
-          startedAt: "2026-01-01T00:00:00.000Z",
+          startedAt: new Date().toISOString(),
           hostname: "old-host",
           uid: process.getuid?.() ?? 0,
           listen: "127.0.0.1:6767",
@@ -169,7 +169,7 @@ describe("pid-lock ownership", () => {
         pidPath,
         JSON.stringify({
           pid: process.pid,
-          startedAt: "2026-01-01T00:00:00.000Z",
+          startedAt: new Date().toISOString(),
           hostname: "old-host",
           uid: process.getuid?.() ?? 0,
           listen: "127.0.0.1:6767",

packages/server/src/server/pid-lock.ts

@@ -2,7 +2,7 @@ import { open, readFile, unlink, utimes } from "node:fs/promises";
 import type { FileHandle } from "node:fs/promises";
 import { ensurePrivateDirectory } from "./private-files.js";
 import { join } from "node:path";
-import { hostname } from "node:os";
+import { hostname, uptime } from "node:os";
 import { z } from "zod";
 
 export const pidLockInfoSchema = z.object({
@@ -40,7 +40,7 @@ const PID_LOCK_HEARTBEAT_INTERVAL_MS = 30_000;
 const PID_LOCK_READ_RETRY_ATTEMPTS = 10;
 const PID_LOCK_READ_RETRY_DELAY_MS = 50;
 
-export function isPidRunning(pid: number): boolean {
+function isPidRunning(pid: number): boolean {
   try {
     process.kill(pid, 0);
     return true;
@@ -49,6 +49,30 @@ export function isPidRunning(pid: number): boolean {
   }
 }
 
+// `uptime()` reports whole seconds on some platforms, so the derived boot instant carries
+// about a second of error. This covers that resolution and nothing more: every extra second
+// is a window in which a lock written just before a reboot still reads as current.
+const BOOT_INSTANT_TOLERANCE_MS = 5_000;
+
+function precedesThisBoot(startedAt: string): boolean {
+  const stamped = Date.parse(startedAt);
+  if (Number.isNaN(stamped)) return false;
+  return stamped < Date.now() - uptime() * 1000 - BOOT_INSTANT_TOLERANCE_MS;
+}
+
+/**
+ * Whether the process that wrote this lock is still running.
+ *
+ * A PID alone does not identify the supervisor: the operating system hands the number to
+ * something else once the supervisor is gone, and a reboot reassigns it freely. A process
+ * cannot predate the boot it runs under, so a lock stamped before this boot is abandoned
+ * however alive its PID looks.
+ */
+export function isPidLockOwnerRunning(lock: PidLockInfo): boolean {
+  if (precedesThisBoot(lock.startedAt)) return false;
+  return isPidRunning(lock.pid);
+}
+
 function getPidFilePath(paseoHome: string): string {
   return join(paseoHome, "paseo.pid");
 }
@@ -105,7 +129,7 @@ async function clearExistingPidLock(
   existingLock: PidLockInfo,
   lockOwnerPid: number,
 ): Promise<"already_owned" | "cleared"> {
-  const lockOwnerRunning = isPidRunning(existingLock.pid);
+  const lockOwnerRunning = isPidLockOwnerRunning(existingLock);
   if (existingLock.pid === lockOwnerPid && lockOwnerRunning) {
     await touchPidLockFile(pidPath);
     return "already_owned";
@@ -116,7 +140,7 @@ async function clearExistingPidLock(
   if (
     !confirmedLock ||
     !isSamePidLock(existingLock, confirmedLock) ||
-    isPidRunning(confirmedLock.pid)
+    isPidLockOwnerRunning(confirmedLock)
   ) {
     throw new PidLockError("PID lock changed while checking whether it was abandoned");
   }
@@ -339,7 +363,7 @@ export async function isLocked(
   if (!info) {
     return { locked: false };
   }
-  if (!isPidRunning(info.pid)) {
+  if (!isPidLockOwnerRunning(info)) {
     return { locked: false, info };
   }
   return { locked: true, info };