fix(fleet): make restart-lock release/break ownership-safe (review #680)

Addresses the reviewer's blocker (comment 15915): release() unconditionally
unlinked restart.lock, so after a stale/max-wait break an OLD owner could
delete a NEWER owner's lock, letting a third restart interleave and defeating
the guard.

- Each acquire writes a unique owner token (randomUUID) into the lock file.
- release() only unlinks while that token is still on disk; once another caller
  has broken and re-owned the lock, the timed-out original owner's release() is
  a no-op and leaves the new owner's lock intact.
- Breaking a stale/timed-out lock now takes ownership atomically via
  write-temp + rename (atomic replace) instead of a blind unlink-then-recreate;
  a breaker that loses a concurrent takeover reads back a foreign token and
  keeps waiting rather than assuming ownership.

Regression test (does not let a timed-out owner drop a lock another restart
broke and re-owned) reproduces the three-restart interleave: R1 hangs (stale),
R2 breaks + re-owns, R1.release() must NOT drop R2's lock. Fails on the old
blind-unlink path (ENOENT), passes now. Also adds explicit single-agent
restart-path guard coverage (review should-fix).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

packages/mosaic/src/commands/fleet.spec.ts

@@ -4,6 +4,7 @@ import { dirname, join, resolve } from 'node:path';
 import { Command } from 'commander';
 import { afterEach, describe, expect, it, vi } from 'vitest';
 import {
+  acquireRestartLock,
   addAgentToRoster,
   buildAgentSendCommand,
   buildAgentWatchAttachCommand,
@@ -818,6 +819,93 @@ describe('fleet command construction', () => {
     }
   });
 
+  it('guards the single-agent restart path behind the in-flight restart lock', async () => {
+    const home = await tempDir();
+    const rosterPath = join(home, 'fleet', 'roster.yaml');
+    await mkdir(join(home, 'fleet'), { recursive: true });
+    await writeFile(
+      rosterPath,
+      ['version: 1', 'transport: tmux', 'agents:', '  - name: coder0', '    runtime: codex'].join(
+        '\n',
+      ),
+    );
+
+    // A full restart is mid-flight (lock held); a single-agent restart re-enters.
+    const lockPath = restartLockPath(home);
+    await mkdir(dirname(lockPath), { recursive: true });
+    await writeFile(lockPath, `4242\n${Date.now()}\n`);
+
+    const events: string[] = [];
+    const runner: CommandRunner = async (command, args) => {
+      events.push(`run:${args[args.length - 1]}`);
+      return { stdout: '', stderr: '', exitCode: 0 };
+    };
+    let sleeps = 0;
+    const sleepFn: SleepFn = async () => {
+      sleeps += 1;
+      events.push(`sleep:${sleeps}`);
+      await rm(lockPath, { force: true });
+    };
+
+    const program = new Command();
+    program.exitOverride();
+    registerFleetCommand(program, { runner, sleepFn, mosaicHome: home });
+
+    try {
+      await program.parseAsync(['node', 'mosaic', 'fleet', 'restart', 'coder0']);
+
+      // The single-agent restart waits for the in-flight restart before acting.
+      expect(sleeps).toBeGreaterThan(0);
+      const firstSleep = events.findIndex((e) => e.startsWith('sleep:'));
+      const firstRun = events.findIndex((e) => e.startsWith('run:'));
+      expect(firstSleep).toBeGreaterThanOrEqual(0);
+      expect(firstRun).toBeGreaterThan(firstSleep);
+      // Only the named agent is restarted; the holder is untouched.
+      expect(events).toContain('run:mosaic-agent@coder0.service');
+      expect(events).not.toContain('run:mosaic-tmux-holder.service');
+    } finally {
+      await rm(home, { recursive: true, force: true });
+    }
+  });
+
+  it('does not let a timed-out owner drop a lock another restart broke and re-owned', async () => {
+    const home = await tempDir();
+    const runDir = join(home, 'fleet', 'run');
+    await mkdir(runDir, { recursive: true });
+    const lockPath = restartLockPath(home);
+    const tokenOf = async (): Promise<string> => {
+      const raw = await readFile(lockPath, 'utf8');
+      return raw.split('\n')[2]?.trim() ?? '';
+    };
+    const sleepFn = vi.fn<SleepFn>(async () => {});
+
+    // R1 acquires the lock and begins a restart that then hangs.
+    const r1 = await acquireRestartLock(home, sleepFn);
+    const tokenR1 = await tokenOf();
+    expect(tokenR1).not.toBe('');
+
+    // The hung R1 leaves a stale lock: rewrite its timestamp into the past while
+    // preserving R1's token — exactly the on-disk state a stuck owner leaves.
+    await writeFile(lockPath, `4242\n${Date.now() - RESTART_LOCK_STALE_MS - 1_000}\n${tokenR1}\n`);
+
+    // R2 re-enters, sees the stale lock, and atomically takes ownership.
+    const r2 = await acquireRestartLock(home, sleepFn);
+    const tokenR2 = await tokenOf();
+    expect(tokenR2).not.toBe(tokenR1);
+    expect(sleepFn).not.toHaveBeenCalled();
+
+    // R1 finally finishes and releases. It must NOT delete R2's lock — otherwise
+    // a third restart (R3) could acquire and interleave with R2 still running.
+    await r1.release();
+    expect(await tokenOf()).toBe(tokenR2);
+
+    // R2 releases cleanly and the lock is gone.
+    await r2.release();
+    await expect(readFile(lockPath, 'utf8')).rejects.toMatchObject({ code: 'ENOENT' });
+
+    await rm(home, { recursive: true, force: true });
+  });
+
   it('attempts every agent and the holder during fleet stop even when an agent stop fails', async () => {
     const home = await tempDir();
     const rosterPath = join(home, 'fleet', 'roster.yaml');