node --report-on-process-timeout writes truncated report when a Worker is blocked

[trivikr]

Version

main

Platform

All

Subsystem

report

What steps will reproduce the bug?

import { Worker } from 'node:worker_threads';

new Worker(`require('node:child_process').execFileSync('sleep', ['30']);`, { eval: true });

Run with

$ node --process-timeout=1s --report-on-process-timeout repro.js

How often does it reproduce? Is there a required condition?

Always

What is the expected behavior? Why is that the expected behavior?

A complete, valid JSON report is written. Blocked Workers could be left out or marked as unresponsive. A stuck Worker is exactly the case where you'd want the report. The forced-exit message should also start on its own line.

What do you see instead?

(node:2291000) Process timed out after 1s (--process-timeout). Exiting with code 124.
Main thread was not executing JavaScript.
Resources keeping the event loop alive:
    Worker (thread 1, name 'WorkerThread')

Writing Node.js report to file: report.20260925.212601.2291000.0.001.json(node:2291000) The process did not finish exiting within 5000ms after --process-timeout expired. Forcing exit.

$ node -e "JSON.parse(require('fs').readFileSync('report.20260925.212601.2291000.0.001.json'))"
SyntaxError: Expected property name or '}' in JSON at position 8192 (line 338 column 9)

Additional information

No response

[jasnell]

@trivikr ... i'm going to assume since you assigned this to yourself you're going to take a stab at fixing... I did a quick check and this is what I found:

The bug is in node_report.cc, not the watchdog. The report waits with no time limit for every Worker to respond to its interrupt. A Worker blocked in a native synchronous call never responds. Any main-thread report hangs forever in this case; --process-timeout just turns the hang into a truncated file. Specifically, node_report.cc hands and the process timeout grace period cuts it off.

[trivikr]

Yes. A proposed a 2-second time limit for worker threads in #66304