Skip to content

Cloudflare Workers Resource Limits & OpenNext Configuration

Overview

Cloudflare Workers don't have a "reboot" mechanism when hitting resource limits. When a Worker exceeds CPU or memory limits, Cloudflare terminates execution and returns Error 1102. Each new request starts a fresh Worker instance automatically.

Solutions

1. Increase CPU Limits

Add the limits configuration to wrangler files. Paid Workers can use up to 5 minutes (300,000ms) of CPU time.

For apps/frontend/wrangler.jsonc: 

{
  "main": ".open-next/worker.js",
  "limits": {
    "cpu_ms": 60000
  },
  // ... rest of config
}

For apps/producer/wrangler.jsonc: 

{
  "name": "piquetickets-producer",
  "main": ".open-next/worker.js",
  "limits": {
    "cpu_ms": 60000
  },
  // ... rest of config
}

Start with 60 seconds (60000ms) and adjust based on actual needs.

2. Update Compatibility Date

Update compatibility_date to 2025-05-05 or later for better API support:

"compatibility_date": "2025-05-05"

3. Multi-Worker Setup (Advanced)

For better resource isolation, OpenNext supports splitting your app into separate middleware and server workers. This reduces the memory footprint of each worker.

Benefits: - Reduced memory footprint per worker - Improved cold start performance - Better handling of ISR/SSG requests

Limitations: - Cannot be used with preview URLs - Cannot use standard @opennextjs/cloudflare deploy command - Requires manual deployment with Wrangler CLI

See: https://opennext.js.org/cloudflare/howtos/multi-worker

4. Analyze and Optimize Bundle Size

cd apps/frontend && npx opennextjs-cloudflare build
# Then analyze .open-next/server-functions/default/handler.mjs.meta.json

Use the ESBuild Bundle Analyzer to identify large dependencies.

Size Limits: - Free Plan: 3 MiB compressed - Paid Plan: 10 MiB compressed

5. Monitoring

Use wrangler tail to debug in real-time: 

wrangler tail --env production

Sentry is already configured for error capture.

Common Causes of Error 1102

  1. Heavy computation or long-running tasks
  2. Tight loops or unbounded recursive calls
  3. Hanging Promises that stall execution
  4. Large bundle sizes causing slow cold starts

Best Practices

  • Break down lengthy tasks into smaller, asynchronous chunks
  • Avoid blocking patterns
  • Stream data processing where possible
  • Ensure all Promises resolve or reject appropriately
  • Move heavy computation to backend services or scheduled jobs

References