macOS TCP Networking Bug: A Ticking Time Bomb That Detonates After Exactly 49 Days

Researchers at Photon have discovered a critical bug in macOS networking: TCP connections stop working entirely after the system has been running for exactly 49.7 days — a classic integer overflow that turns any long-running Mac into an isolated island with no network access.

The Bug

The discovery reveals a fundamental flaw in macOS's TCP implementation:

• Trigger: System uptime reaches approximately 49.7 days (4,294,967,295 milliseconds — the maximum value of a 32-bit unsigned integer)
• Effect: All TCP networking stops functioning — no web, no email, no API calls, no SSH
• Cause: Integer overflow in a millisecond counter used by the TCP stack
• Severity: Critical for servers, critical infrastructure, and always-on Macs
• Fix: A simple reboot resolves the issue — until 49.7 days later

Why 49.7 Days?

This is a classic computer science problem:

• 2^32 milliseconds = 4,294,967,295 ms ≈ 49.7 days
• Integer overflow: When the counter wraps around, calculations produce incorrect values
• Historical precedent: Similar bugs have affected GPS systems (week rollover), Unix time (Y2038 problem), and various network protocols

Who Is Affected?

Any Mac running for extended periods:

• Mac Mini servers: Often run for months without rebooting
• macOS CI/CD runners: Build servers that stay online continuously
• Development machines: Developers who rarely restart their Macs
• Apple Silicon Macs: All architectures affected
• macOS Sequoia and earlier: Confirmed affected versions

The Impact

For different use cases, the impact varies:

• Personal Mac: Annoying but fixable with reboot
• Mac server: Complete service outage
• CI/CD pipeline: Failed builds, lost productivity
• IoT/edge Mac: Remote devices losing connectivity
• Production services: Potential data loss, SLA violations

The Research

The team at Photon discovered the bug while investigating network issues with their own long-running Mac servers:

• Discovery method: Systematic debugging of unexplained network failures
• Confirmation: Reproduced reliably on multiple Mac systems
• Disclosure: Published as a public service to the Mac community
• HN reception: 86 points, 44 comments within hours of posting

What Apple Should Do

The fix is straightforward:

1. Use 64-bit counters instead of 32-bit for time calculations
2. Add monitoring for uptime-based issues in the networking stack
3. Implement automatic recovery when overflow is detected

Broader Implications

This bug highlights persistent challenges in software engineering:

• Integer overflow remains a problem even in 2026, despite being a well-known class of bug
• Long-running systems expose bugs that testing doesn't catch
• Apple's QA process should include extended uptime testing
• Classic bugs like Y2K and GPS week rollover keep recurring in new forms