A macOS kernel bug can cause OpenClaw to stop working after 49.7 days

What happened

It has been reported that a 32-bit unsigned integer overflow in Apple's XNU kernel causes the internal TCP timestamp clock to wrap after exactly 49 days, 17 hours, 2 minutes, and 47 seconds of continuous uptime. When that counter rolls over, TIME_WAIT entries stop expiring. The result? Ephemeral ports slowly run out and new TCP connections fail — though ICMP (ping) still replies. In short: the network looks alive, but everything that depends on TCP quietly dies. The only widely known remedy is a reboot.

How it was found

Photon, which runs a fleet of always-on Macs to monitor iMessage, allegedly discovered the problem when several machines stopped handling their monitoring traffic at the 49.7‑day mark. They reproduced the behavior on two machines and traced the root cause to a single comparison in the XNU kernel source that assumes the millisecond tick never wraps. It has been reported that this tick is stored in a uint32_t and overflows at 2^32 milliseconds — classic Pac‑Man kill‑screen territory for modern kernels.

Who's affected — and what to do

Anyone running long-lived macOS instances could be hit: monitoring fleets, lab machines, CI runners, kiosks, or any Mac that’s rarely rebooted. Want a practical fix? Reboot before the 49.7‑day mark or schedule periodic restarts. Tinkering with TCP internals on macOS isn't a realistic stopgap for most users, so rebooting is the pragmatic workaround until Apple issues a patch. It has been reported that Photon’s mitigation is to automate reboots; simple, ugly, effective.

This is one of those bugs that reads like tech archaeology — a tiny integer, an old assumption, and suddenly modern services are stranded. Who would have thought a 32‑bit millisecond counter would still be throwing curveballs in 2026? Keep an eye on your uptimes. Are your Macs silently counting down?

Sources: photon.codes, Hacker News