[Monitor]

It could be a software problem, too.

I worked on embedded systems for many years, though not for NASA. And one thing we always included was a watchdog timer.

Essentially, the watchdog timer is hardware that resets the computer every X seconds, or minutes, or hours (depending on the design), and if the software that's running on the computer doesn't occasionally reset the watchdog timer (because, perhaps it's caught in an infinite loop) then the watchdog timer will reset the computer. If everything is working perfectly, the software always resets the watchdog timer before the watchdog timer can reset the computer.

Considering the article specifically mentions times of 15 minutes and an hour, I suspect it's the watchdog timer that's resetting the computer.

Added to this all is that Spirit is running a multi-tasking, multi-threaded, realtime operating system. As such, the problem could be as simple as a thread having too high a priority (higher than the thread for resetting the watchdog timer), which is grabbing almost all the CPU cycles, thus slowing down communications and pre-empting the watchdog reset thread.

[muskogee]

If this is software, then competing OS's would have been nice here. One OS tries to complete the commands sent and prevent a toggle switch from being mechanically pushed, the other OS tries to mechanically push the toggle switch and complete the commands sent. If there is a weakness in the first OS, you get the second OS.

[blowfish]

>I worked on embedded systems for many years, though not for NASA. And one thing we always included was a watchdog timer.

Excellent point.