It's my understanding the problem has to do with the file / memory management code within VxWorks.
Here's another article on the subject: Thinking on Mars: The Brains of NASA's Red Planet Rovers
From what I'm hearing, the problem is not the OS, which is doing exactly what it's supposed to do.
The problem is the engineers that set it up. Apparently when the flash device was formated, it was formated with the expectation of only holding a certain number of files. When they exceeded that number, the file handler in the OS refused to access any more files. IOW, it operated as designed.
The fix that they are setting up for this weekend is basically a format and reinstall. Once the filesystem is set up properly, the rover shouldn't have this kind of problem again.
A lot of Unix engineers are familiar with this kind of problem. Anyone who has set up a system that grew much faster than expected has probably run into the same issue.