If I remember correctly, two weeks before this failure the Israelis had informed the US Army about the clocks losing accuracy (20%) after 8 hours and were rebooting their systems regularly to offset the skew. The software developers had estimated that a 50% loss in accuracy would mean that the system could not track the missile effectively. After analyzing the data Israel provided the Army determined 50% loss would occur at approximately 20 hours… the one at Dhahran had been running for over 100 hours.
I remember the same discussion. This seems to be more thorough coverage: