I'm running "Bitcoin version 0.3.24-beta" on a couple Linux servers. About once a week, the daemon crashes and I have to restart it. In every instance, the last entry in debug.log is something like the following:
Added time data, samples 28503, offset +0 (+0 minutes)
-747598944 -747551416 -747428248 ... -201976 -1940
The daemon seems to crash while it's printing the entire list of time samples. In the above quote, the median offset was "+0" but "-1940" was the last sample it printed.
I haven't noticed any other patterns. What else can I do to help narrow down the problem?