If properly installed, at least on CentOS, Asterisk will be restarted by safe_asterisk, if it crashes. This does not apply if it deadlocks (continues to run but becomes unresponsive), but I think your problem description would have hinted at that if it had been the case.
From /var/log/asterisk/full, you are looking for ERRORs and WARNINGs near the time of the crash, or failing that, anything in common between crashes. However, you really should be getting a backtrace from gdb. Google for “asterisk wiki backtrace”.
Chances are, though, that you don’t have a version built to support debugging, and in any case, it is very unlikely that you are relying on anything that was broken in your version and is fixed in 22.214.171.124, so you should update to the end of line 1.6.2 version and ensure it meets the debugging requirements from that Google search.