> I'm investigating this ..

Problem fixed, most probably.
The glitch was the handling of bogus keep-alive ; if the 
remote server is closing the connection after the first 
response, given a valid "Connection: keep-alive", then 
httrack was trying to send the second request anyway, 
causing a SIGPIPE.
But I did not fully understand why the signal was sent 
despite of the signal() trap, weird.

Anyway this is a weird server behavious (IIS, as usual)

I could not explain the other SIGALARM problem, however - 
might be related to this issue.