Why does NCSA HTTPd core dump when I try to restart it?
NCSA HTTPd 1.4.x did some things in its restart code which are generally considered a bad thing to do. Most platforms allow it anyways, but under certain systems (most notably SunOS 4.1.3 and Solaris 2.x) using gcc with -O2 optimization, the restart code causes a SIGBUS. NCSA HTTPd 1.4.2 worked “around” this problem by changing the default compile to -O optimization. NCSA HTTPd 1.5.x had a complete change in the way in which restarts were handled to work in a fashion which didn’t bend the rules, and there should be no more problems with the restart. If you use the -d or -f command line options and attempt to restart, the restart may fail if the paths you used were not full paths. There is an attempt at a work around in 1.5.x. Some people have complained about the lack of information when a restart fails. Because of the nature of the daemon, it is generally not possible to provide the information to the terminal from which the HUP was sent (and is not always wanted, either). Instead, we
