Okay, okay, "me bad," I know that I am doing something where I easily deserve to be getting strange and cryptic error-messages. What I have not figured out in this case is how to go about diagnosing ... the utter and complete lack of any messages at all!

I have built an Apache 1.3 server from-source, and apachectl checkconfig runs clean. But when I try apachectl start I get the message "httpd could not be started," and there is an error-log file, and that error-log is completely empty. (Yes, this error-log is known to be brand-new, newly-created... known to be the log-file, and known to be empty.)

gdb tells me that the httpd program has ended with return-code "1" and that it did so completely normally. No segfault, no core dump, and no messages anywhere.

How might one diagnose this problem? Remember, I do have the source and I do, basically, know what I am doing. I'm sure that it basically comes down to an exit(1); but with no preceding sprintf()s that would have written to the logfile.

I guess my first thought is... does Apache have some "really great tell-you-what-I'm-doing-now" logging that I might turn on, so that it can give me a chatty play-by-play leading up to the problem?

Remember: I'm doing a custom build, from scratch, in a "nice out-of-the-way place," just for testing. I can start up the 'standard' Apache server on the box at any time, so I'm not dead in any water.

According to this, you should check the Errorlog Directive. If your config shows the /path/to/errorlog, try changing it to point to syslog instead. It might provide more information.