I have a test that involves making nginx kill itself in the init_worker phase (a worker kills the master process and then kill itself, just in case), so my test look like that
--- wait: 1
--- error_log
[crit]
--- must_die: 0
but the error_log started to randomly fail at some point, I tracked this down to the fact that the error_log doesn't get the critical log lines, but they do appear in the t/servroot/error.log file, so I guess they are not written yet at that point. Setting a ridiculously high wait doesn't help as it seem to be ignored in that case.
I have a test that involves making nginx kill itself in the
init_workerphase (a worker kills the master process and then kill itself, just in case), so my test look like thatbut the
error_logstarted to randomly fail at some point, I tracked this down to the fact that theerror_logdoesn't get the critical log lines, but they do appear in thet/servroot/error.logfile, so I guess they are not written yet at that point. Setting a ridiculously high wait doesn't help as it seem to be ignored in that case.