||Posted: 2011-08-18 17:40:19
I Have been searching in vain for an answer to my problem - why is my Apache restarting ever day or two (or sometimes once a week). I have had LogLevel set at "devel" and the only thing I saw was - that apache was spawning new children 2 hrs before the crash occasionally but just 2-5 minutes before the crash - it would spawn them every few seconds and finally reach MaxClients setting and hang there forever until restarted.
The problem is:
We are getting
child pid 7459 exit signal Bus error (7)
child process 7518 still did not exit, sending a SIGTERM
Error messages just before the hangup. Examining what pages were being accessed just before it gave no results because sometimes those errors happen an hour before hangup. But Apache can raise from 10 requests to 300 requests in less than 3 minutes and that is it.
||Posted: 2011-08-23 03:48:20
Watch your MaxClients setting. Do you get Apache warnings about MaxClients setting reached often?
Memory and hardware problems would cause crashes and corruption in random
applications, not the same application in the same way over and over. So you
need to find out whether the error is in random or the same place.
You're going to need to get a core dump from apache (recompile your apache to have it); to find out what is
causing an apache worker to exit unexpectedly. Enable coredumps and the next
time the crash occurs, a core file will be dump, which can be examined with
gdb. This will give a stack backtrace, helping to identify the area of crash.
To read the core dump, this works (having in mind that that is where the core dump is sent):
gdb /usr/local/apache/bin/httpd /tmp/$core_dump_file
Some line will say that there was a problem and you'll know what it is.
You do Not have to register in order to post replies or topics. However, if you wish to receive an email alert when a new entry is posted on this topic, you will first have to Sign-Up
and confirm your registration.