[Sorry, I posted with the wrong email account] > I'm not familiar with apache internals, nor LDAP, but I can try to > help begin diagnosis. Thanks. > Could you provide a strace of the process when it is in that state? > (strace -p <pid>)? Maybe also a strace of the process beginning > before it is in that state, and ending sometime after it has gone into > that state. Here it is. I'm not familiar with bug reporting w/ Debian: I've gzipped the log, to keep this email small. If I should have sent it as plain ASCII, please let me know. Here's what I did to obtain this trace: /etc/init.d/apache2 start # use prefork ps ax | grep apache2 # to get the PIDs -> 23224 to 23229 sudo strace -p 23224 ... -p 23229 2>&1 | tee ~/apache2.strace I then run a small script that use wget to stress the webserver from another console, until the script got deadlocked (as the server does not answer the request when it consumes 100% CPU), then broke the strace execution (Ctrl+C) top # to get the PID of the process eating up the CPU time grep [pid] ~/apache2.strace | gzip -c > apache2.strace.23229.gz I then restarted strace -p 23229, but strace does not show any other trace for this process. I'll come back later in another email with the GDB trace, if I manage to get it. Thanks for your support, Emmanuel
Attachment:
apache2.strace.23229.gz
Description: Binary data