Hello,
Four days ago, Apache suddenly started to not kill some of it's older child processes. Since the processes are unresponsive to new reqeusts, it then spawns more processes, leading to more than 150 httpd processes running. Mysql eventually ran out of connections, causing my clients to start emailing about their sites being down.
I have since setup a script to scan for the number of Apache processes, and when it hits 50, restart Apache, but this is hardly a solution.
If anyone has any suggestions for why Apache would not kill off certain child processes, please let me know. The error_log shows nothing of importance, even when I set it to debug mode. The only notable entries are ones mentioning that the other user closed the connection abruptly:
(104)Connection reset by peer: core_output_filter: writing data to the network
(32)Broken pipe: core_output_filter: writing data to the network
When I restart apache, I get dozens of these (well as many processes halted without releasing memory or accepting new connections):
child process 19979 still did not exit, sending a SIGTERM
I don't get any corresponding PHP errors. Before this happened I updated spamassassin, which should have nothing to do with Apache. I have since updated PHP to version 5.04, and Apache to the latest stable release for rhel3 (2.0.46-61) in ART's repository in an attempt to fix the problem. MySQL is 4.1.21
Everything is working great aside from this problem. If anyone has any suggestions, I would appreciate the help!
Jordan
Four days ago, Apache suddenly started to not kill some of it's older child processes. Since the processes are unresponsive to new reqeusts, it then spawns more processes, leading to more than 150 httpd processes running. Mysql eventually ran out of connections, causing my clients to start emailing about their sites being down.
I have since setup a script to scan for the number of Apache processes, and when it hits 50, restart Apache, but this is hardly a solution.
If anyone has any suggestions for why Apache would not kill off certain child processes, please let me know. The error_log shows nothing of importance, even when I set it to debug mode. The only notable entries are ones mentioning that the other user closed the connection abruptly:
(104)Connection reset by peer: core_output_filter: writing data to the network
(32)Broken pipe: core_output_filter: writing data to the network
When I restart apache, I get dozens of these (well as many processes halted without releasing memory or accepting new connections):
child process 19979 still did not exit, sending a SIGTERM
I don't get any corresponding PHP errors. Before this happened I updated spamassassin, which should have nothing to do with Apache. I have since updated PHP to version 5.04, and Apache to the latest stable release for rhel3 (2.0.46-61) in ART's repository in an attempt to fix the problem. MySQL is 4.1.21
Everything is working great aside from this problem. If anyone has any suggestions, I would appreciate the help!
Jordan