Hello, I found an issue in mod_fcgid; when a lot of pending requests are piled up due to fcgi programs running too slow, the fcgi processes can fail to be collected by mod_fcgid and become zombies. And if the number of such zombies gets higher, the httpd itself terminates.
http://httpd.apache.org/mod_fcgid/mod/mod_fcgid.html#examples I tried the example programs in the page above, both PHP and Perl scripts, with sleep(3) inserted to make them slow. I could reproduce the problem by requesting those programs 800 times per minute (with 20 threads, each requesting 40 times per minute). Zombies were still not collected even if I set "FcgidZombieScanInterval 3" explicitly. Versions which reproduces the issue: Apache 2.4.16, mod_fcgid 2.3.9, CentOS6.6 Apache 2.4.10, mod_fcgid 2.3.7, CentOS6.6 Versions which don't reproduce the issue: Apache 2.2.25, mod_fcgid 2.3.7, CentOS6.6 My configuration for mod_fcgid: FcgidMaxProcessesPerClass 200 FcgidIOTimeout 300 FcgidIdleTimeout 180 FcgidIPCDir /home/www/var/apache24/fcgidsock FcgidProcessTableFile /home/www/var/apache24/fcgid_shm # 128MiB. FcgidMaxRequestLen 134217728 For now I am periodically graceful-restarting the httpd to collect zombies. Please let me know if there is a better way to avoid this problem. Thanks, --------------------------------------------------------------------- To unsubscribe, e-mail: users-unsubscr...@httpd.apache.org For additional commands, e-mail: users-h...@httpd.apache.org
