Red Hat Bugzilla – Bug 73220
httpd dies on receipt of a HUP signal
Last modified: 2008-05-01 11:38:03 EDT
From Bugzilla Helper: Description of problem: Apache dies when logs are rotated as part of logrotate.d (the weekly logrotate sends apache a HUP) Version-Release number of selected component (if applicable): [root@tornado linux-2.5]# rpm -q httpd httpd-2.0.40-4 (but this applies to previous versions of httpd-2 as well) How reproducible: Always Steps to Reproduce: 1. Wait for 4am Sunday morning after the weekly logrotate runs 2. Wake up and find that httpd is no longer running ;-) Expected Results: Apache should have rotated logs and not died Additional info: This can be manually triggered: [root@tornado root]# /etc/init.d/httpd status httpd (pid 17755 17754 17753 17752 17751 17748) is running... [root@tornado root]# kill -HUP `cat /var/run/httpd.pid` [root@tornado root]# /etc/init.d/httpd status httpd dead but pid file exists [root@tornado root]# [root@tornado root]# /etc/init.d/httpd restart Stopping httpd: [FAILED] Starting httpd: [ OK ] [root@tornado root]#
Are you using PHP or any other modules? If so, what version of the PHP package?
Yes, using PHP: php-4.2.2-11 php-imap-4.2.2-11 php-mysql-4.2.2-11 php-ldap-4.2.2-11 No other modules other than the default install AFAIK.
Can you still reproduce this if you disable PHP? e.g. # cd /etc/httpd/conf.d # mv php.conf php.conf.disabled # service httpd restart
You've obviously struck something like this before ;-) Yes it seems to be fine if PHP is disabled...
Yes, we've seen this before... what system are you testing on here - Limbo, Null, 7.3?
Yes, running Null (with all up2date updates applied).
*grumble* ok I can finally reproduce this. I still say it was something in httpd though Basically it'll trigger on the *SECOND* sighup to httpd not the first and only if libphp4.so is loaded. Oddly the segv is NOT in any php related code. Smells like a httpd bug, however, I've been advised that disabling zts support in php fixes this. ie service httpd start (ok) apachectl restart (ok) apachectl restart (fails) Humm,... compiled without zts,.. php segv's httpd segv's worse state of affairs. ... and this is going to have to be fixed in <10 hours??!? God, I'm so doomed. 8/ Phil =--=
Fixed in 4.2.2-8.0.5 though root cause not found
We're actually still tracking this down - what other modules do you have loaded? Results of "ls /etc/httpd/conf.d" could be useful data.
*** This bug has been marked as a duplicate of 66613 ***