The default logrotate scripts (in part. /etc/logrotate.d/apache) do not function properly. The post-rotate script either does a: /usr/bin/killall -HUP httpd in RH 6.0 or a /usr/bin/killall -USR1 httpd in RH 6.1. Presumably these commands should cause apache to close and re-open its log files. Unfortunately, at least in the case of RH 6.1, it kills apache. My view is the post-rotate scripts should function like: /etc/rc.d/init.d/httpsd stop sleep 1 /etc/rc.d/init.d/httpsd start That way any customizations the user has made in the scripts which are designed to stop and start apache are executed. In any event, the 'USR1' signal is RH 6.1 should be changed to something that doesn't kill apache. Regards, Dr. Dre
This might well have been an Apache bug. Do you see this same behavior using the newer packages included with 6.2 (ftp://ftp.redhat.com/redhat/redhat-6.2/i386/)?
Created attachment 48428 [details] force to use pkill -HUP httpd
The logrotate script has done: /bin/kill -HUP `cat /var/run/httpd.pid 2>/dev/null` 2> /dev/null || true for a long while now.