+++ This bug was initially created as a clone of Bug #894955 +++ Description of problem: OpenShift Enterprise ships a cartridge for JBoss EAP that requires a JBoss EAP subscription and the EAP channel ships an incompatible Apache init script. This will affect any customer that runs the stock httpd with a custom configuration and pidfile. When the OpenShift Broker service is run the following process is running: /usr/sbin/httpd -C Include /var/www/openshift/broker/httpd/broker.conf -f /var/www/openshift/broker/httpd/httpd.conf The openshift-broker service script is based of the RHEL6 httpd service script. It creates a pidfile in /var/www/openshift/broker/httpd/run/httpd.pid and uses that pid to only kill the processes it started. The JBoss httpd service script does not pass the "-p {pidfile}" flag to killproc and therefore it kills everything that matches /usr/sbin/httpd. This will stop both the openshift-console and openshift-broker services. Version-Release number of selected component (if applicable): httpd-2.2.22-14.ep6.el6 (EAP) Additonal Info: This exact same problem exist for EWS's httpd: httpd-2.2.17-15.4.ep5.el6 --- Additional comment from Brenton Leanhardt on 2013-01-14 01:53:42 EST --- Created attachment 678031 [details] use the pidfile like rhel6 This patch makes the JBoss httpd service script almost exactly like the RHEL6 service script. The only modification I made was to re-add the check13 function to maintain that functionality if that is required for JBoss. I believe there were tabs in the original JBoss httpd service script which makes it look like the check13 function changed. That is not the case. --- Additional comment from Brenton Leanhardt on 2013-01-14 01:56:52 EST --- This is a little more of an edge case for OpenShift than I originally thought. I'm lowering the severity to medium. --- Additional comment from Brenton Leanhardt on 2013-01-15 00:22:47 EST --- After some more investigation was performed on this by the OpenShift we realized this affects more than just the OpenShift web services but also the end user applications that make use of httpd (the vast majority of applications). Whenever the JBoss httpd is restarted all the user applications will be stopped. --- Additional comment from Brenton Leanhardt on 2013-01-20 22:30:12 EST --- *** Bug 893884 has been marked as a duplicate of this bug. *** --- Additional comment from Brenton Leanhardt on 2013-02-13 17:48:32 EST --- We're writing a KBase article to help users work around this issue: https://access.redhat.com/knowledge/articles/311023 --- Additional comment from Jimmy Wilson on 2013-04-02 14:54:34 EDT --- Fixing version to 6.0.1 since that's where the problem exists. I'm proposing this for 6.1.0, if we can squeeze it in. @Fernando, I think that's your call.
I cloned this bug for EWS as the same problem exists. I'm proposing for 2.0.1.
Yeah, we'll fix this for 2.0.1 :-)
Issue too complex to automate, will be tested manually
fixed in branches, after 2.0.1 ER1 release this could be tested.
Added DocText. @Wei Nan Li, can you please review the Doc Text content?
VERIFIED on httpd-2.2.22-17.ep6.el5 and httpd-2.2.22-17.ep6.el6
Apologies for the bz spam. An element in the doc text field is breaking the release notes tool, so I'm troubleshooting.