** Description changed: [Problem] Apache OCF resource agent doesn't work when SSL module is enabled [Impact] This bug impacts all users that want to set up highly available Apache service with SSL, using Pacmaker or RedHat Cluster Suite tools. [Development Fix] Development branch isn't fixed yet either. Proposed solution is an upstream patch for this problem. [Stable Fix] Attached patch - resource-agents-apache.debdiff - is an oneliner, authored by upstream and part of the new upstream versions. [Test Case] To reproduce the problem, follow these steps on a clean 12.04 install: 1. apt-get install pacemaker apache2 2. service apache2 stop ; a2enmod ssl 3. update-rc.d -f apache2 remove 4. sed -i 's/START=no/START=yes/' /etc/default/corosync ; service corosync start - 5. sleep 90 ; crm configure primitive apache2 ocf:heartbeat:apache params \ + 5. sleep 90 ; crm configure property stonith-enabled=false ; crm configure property no-quorum-policy=ignore + 6. crm configure primitive apache2 ocf:heartbeat:apache params \ configfile="/etc/apache2/apache2.conf" - 6. sleep 21 ; crm status | grep 'unknown exec error' - 7. echo $? + 7. sleep 21 ; crm status | grep 'unknown exec error' + 8. echo $? Broken Behavior: last step (echo $?) will return 0, apache will not be running Fixed Behavior: last step (echo $?) will return 1, apache will be running [Regression Potential] There are no known regressions. This patch impacts only this resource agent; and since it's not working now it would hardly introduce any worse side effects. [Original Report] Summary: the apache heartbeat RA that ships with both the universe package and PPA does not include a line to guarantee the existence of /var/run/apache2, which is usually done by apache2ctl. Since crm calls the RA, which in turn directly calls the apache2 daemon, this bit of housekeeping never gets done. I originally raised this in https://github.com/ClusterLabs/resource- agents/issues/70 but then realized this PPA isn't in sync with what they're doing there. (I'm guessing this package is downstream from that?) The line in question, at that remote repo, is https://github.com/ClusterLabs/resource- agents/blob/master/heartbeat/apache#L177 I discovered this bug in the course of setting up a Linux HA cluster locally. The line I'm proposing be added is also absent from the universe cluster-agents package that ships with lucid. (See http://bazaar.launchpad.net/~ubuntu-branches/ubuntu/lucid/cluster-agents /lucid-proposed/view/head:/heartbeat/apache#L399) I'd submit a patch but I am not much up to speed on bzr... it's one line though so not too complex.
-- You received this bug notification because you are a member of Ubuntu High Availability Team, which is subscribed to cluster-agents in Ubuntu. https://bugs.launchpad.net/bugs/980180 Title: Apache RA doesn't create /var/run/apache2 [SRU] Status in “cluster-agents” package in Ubuntu: Confirmed Status in “resource-agents” package in Ubuntu: Confirmed Status in “cluster-agents” source package in Precise: Confirmed Status in “resource-agents” source package in Precise: Fix Committed Status in “cluster-agents” source package in Quantal: New Status in “resource-agents” source package in Quantal: Incomplete Bug description: [Problem] Apache OCF resource agent doesn't work when SSL module is enabled [Impact] This bug impacts all users that want to set up highly available Apache service with SSL, using Pacmaker or RedHat Cluster Suite tools. [Development Fix] Development branch isn't fixed yet either. Proposed solution is an upstream patch for this problem. [Stable Fix] Attached patch - resource-agents-apache.debdiff - is an oneliner, authored by upstream and part of the new upstream versions. [Test Case] To reproduce the problem, follow these steps on a clean 12.04 install: 1. apt-get install pacemaker apache2 2. service apache2 stop ; a2enmod ssl 3. update-rc.d -f apache2 remove 4. sed -i 's/START=no/START=yes/' /etc/default/corosync ; service corosync start 5. sleep 90 ; crm configure property stonith-enabled=false ; crm configure property no-quorum-policy=ignore 6. crm configure primitive apache2 ocf:heartbeat:apache params \ configfile="/etc/apache2/apache2.conf" 7. sleep 21 ; crm status | grep 'unknown exec error' 8. echo $? Broken Behavior: last step (echo $?) will return 0, apache will not be running Fixed Behavior: last step (echo $?) will return 1, apache will be running [Regression Potential] There are no known regressions. This patch impacts only this resource agent; and since it's not working now it would hardly introduce any worse side effects. [Original Report] Summary: the apache heartbeat RA that ships with both the universe package and PPA does not include a line to guarantee the existence of /var/run/apache2, which is usually done by apache2ctl. Since crm calls the RA, which in turn directly calls the apache2 daemon, this bit of housekeeping never gets done. I originally raised this in https://github.com/ClusterLabs/resource- agents/issues/70 but then realized this PPA isn't in sync with what they're doing there. (I'm guessing this package is downstream from that?) The line in question, at that remote repo, is https://github.com/ClusterLabs/resource- agents/blob/master/heartbeat/apache#L177 I discovered this bug in the course of setting up a Linux HA cluster locally. The line I'm proposing be added is also absent from the universe cluster-agents package that ships with lucid. (See http://bazaar.launchpad.net/~ubuntu-branches/ubuntu/lucid/cluster- agents/lucid-proposed/view/head:/heartbeat/apache#L399) I'd submit a patch but I am not much up to speed on bzr... it's one line though so not too complex. To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/cluster-agents/+bug/980180/+subscriptions _______________________________________________ Mailing list: https://launchpad.net/~ubuntu-ha Post to : [email protected] Unsubscribe : https://launchpad.net/~ubuntu-ha More help : https://help.launchpad.net/ListHelp
