Red Hat Bugzilla – Bug 1212158
notifications are not being sent by OpenStack services because notification_driver is not defined in their config files
Last modified: 2016-04-07 17:37:05 EDT
Description of problem: Notifications (https://wiki.openstack.org/wiki/SystemUsageData) sent to the notifications.info queue which are consumed by Ceilometer or ManageIQ are not being sent by OpenStack services. The services configuration file are leaving the notification_driver undefined. I believe this is a problem on both the undercloud and overcloud. For nova and heat, I had to manually change the config files and restart services to start getting notifications. /etc/heat/heat.conf notification_driver=glance.openstack.common.notifier.rpc_notifier notification_topics=notifications /etc/nova/nova.conf notification_driver=messaging notification_topics=notifications Ideally, puppet would set the appropriate notification_driver and any other settings for all services. There might be additional work needed upstream because not every puppet module allows you to set the notification driver. For example nova does, but heat and ironic does not. Version-Release number of selected component (if applicable): instack-0.0.6.4-g57c723a.el7.centos.noarch instack-undercloud-2.0.0-dev1606.el7.centos.noarch openstack-tripleo-puppet-elements-0.0.0-post45.el7.centos.noarch How reproducible: always Steps to Reproduce: 1. Deploy instack undercloud 2. Observe notification.info queue, from RabbitMQ management UI. No messages flow into the queue when you deploy the overcloud. Actual results: Notifications are not sent Expected results: Notifications are sent Additional info:
Correction. The heat configuration file should have the following lines. /etc/heat/heat.conf notification_driver=messaging notification_topics=notifications
Urgent for OSP8, not a blocker (removing a flag).
I believe "notification_driver=messaging" needs to be added to more than just nova and heat configuration file. When adding "notification_driver=messaging" only in Nova, I am not getting all instances events in Cloudforms. As an example, Cloudforms will get messages about an instance power off/on. But Cloudforms will not see a new instances booting up. I have added "notification_driver=messaging" to all configuration files: nova.conf, neutron.conf, heat.conf, keystone.conf, glance.conf, etc... And now Cloudforms can see all messages, including newly created instanced directly from Openstack. Somebody should verify exactly what list of messages are required by Cloudforms or make sure all services have this setting (unless this is causing an issue).
The related patch adds the relevant configuration for Keystone, Glance, Nova, Heat, and Neutron, so I think we're okay here!
See below output all relevant components configured with messaging /etc/glance/glance-api.conf:508:notification_driver =messaging /etc/heat/heat.conf:401:notification_driver =messaging /etc/keystone/keystone.conf:238:notification_driver =messaging /etc/neutron/neutron.conf:590:notification_driver=messaging /etc/neutron/neutron.conf:1123:notification_drivers = message_queue /etc/nova/nova.conf:1485:notification_driver =messaging /etc/puppet/hieradata/puppet-stack-config.yaml:223:keystone::notification_driver: messaging /etc/puppet/hieradata/puppet-stack-config.yaml:316:nova::notification_driver: messaging
*** Bug 1314616 has been marked as a duplicate of this bug. ***
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://rhn.redhat.com/errata/RHEA-2016-0604.html