Puppet keystone cannot restart HTTPD managed by Pacemaker

[Adam Young]

when calling puppet from tripleo, tried to call:

 class { '::keystone::wsgi::apache':
    ssl => false,
  }

but that seems to trigger puppet into restarting httpd, and that service file won't exist anymore and thus will fail. That is because httpd will be managed by pacemaker instead. That is actually done here:

https://github.com/openstack/tripleo-heat-templates/blob/master/puppet/manifests/overcloud_controller_pacemaker.pp#L1501

this seems to be the problem https://github.com/openstack/puppet-keystone/blob/master/manifests/wsgi/apache.pp#L168 as using the ~> triggers a refresh of apache, which puppet tries to do via systemd

[Richard Megginson (rmeggins)]

Instead of::

    Package['keystone'] ~> Service['httpd']

we should use::

    Package['keystone'] ~> Class['::apache::service']

See http://eavesdrop.openstack.org/irclogs/%23tripleo/%23tripleo.2015-12-10.log.html#t2015-12-10T17:38:16 and later for the discussion about what this issue is and how it might be solved.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/276285

[OpenStack Infra (hudson-openstack)]

Change abandoned by ayoung (<email address hidden>) on branch: master
Review: https://review.openstack.org/276285
Reason: No longer needed. If we do, need something like this, we'll start from the top

[OpenStack Infra (hudson-openstack)]

Change abandoned by ayoung (<email address hidden>) on branch: master
Review: https://review.openstack.org/276285

[Cédric Jeanneret deactivated (cjeanneret-c2c-deactivated)]

Dear Stackers,

This bug still exists in Pike, at least when we use TripleO.

For example, I had to update a credential in our keystone domain, and had to manually restart httpd service on all our controllers.

Would be great if this issue could be squashed properly :).

Thank you!

Cheers,

C.

[Takashi Kajinami (kajinamit)]

I'll close this bug as won't fix. If any different treatment is required then you can use manage_service=false and implement your own logic to trigger restart.