The instance method for several types (e.g. keystone_tenant) using the openstack provider only caches the instance on its first execution, which does not account for potential changes to the tenant. The following snippet should create a tenant and a Neutron network associated to that tenant:
keystone_tenant { 'test1':
ensure => present,
enabled => true,
description => 'default tenant',
}
neutron_network { 'test1-network':
ensure => present,
router_external => true,
tenant_name => 'test1',
}
With the old keystone provider, test1-network was correctly associated to the tenant, while it is not with the new. Looking at the way puppet-neutron gets the tenant-id from the tenant name:
def self.get_tenant_id(catalog, name)
instance_type = 'keystone_tenant'
instance = catalog.resource("#{instance_type.capitalize!}[#{name}]")
if ! instance
instance = Puppet::Type.type(instance_type).instances.find do |i|
i.provider.name == name
end
end
if instance
return instance.provider.id
else
fail("Unable to find #{instance_type} for name #{name}")
end
end
Since Keystone_tenant[test1] is already in the catalog, it is simply getting the id. This is broken by the new implementation. It could also be fixed directly in the puppet-neutron code, but there is a chance that other modules rely on this.
Fix proposed to branch: master
Review: https:/