puppet-octavia/releasenotes/notes/set-common-configuration-options-across-services-91a9057d62b52de0.yaml
Tobias Urdin 0ea4e10dd3 Fix octavia::controller redeclaration
It's impossible to use this new class because you
cannot use any services combined with and specify
the parameters in octavia::controler since you get
redeclaration issues.

The class should only be included from octavia::worker
if it doesn't exist to support the old use case.

If you want to use this class it's required that you
define octavia::controller *before* octavia::worker
otherwise you'll end up with another redeclaration
issue since we include the controller unless its
defined before.

Depends-On: https://review.openstack.org/#/c/618915/
Change-Id: If5b4014d2129879199798d440ae604c03b01ab3a
2018-11-20 16:26:09 +01:00

35 lines
1.3 KiB
YAML

---
deprecations:
- |
The following octavia::worker parameters are deprecated and have been
moved to octavia::controller class. When you start using octavia::controller
make sure it's defined before octavia::worker.
* amp_flavor_id
* amp_image_tag
* amp_secgroup_list
* amp_boot_network_list
* loadbalancer_topology
* amphora_driver
* compute_driver
* network_driver
* amp_ssh_key_name
* enable_ssh_access
* timeout_member_connect
* timeout_member_data
* timeout_tcp_inspect
fixes:
- |
There are a couple of configuration options that need to be set not only
for the worker but also for other Octavia services. For example, on a
composable node deployment where the API runs on a separate node than
the rest of the Octavia services, the network driver was not being set
(hence defaulting to noop driver) while for the worker the driver was
allowed_address_pairs_driver. Another example is the database that was only
being set for the API service. Such configuration misalignment and
omissions lead to operate Octavia services and its resources.
upgrade:
- |
If you want to use the new octavia::controller class you must define it
before the octavia::worker class.