Im putting it automation blocker, as this is blocking create vm flow

In 3.5, the db schema script just sets the management network name to be ovirtmgmt:

select fn_db_add_config_value('ManagementNetwork','ovirtmgmt','general');

and then a module from the downstream-only package rhevm-setup-plugins renames it:

2015-03-02 23:05:16 DEBUG otopi.context context._executeMethod:138 Stage misc METHOD otopi.plugins.ovirt_engine_setup.rhevm_setup_plugins.distro-rhel.config.Plugin._network
2015-03-02 23:05:16 DEBUG otopi.ovirt_engine_setup.database database.execute:167 Database: 'None', Statement: '
                select RenameManagementNetwork(%(name)s)
            ', args: {'name': 'rhevm'}
2015-03-02 23:05:16 DEBUG otopi.ovirt_engine_setup.database database.execute:217 Result: [{'renamemanagementnetwork': ''}]

(Sadly, the log does not show if it worked or not etc., but I guess it did).

In 3.6, we have quite many other places dealing with this name:

INSERT INTO network (id, name, description, type, addr, subnet, gateway, vlan_id, stp, storage_pool_id, mtu, vm_network, provider_network_provider_id, provider_network_external_id, free_text_comment) VALUES ('00000000-0000-0000-0000-000000000009', 'ovirtmgmt', 'Management Network', NULL, NULL, NULL, NULL, NULL, false, '00000001-0001-0001-0001-0000000001f5', NULL, true, NULL, NULL, NULL);

INSERT INTO vnic_profiles (id, name, network_id, port_mirroring, custom_properties, description, _create_date, _update_date, network_qos_id) VALUES ('0000000a-000a-000a-000a-000000000398', 'ovirtmgmt', '00000000-0000-0000-0000-000000000009', false, NULL, NULL, '2015-02-02 13:47:16.002849+02', NULL, NULL);

select fn_db_add_config_value('ManagementNetwork','ovirtmgmt','general');

update network_cluster
set management = true
where network_id in (select id
                     from network
                     where name=(select COALESCE((select option_value
                                                  from vdc_options
                                                  where option_name='ManagementNetwork'
                                                  and version='general'),
                                                 'ovirtmgmt')));

Some options I can think of:

1. Just update the rhevm-setup-plugins package to updates these places too

2. Make all the other callers in dbscripts parametric, and pass them somehow the correct name

3. Split the schema script to two parts - one sets the name, the other includes all the code that uses this name, and use the name in the db (and not hard coded). Then make the rhevm-setup-plugins plugin run in between the two.

(1.) is simplest, but is not that scalable - it will break if more such places are added in the future.

Sandro - what do you think?

There is, btw, another option - to just give up on branding the network name and leave it ovirtmgmt in downstream. We'll then have to deal with upgrades, might not be trivial either.

I suggest to go with 1 for allowing automation to continue to test the rest of the code and ask PM and infra about comment #3 and options 2 and 3.

It seems to me like it should have worked - RenameManagementNetwork does change all places. Looking further.

Seems like a bug in [1], setting the version to 'genenral' instead of 'general'.

Eli will fix by adding a new update file.

[1] https://gerrit.ovirt.org/32810

Hi Nelly,

I still encounter this issue in Red Hat Enterprise Virtualization Manager Version: 3.6.0-0.18.el6(rhevm-backend-3.6.0-0.18.el6.noarch.rpm). 

Please let me know which rhevm-backend version includes this fix.

Thanks!

Hui Wang

Eli/Simone, can you answer Hui's question in comment 8?

The issue Didi was talking about has been fixed here:
https://gerrit.ovirt.org/#/c/39885/
a few month ago.

On 3.6 the management network is called by default ovirtmgmt also downstream and the user can also change it to a different value. Please see:
https://bugzilla.redhat.com/show_bug.cgi?id=1231799#c5