"neutron net-list" : deprecated and wrong return code

[vercetty92360]

Hi,

First thx a lot for your work, I don't have yet a functionnal openstack PIKE cluster, but from what I saw everything is clear and simple.

Please let me share with you some info you should maybe consider, and one issue I have.

• I needed to modify line 101 of roles/database/tasks/main.yml

from:
"shell: rabbitmqctl cluster_status | grep running_nodes | grep {{ item }}"

to:
"shell: rabbitmqctl cluster_status | grep {{ item }}"

to have my deployment working. I was setting up a 3 nodes cluster, and when I do "rabbitmqctl cluster_status" from each node, the related hostname was not in the same line (it was in "running nodes" section, but on the line just after).

Example from my node openstack-1:

#rabbitmqctl cluster_status
Cluster status of node 'rabbit@openstack-1' ...
[{nodes,[{disc,['rabbit@openstack-1']},
{ram,['rabbit@openstack-3','rabbit@openstack-2']}]},
{running_nodes,['rabbit@openstack-3','rabbit@openstack-2',
'rabbit@openstack-1']},
{cluster_name,<<"rabbit@openstack-3.mydomain">>},

and from my node openstack-2:

#rabbitmqctl cluster_status
Cluster status of node 'rabbit@openstack-2' ...
[{nodes,[{disc,['rabbit@openstack-1']},
{ram,['rabbit@openstack-3','rabbit@openstack-2']}]},
{running_nodes,['rabbit@openstack-3','rabbit@openstack-1',
'rabbit@openstack-2']},
{cluster_name,<<"rabbit@openstack-3.mydomain">>},
{partitions,[]},
{alarms,[{'rabbit@openstack-3',[]},
{'rabbit@openstack-1',[]},
{'rabbit@openstack-2',[]}]}]

So I agree the change I did is really dirty but it was a simple workaround for me.

• now I have real issue with some others things: the neutron commands, which is now deprecated and never return anything. (return code is not 0)

Example:

grep "neutron " roles/neutron/tasks/controller.yml | grep "command"
- command: "openstack role add --project service --user neutron admin"
- command: "openstack service create --name neutron --description 'OpenStack Networking' network"
command: "neutron net-list"
- command: "neutron net-create --shared --provider:physical_network {{ provider_network }} --provider:network_type flat {{ provider_network }} --router:external True"
- command: "neutron subnet-create --name {{ provider_network }} --allocation-pool {{ provider_allocation_pool }} --dns-nameserver {{ provider_nameserver }} --gateway {{ provider_gateway }} {{ provider_network }} {{ provider_network_cidr }}"
- command: "neutron net-create {{ selfservice_network }}"
- command: "neutron subnet-create --name {{ selfservice_network }} --dns-nameserver {{ selfservice_nameserver }} --gateway {{ selfservice_gateway }} {{ selfservice_network }} {{ selfservice_network_cidr }}"
- command: "neutron router-create router"
- command: "neutron router-interface-add router {{ selfservice_network }}"
- command: "neutron router-gateway-set router {{ provider_network }}"

all the commands "neutron something" are not working anymore so impossible to run the playbook.

Do you have any change to commit about that?

Thx a lot in advance

[vercetty92360]

by the way, I tested to change "neutron net-list" with "openstack network list", but now the pb is that it try to reach the failover vip, which is mounted on a node where neutron is not listening on port 9696 (the ansible playbook always change the VIP's position to a node where neutron is not listening...)

Is it normal? do I miss something?

Thx a lot

[sergevs]

Hello,

Thanks for your feedback. I'll try to take a look on the issues by the end of this week.

all the commands "neutron something" are not working anymore
What's version of openstack you are trying to deploy ?

[vercetty92360]

Hi,

Sorry I was off^^

I'm trying to deploy Pike.

by the way, it seems that only replacing "neutron net-list" by "openstack network list" is enough, the other neutron commands was ok in my deployment.

i got another error on a second deployment when the playbook try to create the swift user. I didn't have this issue on the first deployment so I'm not sure what is the pb. If I understand the user database is shared between nodes, but it seems it try to create the user on each node.

TASK [swift : command] **************************************************************************************************************************************************************
changed: [openstack-2]
fatal: [openstack-1]: FAILED! => {"changed": true, "cmd": ["openstack", "user", "create", "--domain", "default", "--password", "soleil", "swift"], "delta": "0:00:02.220937", "end": "2018-03-09 14:52:40.492509", "msg": "non-zero return code", "rc": 1, "start": "2018-03-09 14:52:38.271572", "stderr": "Conflict occurred attempting to store user - Duplicate entry found with name swift at domain ID default. (HTTP 409) (Request-ID: req-b8aff5f9-28b5-43d8-a997-64f146af8fca)", "stderr_lines": ["Conflict occurred attempting to store user - Duplicate entry found with name swift at domain ID default. (HTTP 409) (Request-ID: req-b8aff5f9-28b5-43d8-a997-64f146af8fca)"], "stdout": "", "stdout_lines": []}
fatal: [openstack-3]: FAILED! => {"changed": true, "cmd": ["openstack", "user", "create", "--domain", "default", "--password", "soleil", "swift"], "delta": "0:00:02.239136", "end": "2018-03-09 14:52:40.545099", "msg": "non-zero return code", "rc": 1, "start": "2018-03-09 14:52:38.305963", "stderr": "Conflict occurred attempting to store user - Duplicate entry found with name swift at domain ID default. (HTTP 409) (Request-ID: req-6ccaa766-0625-498f-9cfe-3203aad23b49)", "stderr_lines": ["Conflict occurred attempting to store user - Duplicate entry found with name swift at domain ID default. (HTTP 409) (Request-ID: req-6ccaa766-0625-498f-9cfe-3203aad23b49)"], "stdout": "", "stdout_lines": []}

I take the opportunity to ask you 2 quick questions :

1. is it possible to use cinder with a shared storage with your playbooks? I didn't find something related to that, and this is what i'm trying to do.

FYI I am using a pacemaker cluster with cluster lvm.

But when the VG is created, it works only on one node and fail on the others (normal).
And I noticed that just after the no blocking error saying that the VG already exists on the other nodes (as it have been created on only one node and propagated to the other via cluster lvm), all the next ansible actions are only done on the node were the vg have been successfully created, so a lot of things are missing on the other nodes (all my nodes are both controller and compute, at least this is what I'm trying to do)

2. is there a correct way if one day we want to add a node in the openstack cluster without having too much troubles on the already running cluster? (same thing with both controler and compute role for the new node)

Thx a lot dude