It's possible to remove nodegroup from operational environment even if there are ready nodes in that group

[Artem Panchenko]

Fuel version info (8.0 build #200): http://paste.openstack.org/show/479824/

It's possible to remove nodegroup from operational environment even if there are ready nodes in that group:

[root@nailgun ~]# fuel nodegroup
id | cluster_id | name
---|------------|---------------
1 | 1 | default
2 | 1 | group-custom-1
[root@nailgun ~]# fuel nodes
id | status | name | cluster | ip | mac | roles | pending_roles | online | group_id
---|--------|---------------------|---------|-------------|-------------------|------------|---------------|--------|---------
1 | ready | slave-02_controller | 1 | 10.109.10.3 | 64:9e:86:f3:42:42 | controller | | True | 2
4 | ready | slave-01_compute | 1 | 10.109.5.3 | 64:f7:5a:de:d8:c3 | compute | | True | 1
5 | ready | slave-03_cinder | 1 | 10.109.5.4 | 64:6b:3c:3b:63:03 | cinder | | True | 1
3 | ready | slave-06_controller | 1 | 10.109.10.5 | 64:02:97:06:36:0a | controller | | True | 2
2 | ready | slave-04_controller | 1 | 10.109.10.4 | 64:db:d1:f3:29:77 | controller | | True | 2
[root@nailgun ~]# fuel nodegroup --delete --group 2
Node group with id=2 was deleted!
[root@nailgun ~]# fuel nodes
id | status | name | cluster | ip | mac | roles | pending_roles | online | group_id
---|--------|---------------------|---------|-------------|-------------------|------------|---------------|--------|---------
5 | ready | slave-03_cinder | 1 | 10.109.5.4 | 64:6b:3c:3b:63:03 | cinder | | True | 1
1 | ready | slave-02_controller | 1 | 10.109.10.3 | 64:9e:86:f3:42:42 | controller | | True | None
4 | ready | slave-01_compute | 1 | 10.109.5.3 | 64:f7:5a:de:d8:c3 | compute | | True | 1
3 | error | slave-06_controller | 1 | 10.109.10.5 | 64:02:97:06:36:0a | controller | | True | None
2 | ready | slave-04_controller | 1 | 10.109.10.4 | 64:db:d1:f3:29:77 | controller | | True | None

The same action could be performed using Fuel GUI.

Steps to reproduce:

1. Create cluster
2. Add new nodegroup to it
3. Add nodes from both nodegroups to the cluster
4. Deploy cluster
5. Try remove custom nodegroup using CLI/GUI/API

Expected result: nodegroup removal is prohibited, because there are 'ready' nodes from that group

Actual result: nodegroup is removed, 'ready' nodes are marked as 'error'. If you create a new nodegroup and assign 'error' nodes to it, then their state will be changed to 'discover'

[Mikhail Zhnichkov (mzhnichkov)]

within bug #1529591 it is was fixed
Fix committed b5f606a26416a866f14f2a4100bd5493854b9c9c