[nailgun] Reuse node names instead keeping/wasting them

[Aleksandr Shaposhnikov]

I worried about this behavior. Expected behavior for me is to allow manually delete at "detected nodes screen" so after they got deleted they could be reclaimed by new nodes.

[Mike Scherbakov (mihgen)]

Can you please provide more information on the flow? Do you mean removal of bootstrap nodes here? Then what do you mean by "reclaiming" by new nodes?

[Aleksandr Shaposhnikov (alashai8)]

I'm sorry that my vision of expected behavior wasn't clear.
Here is steps to reproduce:
1. Deploy fuel masternode.
2. Boot up couple nodes. They will be assigned node-1 and node-2.
3. Deploy some small setup like multinode + 1 compute.
4. Wait for deployment will be finished.
5. Delete cluster.
6. Nodes gets released but will receive names node-3 and node-4 accordingly.

Expected behavior: node-1 and node-2 after cloud deletion or after deleting them on special page.

[Dmitry Pyzhov (dpyzhov)]

I think it is a good behavior. If you do not reuse hostnames, then you do not have a mess with clusters. For example, you do not need to check if the node still in one cluster or it was moved to another environment.

We have enough integer numbers.

[Mike Scherbakov (mihgen)]

Alex & Dmitry, please come up with a blueprint for this. UX is still not fully clear to me..

[Dmitry Pyzhov (dpyzhov)]

Looks like it has been fixed by https://blueprints.launchpad.net/fuel/+spec/node-naming

[Ksenia Svechnikova (kdemina)]

MOS 7.0 build:301

Steps:

1. Deploy fuel master node.
2. Deploy env with 5 nodes (1,2,3,4,5)
3. Delete cluster
4. Create cluster and add new 5 nodes and by default the hostname will have unique ids:
[root@fuel-lab5-mos6 ~]# fuel node
id | status | name | cluster | ip | mac | roles | pending_roles | online | group_id
---|----------|-----------------------------|---------|--------------|-------------------|-------|-------------------|--------|---------
11 | discover | cz7310-kvm.host-telecom.com | 2 | 172.16.40.70 | 0c:c4:7a:13:48:42 | | controller, mongo | True | 2
8 | discover | cz7369-kvm.host-telecom.com | 2 | 172.16.40.80 | 0c:c4:7a:15:00:a8 | | controller, mongo | True | 2
7 | discover | cz7387-kvm.host-telecom.com | 2 | 172.16.40.79 | 0c:c4:7a:17:8d:a0 | | controller, mongo | True | 2
10 | discover | cz7368-kvm.host-telecom.com | 2 | 172.16.40.75 | 0c:c4:7a:15:01:3c | | cinder, compute | True | 2
9 | discover | cz7309-kvm.host-telecom.com | 2 | 172.16.40.72 | 0c:c4:7a:13:48:62 | | cinder, compute | True | 2

Node's id change after any node's deletion, it's expected behavior.
Default hostname is set according to the node ID
But with the realization of BP https://blueprints.launchpad.net/fuel/+spec/node-naming user now have a possibility to set custom hostname before node provisioning. It also could be node-1, node-2, but still id will be unique.

I assume that this issue could be marked as Fix Released