"fuel node --deploy" allows run redeploy without provision

[Alexander Gubanov]

"fuel node --deploy" command allows run redeploy for node in non-provision state (ready), but after nothing happen -
"fuel node" or fuel UI just show reinstallation and few minutes later it becomes in ready state.

Env: mos 8.0 (build 402) neutron vlan, 3 controller nodes, 2 compute with cinder

Steps to reproduce: fuel node --node-id <node_id> --deploy

Details here: http://pastebin.com/6Ft2p2TB

[Arthur Svechnikov (asvechnikov)]

@ogubanov, Please could you provide the expected behavior?

[Oleg S. Gelbukh (gelbuhos)]

In my understanding, the expected behaviour is to run deployment tasks on the node. If nothing changed in settings and node state, don't do anything (just like proper Puppet modules do). Otherwise, apply changes accordingly.

I believe, this behaviour is required to change OpenStack settings on nodes (per BP https://blueprints.launchpad.net/fuel/+spec/openstack-config-change)

[Alexander Gubanov (ogubanov)]

Sure!

This docs https://docs.mirantis.com/openstack/fuel/fuel-7.0/user-guide.html (Node reinstallation)
says that we shoud run "provision" and after run "deploy"

so If node is in non-provision state I expected to see somethink like this:

[root@nailgun ~]# fuel node --node-id 4 --deploy
400 Client Error: Bad Request (Deployment operation cannot be started. Nodes with uids [4] are not provisioned yet. )

[Arthur Svechnikov (asvechnikov)]

@Oleg for OpenStack Config feature there are dedicated command 'fuel openstack-config --execute --env', also --deploy should work fine but longer.

In my understanding we should have an ability to run deploy for already 'ready' node. And if nothing changed then nothing happened, and visa versa.

P.S. For node reinstallation feature there strict workflow set keep flags -> provision -> deploy, if user run deploy for specific node it doesn't mean that user tries to use rollback feature

[Arthur Svechnikov (asvechnikov)]

Since I don't see any reason to change current behavior, because we can run --deploy for already deployed node. If I'm wrong feel free reopen issue.