Bug 1597270

Summary:	[Kubevirt APB] When installing with a version it is not passed to kubevirt-ansible
Product:	Container Native Virtualization (CNV)	Reporter:	Nelly Credi <ncredi>
Component:	Installation	Assignee:	Sergi Jimenez Romero <sjr>
Status:	CLOSED UPSTREAM	QA Contact:	Lukas Bednar <lbednar>
Severity:	high	Docs Contact:
Priority:	high
Version:	1.1	CC:	fsimonce, lbednar, ncredi, rhallise, sjr
Target Milestone:	---
Target Release:	1.3
Hardware:	Unspecified
OS:	Unspecified
Whiteboard:
Fixed In Version:	1.3	Doc Type:	If docs needed, set a value
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2018-12-04 16:59:02 UTC	Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description Nelly Credi 2018-07-02 12:21:14 UTC

Description of problem:
I got into a situation where kubevirt-ansible wasnt updated with latest kubvirt releases (so it was on 0.6 by default) and while the kubevirt apb got version 0.7.0-alpha.2, i ended up with a mismatch
(in this specific case crds were ovm & vm, instead of vm & vmi) 


Version-Release number of selected component (if applicable):
0.7.0-alpha.2

How reproducible:
100%

Steps to Reproduce:
1. deploy latest kubevirt apb, while kubevirt-ansible as an old version
2.
3.

Actual results:
a mismatch of versions

Expected results:
i would expect the apb to be able to set the version when calling the provisioning playbook in kubevirt-ansible

Additional info:
Ryan already started looking at it, and said that the 
command being run is 'ansible-playbook -i /etc/ansible/hosts playbooks/provision.yml -e kubevirt_version='0.7.0-alpha.2' ....'.  Those vars should be getting to the kubevirt-ansible tasks.

Comment 1 Sergi Jimenez Romero 2018-07-04 08:52:34 UTC

Nelly,

I'm just starting to have a look to it, I think it'd be a matter of making kubevirt-apb to pass the version that it gets from user (or the default value) down to kubevirt-ansible.

Going to test: https://github.com/tripledes/kubevirt-apb/commit/5637d63176fceea2198895b881fb4a491b2f57c2 that and will report back ASAP.

Comment 2 Sergi Jimenez Romero 2018-07-05 07:05:19 UTC

PR sent:

* https://github.com/ansibleplaybookbundle/kubevirt-apb/pull/74

After applying the PR, the images pulled are the correct ones, e.g. using version 0.6.0 on the APB form results in the correct images being pulled and started:

docker.io/kubevirt/virt-api                                     v0.6.0                           d27daff06b15        3 weeks ago
docker.io/kubevirt/virt-handler                                 v0.6.0                           27449e9e6655        3 weeks ago
docker.io/kubevirt/virt-controller                              v0.6.0                           b7849a90761c        3 weeks ago

Comment 3 Sergi Jimenez Romero 2018-07-05 21:09:13 UTC

The PR was merged.

Comment 4 Sergi Jimenez Romero 2018-07-24 10:23:21 UTC

As the PR was already merged, I was wondering:

@Lukas, do you need to test anything?

@Nelly, do you still see the reported issue?

Comment 5 Lukas Bednar 2018-08-09 13:09:57 UTC

I am trying to verify this bug, but I am hitting new issue https://bugzilla.redhat.com/show_bug.cgi?id=1614322 .

Comment 6 Lukas Bednar 2018-10-08 11:46:18 UTC

I am not sure how to verify this bug at the moment.

It seems to be u/s issue - I don't remember if we were able to reproduce it on d/s.

We don't pass `kubevirt_version` parameter to d/s APB, we use docker_tag & registry_url & registry_namespace instead.

So I can do verification with u/s APB if you like - that means it is not 1.2 version as it is targeted in this bug.

Any thoughts ?

Comment 7 Nelly Credi 2018-10-08 11:56:07 UTC

yes, i believe the only way to verify it is to try and pass in the u/s APB a version which is not updated in kubevirt-ansible (v0.9 for example)
and make sure that the APB calls kubevirt-ansible with the version param

Comment 8 Lukas Bednar 2018-10-09 08:14:28 UTC

I gave a try and failed with other issue, it seems that u/s apb is broken. Or interface was changed.

[root@cnv-executor-lbednar-master1 ~]# oc get clusterserviceclass -o custom-columns=Name:spec.externalName,externalID:spec.externalID,Broker:spec.clusterServiceBrokerName
Name                              externalID                             Broker
dockerhub-virtualization          034b5848406905aea463db2ec2e6d876       ansible-service-broker

[root@cnv-executor-lbednar-master1 ~]# cat kubevirt-apb-us.yml 
---
apiVersion: servicecatalog.k8s.io/v1beta1
kind: ServiceInstance
metadata:
  name: kubevirt
  namespace: kube-system
spec:
  clusterServiceClassExternalName: dockerhub-virtualization
  clusterServicePlanExternalName: default
  parameters:
    admin_user: test_admin
    admin_password: 123456
    version: "0.9.0"

oc create -f kubevirt-apb-us.yml

[root@cnv-executor-lbednar-master1 ~]# oc logs -n dockerhub-virtualization-prov-td669 bundle-a92c85ef-85c0-4991-81da-8452714746f1
TASK [kubevirt : Render KubeVirt Yaml] *****************************************
fatal: [localhost]: FAILED! => {"changed": false, "msg": "AnsibleUndefinedVariable: 'image_pull_policy' is undefined"}

PLAY RECAP *********************************************************************
localhost                  : ok=8    changed=5    unreachable=0    failed=1

Comment 9 Ryan Hallisey 2018-10-12 14:14:32 UTC

https://github.com/kubevirt/kubevirt-ansible/pull/407 will fix latest.

Comment 10 Federico Simoncelli 2018-10-15 20:54:23 UTC

Sergi please fill in the "Fixed In Version" field.

Comment 11 Sergi Jimenez Romero 2018-10-16 12:40:48 UTC

Federico, added 1.3 (current target), not sure what version was it, moved on to supportability team and lost track of this bug.

Comment 14 Nelly Credi 2018-12-04 16:59:02 UTC

This bug is an upstream bug,
i was unable to verify it in this version timeframe
and it will become irrelevant in the next version

closing as upstream