Bug 1636618

Summary:	[GSS] uninstall.yml does not clean up after CNS/OCS installations
Product:	OpenShift Container Platform	Reporter:	Anton Mark <amark>
Component:	Installer	Assignee:	Jose A. Rivera <jarrpa>
Status:	CLOSED ERRATA	QA Contact:	Rachael <rgeorge>
Severity:	medium	Docs Contact:
Priority:	unspecified
Version:	3.10.0	CC:	aos-bugs, jarrpa, jialiu, jokerman, madam, mmccomas, pprakash
Target Milestone:	---
Target Release:	3.10.z
Hardware:	x86_64
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	No Doc Update
Doc Text:	undefined	Story Points:	---
Clone Of:		Environment:
Last Closed:	2019-07-24 13:47:18 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 Anton Mark 2018-10-05 21:35:35 UTC

Description of problem:
Directories and Gluster/Heketi configuration files are left behind on nodes where OCS/CNS was installed after uninstall playbook is run.

This will cause a side effect of re-installation of OCS/CNS on those nodes to show orphaned volumes in "gluster volume list". Running OCS/CNS uninstall playbook will avoid this situation. But, would also result in the whole CNS cluster being uninstalled. Sometimes it's necessary for a customer to remove and replace a single node.


Version-Release number of the following components:
openshift-ansible-3.10.47-1.git.0.95bc2d2.el7_5.noarch
ansible-2.4.6.0-1.el7ae.noarch

How reproducible:
Always.

Steps to Reproduce:
1. Install CNS/OCS
2. Run uninstall playbook against CNS/OCS node
3. Glusterfs deamonset HostPath's will remain on node.

Actual results:
Gluster/Heketi pod HostPath directories and files remain on node and are not cleaned up.

Expected results:
Node should be free of any remaining OCS/CNS directories and configuration files.

Additional info:
This has been reproduced multiple times by the customer.

Comment 1 Yaniv Kaul 2018-10-08 13:09:05 UTC

I assume it missed 3.10.z, is that going to 3.11?

Comment 2 Jose A. Rivera 2018-10-10 14:03:43 UTC

The uninstall playbook (which will correctly delete the directories) is in both release-3.11 and release-3.10, it just hasn't been tested. Moving this to MODIFIED.

Comment 3 Yaniv Kaul 2018-10-10 14:16:07 UTC

(In reply to Jose A. Rivera from comment #2)
> The uninstall playbook (which will correctly delete the directories) is in
> both release-3.11 and release-3.10, it just hasn't been tested. Moving this
> to MODIFIED.

Which version is it going to? is there an additional 3.10.z? Is QE expected to test it there, or in 3.11?

Comment 4 Jose A. Rivera 2018-10-10 15:37:15 UTC

It is in 3.10.45 and 3.11.1. I don't know what the QE plans are for testing, not do I know the release schedules for the OCP z-streams off-hand. Scott should be able to help with the latter, at least. In my opinion, OCS QE should primarily be testing on OCP 3.11, but this BZ should also be tested on OCP 3.10 if they're doing any testing on 3.10.

Comment 5 Scott Dodson 2018-10-12 18:20:52 UTC

QE will test each release there's a bug open for.

3.11.1 pre-dates the 3.11 GA build so it should be assumed to be fixed in 3.11 already. 


openshift-ansible-3.10.47-1.git.0.95bc2d2.el7_5 shipped 8 days ago to 3.10.z.

I'm moving this to ON_QA, QE please CLOSED CURRENTRELEASE once verified since the fix has already been released.

Comment 16 errata-xmlrpc 2019-07-24 13:47:18 UTC

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory, and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://access.redhat.com/errata/RHBA-2019:1755