Bug 2104363

Summary:	OSP16->17: Needs appropriate guideline to upgrade stack with some indexes removed
Product:	Red Hat OpenStack	Reporter:	Takashi Kajinami <tkajinam>
Component:	openstack-tripleo-heat-templates	Assignee:	OSP Team <rhos-maint>
Status:	CLOSED DEFERRED	QA Contact:	Joe H. Rahme <jhakimra>
Severity:	high	Docs Contact:
Priority:	high
Version:	17.1 (Wallaby)	CC:	bshephar, igallagh, jkreger, mburns, ramishra, rhos-maint, sbaker
Target Milestone:	---	Keywords:	Documentation, Triaged
Target Release:	---	Flags:	ifrangs: needinfo? (rhos-maint)
Hardware:	Unspecified
OS:	Unspecified
Whiteboard:
Fixed In Version:		Doc Type:	If docs needed, set a value
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2022-09-26 20:21:05 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 Takashi Kajinami 2022-07-06 04:22:12 UTC

Description of problem:

I'm creating this bug here to record the concern while working on [1], so that we evaluate this and establish appropriate guideline/solution before we release 16.2 -> 17.1 upgrade.
[1] https://review.opendev.org/c/openstack/tripleo-heat-templates/+/848699

In RHOSP16.2 and older releases we use removal_policies in heat resource group when removing a node from overcloud.
The indexes which were already removed from the stack are not kept in deployment templates but stored in undercloud heat.

However in RHOSP17 we use ephemeral heat instead, and the removal_plolicies is not persisted and is always populated from templates.

This can cause a problem during upgrade, in case user upgrades RHOSP16 deployment which has a "intermediate" index removed.


Version-Release number of selected component (if applicable):


How reproducible:
Always

Steps to Reproduce:
1. Deploy RHOSP16 with 3 computes (compute-0, compute-1, compute-2)
2. Remove compute-1 
3. Upgrade the deployment to RHOSP17

Actual results:
TBD. This should be tested.

Expected results:
Node index should be kept and the deployment should have compute-0 and compute-2 after upgrade.


Additional info:

Comment 5 Julia Kreger 2022-09-26 20:21:05 UTC

Upon discussion of this among the team, we do not believe this would be an actual issue. As such, we're going to close this out as deferred as upgrade testing is expected to reveal issues.