Bug 1327907

Summary:	OS::TripleO::NodeExtraConfigPost customizations with a CREATE action are also re-run on upgrade
Product:	Red Hat OpenStack	Reporter:	Udi Kalifon <ukalifon>
Component:	rhosp-director	Assignee:	Angus Thomas <athomas>
Status:	CLOSED NOTABUG	QA Contact:	Arik Chernetsky <achernet>
Severity:	high	Docs Contact:
Priority:	high
Version:	8.0 (Liberty)	CC:	bnemec, ccamacho, dbecker, mandreou, mbultel, mburns, mcornea, morazi, rhel-osp-director-maint, rhos-flags, ukalifon
Target Milestone:	zstream	Keywords:	Triaged
Target Release:	8.0 (Liberty)
Hardware:	Unspecified
OS:	Unspecified
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2018-09-07 11:12:56 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:

Description Udi Kalifon 2016-04-17 19:40:58 UTC

Description of problem:
When upgrading 7.3 to 8.0 and the stack has a post-configuration like in the example in http://docs.openstack.org/developer/tripleo-docs/advanced_deployment/extra_config.html (the last example), the post-configuration is run even if it has no UPDATE action specified.


How reproducible:
100%


Steps to Reproduce:
1. deploy a 7.3 overcloud with a post-deployment like in the example in http://docs.openstack.org/developer/tripleo-docs/advanced_deployment/extra_config.html (the last example)
2. delete whatever the post-deployment created, so you would be able to know if it re-ran or not
3. Upgrade to 8.0 according to the guide


Actual results:
The post-configuration was reapplied, even if the upgrade doesn't really create new nodes.


Expected results:
Only if the configuration is mapped to the UPDATE action in addition to the CREATE action, should it be re-applied on upgrades.

Comment 2 Marios Andreou 2016-04-18 12:11:08 UTC

Hey Udi, we just chatted with jistr about this on irc. It seems like another consequence of our explicit setting of OS::Heat::None for the entire OS::TripleO::ControllerPostDeployment: during the first two upgrade steps. SO when it gets unset back to normal on the converge it gets treated as create.

I don't know what the specific extra config being applied here is and if it causes problems. But I think this is a fair point that you would not expect it to run. How about, we note the issue in the docs, that any node extra config will be re-executed on upgrade. If you don't want that to happen then explicitly set the OS::TripleO::NodeExtraConfigPost to OS::Heat::None in the pacemaker_converge.yaml environment file?

Comment 4 Udi Kalifon 2017-09-03 13:20:21 UTC

I think it is a very complicated workaround, in an upgrade process which is already complicated enough... Setting OS::TripleO::NodeExtraConfigPost to OS::Heat::None can also have other implications, otherwise it would already be set like that (as it is in the first 2 steps). But I'm not an expert and it's not my decision to say if documenting it is enough. In any ways it would need a lot of testing.

Comment 5 Carlos Camacho 2018-09-07 11:12:56 UTC

Due to the time of creation of this BZ I'll close it, please feel free to re-open it if it's required.