Bug 1938084

Summary:	[vsphere] installation failed with some ovn network errors
Product:	OpenShift Container Platform	Reporter:	jima
Component:	Networking	Assignee:	Mohamed Mahmoud <mmahmoud>
Networking sub component:	ovn-kubernetes	QA Contact:	huirwang
Status:	RELEASE_PENDING ---	Docs Contact:
Severity:	high
Priority:	unspecified	CC:	anbhat, bgalvani, ckeller, huirwang, rbrattai, thaller, trozet, zzhao
Version:	4.7	Keywords:	Reopened, TestBlocker
Target Milestone:	---
Target Release:	4.8.0
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:	2021-03-27 14:19:49 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 jima 2021-03-12 06:21:03 UTC

Description of problem:
Deploy upi-on-vsphere cluster with ovn network enabled, installation is failed at the step of waiting for bootstrap completed.

Checked on master node, found that service ovs-configuration was started failed, and br-ex link is not up.

Attached gather log for details.

Version-Release number of selected component (if applicable):
4.7.0-0.nightly-2021-03-10-202622

How reproducible:
Always on upi-on-vsphere, didn't find on ipi-on-vsphere

Steps to Reproduce:
1. Deploy upi-on-vsphere + ovn network enabled
2. Installation is failed
3.

Actual results:
Fail to deploy cluster

Expected results:
Cluster is deployed successfully

Additional info:

Comment 8 Thomas Haller 2021-03-20 09:25:11 UTC

> it seems we are missing a plugin
>
> sh-4.4# cat /etc/NetworkManager/NetworkManager.conf 

NetworkManager ships an empty configuration file (or better: with only comments). This is intended.


> In /etc/NetworkManager/NetworkManager.conf 
> [main]
> #plugins=ifcfg-rh
> [...]
> based on the above I think we have to un comment the plugins line and restart NetworkManager

The default on rhel-8 is already "plugins=ifcfg-rh" (which is the same as "plugins=ifcfg-rh,keyile") (while on Fedora 33+ it is "plugins=keyfile,ifcfg-rh").

The setting in NetworkManager.conf is commented out to show what the default is. Uncommenting it, has little effect.


Except, if would have an effect if you had

  /usr/lib/NetworkManager/conf.d/10-some.conf
  > [main]
  > plugins=something

then having 

  /etc/NetworkManager/NetworkManager.conf
  > [main]
  > plugins=ifcfg

would make a difference.


Run `sudo NetworkManager --print-config` to see the configuration.

Comment 14 Mohamed Mahmoud 2021-03-24 17:04:19 UTC

This fix updated the configfile for you 
https://github.com/openshift/machine-config-operator/commit/a2815aa4993954ec6f30785c8c0a1d0a79109a4b
and it was merged to 4.7

Comment 15 Tim Rozet 2021-03-24 17:55:46 UTC

The above fix was merged a few days ago into 4.7: https://github.com/openshift/machine-config-operator/pull/2426

So it makes sense it suddenly started working for you. But you probably need the cloned-mac-address fix as well.

Comment 17 Tim Rozet 2021-03-25 16:44:37 UTC

*** Bug 1942006 has been marked as a duplicate of this bug. ***

Comment 19 Mohamed Mahmoud 2021-04-20 19:14:16 UTC

*** Bug 1949351 has been marked as a duplicate of this bug. ***