Note: This bug is displayed in read-only format because the product is no longer active in Red Hat Bugzilla.

Bug 1948020

Summary:	[RHV-IPI ENV] Node IP is flapping from apiVIP/IngressVIP and DHCP managed IP
Product:	OpenShift Container Platform	Reporter:	Aditya Deshpande <adeshpan>
Component:	Machine Config Operator	Assignee:	Dan Winship <danw>
Status:	CLOSED DUPLICATE	QA Contact:	zhaozhanqi <zzhao>
Severity:	high	Docs Contact:
Priority:	unspecified
Version:	4.6.z	CC:	aconstan, aos-bugs, danw, harpatil, rphillips
Target Milestone:	---
Target Release:	---
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:	2021-04-26 19:08:59 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 Aditya Deshpande 2021-04-09 18:56:33 UTC

Description of problem:
OCP is installed with IPI method on RHEV. 
Upgrade was done from 4.6.18 to 4.6.22 using stable-4.6 channel.
After upgrade, we observed that nodeIPs are flapping to apiVIP for master nodes and IngressVIP for worker nodes in oc get nodes -o wide output.

In below output, 172.26.128.5 is apiVIP which is static IP not managed by DHCP. 
~~~
os4p-s9ds8-master-0         Ready    master         43d   v1.19.0+d46d32f   172.26.128.5   <none>        Red Hat Enterprise Linux CoreOS 46.82.202103110440-0 (Ootpa)   4.18.0-193.47.1.el8_2.x86_64   cri-o://1.19.1-11.rhaos4.6.git050df4c.el8
~~~

From node os4p-s9ds8-master-0:
~~~
2: ens3    inet 172.26.128.54/23 brd 172.26.129.255 scope global dynamic noprefixroute ens3\       valid_lft 178117sec preferred_lft 178117sec
2: ens3    inet 172.26.128.5/32 scope global ens3\       valid_lft forever preferred_lft forever
2: ens3    inet6 fe80::546f:98ff:fecb:e0/64 scope link noprefixroute \       valid_lft forever preferred_lft forever
~~~

172.26.128.5 IP should not be reflected in oc get nodes -o wide output. When it shows in command output, OCP cluster is not working as expected and some CO will go to degraded state.

Sometimes it is observed that 172.26.128.5 apiVIP is taken for pod IPs as well which are static pods to the nodes even at the time oc get nodes is not having that apiVIP. 

~~~
# oc get pods -o wide -A | grep "172.26.128.5 "
openshift-etcd                                     etcd-os4p-s9ds8-master-0                                  3/3     Running       0          67s     172.26.128.5    os4p-s9ds8-master-0         <none>           <none>
openshift-ovirt-infra                              coredns-os4p-s9ds8-master-1                               1/1     Running       0          3d4h    172.26.128.5    os4p-s9ds8-master-1         <none>           <none>
openshift-ovirt-infra                              haproxy-os4p-s9ds8-master-1                               2/2     Running       0          3d4h    172.26.128.5    os4p-s9ds8-master-1         <none>           <none>
openshift-ovirt-infra                              keepalived-os4p-s9ds8-master-1                            2/2     Running       0          3d4h    172.26.128.5    os4p-s9ds8-master-1         <none>           <none>
~~~



Version-Release number of selected component (if applicable):
OCP 4.6.22

How reproducible:


Steps to Reproduce:
1.
2.
3.

Actual results:
Node is flapping in between the api VIP, ingress VIP(static IP) to regular node's IP(DHCP managed IP).

Expected results:
oc get nodes -o wide should not show api VIP ingress VIP.

Additional info:
Note: OpenShiftSDN is being used in OCP cluster.

Comment 6 Dan Winship 2021-04-26 19:08:59 UTC

It appears that we are not correctly configuring hosts on oVirt in 4.6

*** This bug has been marked as a duplicate of bug 1948533 ***