Version:
$ openshift-install version
./openshift-install 4.6.0-0.nightly-2020-10-28-001707
built from commit 843cab21dd43251bee02e278f4bc06164533944f
release image registry.svc.ci.openshift.org/ocp/release@sha256:f0bf8cc2a0f9bea0d3d67af7d30e900841e0a8226ff3fa3880c7dfcb996b1d6c
Platform:
openstack
Please specify:
* UPI (semi-manual installation on customized infrastructure)
What happened?
Ingress operator goes into degraded due to masters hold the ingressVIP(192.168.0.7)
$ oc get co |grep -i -E "authentication|console|ingress"
authentication False True True 3h32m
console 4.6.0-0.nightly-2020-10-28-001707 False True True 3h19m
ingress 4.6.0-0.nightly-2020-10-28-001707 True False True 3h19m
$ for i in `oc get nodes -o name`; do oc debug $i -- chroot /host hostname -I ; done
Starting pod/wj46uos1028b-k5zm7-master-0-debug ...
To use host binaries, run `chroot /host`
192.168.2.22 192.168.0.7 10.128.0.1
Removing debug pod ...
Starting pod/wj46uos1028b-k5zm7-master-1-debug ...
To use host binaries, run `chroot /host`
192.168.3.226 192.168.0.5 10.129.0.1
Removing debug pod ...
Starting pod/wj46uos1028b-k5zm7-master-2-debug ...
To use host binaries, run `chroot /host`
192.168.3.167 10.130.0.1
Removing debug pod ...
Starting pod/wj46uos1028b-k5zm7-worker-0-debug ...
To use host binaries, run `chroot /host`
192.168.2.26 10.129.2.1
Removing debug pod ...
Starting pod/wj46uos1028b-k5zm7-worker-1-debug ...
To use host binaries, run `chroot /host`
192.168.3.58 10.128.2.1
Removing debug pod ...
Starting pod/wj46uos1028b-k5zm7-worker-2-debug ...
To use host binaries, run `chroot /host`
192.168.0.12 10.131.0.1
Removing debug pod ...
$ oc -n openshift-ingress get svc router-default -o wide
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE SELECTOR
router-default LoadBalancer 172.30.49.240 <pending> 80:30332/TCP,443:30195/TCP 134m ingresscontroller.operator.openshift.io/deployment-ingresscontroller=default
$ oc get pods -n openshift-ingress -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
router-default-58dd9f4f54-cqcd2 1/1 Running 0 135m 10.128.2.7 wj46uos1028b-k5zm7-worker-1 <none> <none>
router-default-58dd9f4f54-xjf8p 1/1 Running 0 135m 10.131.0.6 wj46uos1028b-k5zm7-worker-2 <none> <none>
What did you expect to happen?
ingress operator should work well and not be degraded
How to reproduce it (as minimally and precisely as possible)?
not sure
Anything else we need to know?
#Enter text here.
Based on the code I was able to find, I believe that this feature is only supported on the AWS and GCP platforms. It should not change the install behaviour on other platforms as far as I can tell.
To clarify this bug to public readers, this bug is not a duplicate, and setting publish: internal in the install-config will cause OpenStack installations to fail. Note that this is not a bug, but is not supported on the openstack platform and is not expected to be. The publish field will be set to external by default, so please just ignore it. We will be updating the docs to remove any suggestion of setting it with the OpenStack platform. We are following up on that in https://bugzilla.redhat.com/show_bug.cgi?id=1906801.
Closing!