James Harrington created JBCLUSTER-316:
------------------------------------------
Summary: remoteingress to patch ClusterIngress with certificate bundle
Key: JBCLUSTER-316
URL:
https://issues.redhat.com/browse/JBCLUSTER-316
Project: JBoss Clustering
Issue Type: Feature Request
Reporter: James Harrington
Assignee: Paul Ferraro
Once a cluster is installed Hive should patch a cluster's ingresscontroller with the
new certificate bundle from Lets Encrypt to enable the console URL to work as soon as
those certificates are available.
Hive's remoteingress controller
[
here|https://github.com/openshift/hive/blob/master/pkg/controller/remotei...]
used to do exactly this by creating a new ingresscontroller object in the
openshift-ingress-operator namespace but raced with the cloud-ingress-operator to manage
it so it was disabled
[
here|https://gitlab.cee.redhat.com/service/app-interface/-/blob/master/da...].
I’m not sure if we can replace creating the entire ingresscontroller or just patch the new
`.spec.defaultCertificate` to be the new certificate bundle. The piece I don’t know about
is if the cluster-ingress-operator will load in the new certificates or do we have to
completely replace the object? For example cloud-ingress-operator has to replace the
entire ingresscontroller CR to have it change the load balancer from external to internal,
I suspect the cert patch will work since the cert isn’t terminated on the ELB. The
defaultCertificate ref is in the ingresscontroller spec Hive was updating it
[
here|https://github.com/openshift/hive/blob/master/pkg/controller/remotei...]
This controller likely raced with cloud-ingress-operator because they were both watching
and reconciling the same object. Maybe we could enable this controller but disable the
watch when “ActiveAPIURLOverride” is set in the ClusterDeployment? Though that doesn’t
feel right.
--
This message was sent by Atlassian Jira
(v7.13.8#713008)