-
Bug
-
Resolution: Unresolved
-
Major
-
None
-
4.15.0, 4.16.0, 4.17.0
-
Moderate
-
None
-
NE Sprint 256, NE Sprint 257, NE Sprint 258, NE Sprint 259, NE Sprint 260, NE Sprint 261, NE Sprint 262
-
7
-
Rejected
-
False
-
Description of problem:
Cluster-ingress-operator dns_controller should not watch Gateway API DNS and GWAPI service_dns_controller should. In AWS, when a dnsRecord is created in the openshift-ingress namespace for Gateway API, it is seen and acted upon by the watchers in both operator.dns_controller and operator.service_dns_controller. AWS DNS then proceeds to update the dns record for the domain, and delete the dnsRecord with the dnsName matching the domain. Then the dns_controller deletes the dnsRecord from the provider. This repeats again, in a different zone.
Version-Release number of selected component (if applicable):
4.17.0
How reproducible:
Easily
Steps to Reproduce:
1. Have an AWS cluster
2. Create a dnsRecord for a Gateway - running the e2e test https://github.com/openshift/cluster-ingress-operator/pull/1023 will do it
3. Observe that the DNS record is never published for a length of 10 minutes, and net.LookupHost will fail
Actual results:
The DNS appears to be published and then deleted.
Expected results:
When the DNS record is published, don't delete it until requested.
Additional info: