-
Bug
-
Resolution: Done
-
Normal
-
None
-
None
-
None
-
5
-
False
-
None
-
False
-
-
-
GITOPS Sprint 255, GITOPS Service EE Sprint 256, GITOPS Service EE Sprint 257
Description of problem:
Operator overrides dex connector configuration in argocd-cm upon setting .spec.sso.dex.config in ArgoCD CR. This is observed for openshift connector
Slack Discussion: https://redhat-internal.slack.com/archives/C048WUV584F/p1699541069712629
Workaround (for openshift connector only)
Note: We do not recommend this workaround unless it is critical
Save the connector configuration from argocd-cm and add them into ArgoCD CR along with other fields being added to .spec.sso.dex.config
Prerequisites (if any, like setup, operators/versions):
Steps to Reproduce
- Add the below configuration to .spec.sso
provider: dex dex: config: | expiry: idTokens: "1h" signingKeys: "12h"
Actual results:
argocd-cm is reset with
dex.config: | expiry: idTokens: "1h" signingKeys: "12h"
As the connector configuration is not found, dex pod goes into CrashLoopBackOff:
level=fatal msg="malformed Dex configuration found"
Expected results:
Dex configuration in ArgoCD CR is appended to argocd-cm instead of overriding
Reproducibility (Always/Intermittent/Only Once):
Always
Acceptance criteria:
- dex openshift connector configs are not overridden in argocd-cm on setting .spec.sso.dex.config in ArgoCD CR
- Fields from .spec.sso.dex.config are appended to argocd-cm
- dex pod is up and running
Definition of Done:
Build Details:
Additional info (Such as Logs, Screenshots, etc):