Uploaded image for project: 'Red Hat Advanced Cluster Management'
  1. Red Hat Advanced Cluster Management
  2. ACM-3002

Host Inventory Settings on AWS/IPI lead to mis-configuration of IngressController

XMLWordPrintable

    • Icon: Bug Bug
    • Resolution: Done
    • Icon: Undefined Undefined
    • None
    • ACM 2.7.0
    • False
    • None
    • False
    • No

      Description of problem:

      Host Inventory Settings or CIM Enable documentation (technically equal) lead to misconfigured IngressController and so the agent-image-service can not be reached by external HTTP traffic and so the Discovery ISO can not be dowloaded.

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

      • 2.6+ for documentation
      • 2.7 for its UI-counterpart (equal) Host Inventory Settings

      How reproducible:

      • AWS/IPI 100% (i.e. ACM's weekly acm-console-development-cluster)
      • not sure about other cases, they depend on actual IngressController settings

      Steps to Reproduce:

      1. Install ACM and so the Assisted Service is deployed but not functional till AgentServiceConfig resource is created. To do o
      2. Go to All Clusters --> Infrastructure --> Host Inventory
      3. check: the list is empty and the Create button is disabled
      4. Click Host Inventory Settings, a modal is shown
      5. Since the cluster is on AWS, let it configure for AWS
      6. click Configure
      7. Wait till the process is finished (a green Success alert is shown on the Host Inventory page)
      8. create a new Host Inventory
      9. Open Add host via Discovery ISO, pass till "wget" command to download the ISO is shown
      10. copy & run the wget locally
      11. the wget process hangs. No error, no ISO is downloaded

      Actual results:

      The ISO is not downloaded since the HTTP traffic does not reach the agent-image-service

      Expected results:

      The ISO is downloaded

      Additional info:

      In AWS/IPI, the default IngressController has no "spec.domain" set so matching all traffic to *.CLUSTER_DOMAIN.

       

      The configuration above creates a new IngressController matching *nlb-apps.CLUSTER_DOMAIN, so conflicting with the default one.

      Same issue can be when the default (or any other) IngressController has spec.domain set to a conflicting value wit the .nlb-apps..

              mlibra@redhat.com Marek Libra
              mlibra@redhat.com Marek Libra
              Chad Crum Chad Crum
              Votes:
              0 Vote for this issue
              Watchers:
              2 Start watching this issue

                Created:
                Updated:
                Resolved: