Uploaded image for project: 'Hybrid Application Console'
  1. Hybrid Application Console
  2. HAC-5087

Single inflight_checks error can break ClusterList

This issue belongs to an archived project. You can view it, but you can't modify it. Learn more

    • Icon: Bug Bug
    • Resolution: Done
    • Icon: Critical Critical
    • None
    • None
    • Infrastructure
    • False
    • False
    • None

      We've seen a ROSA (HCP) cluster that is accessible in account_mgmt API (status: Active, display_name: greg-828):

      ocm get /api/accounts_mgmt/v1/subscriptions/2Uc0PmBCI7LuLjqxhd0cI5sD6oX

      but absent/inaccessible (404) in clusters_mgmt API:

      ocm get /api/clusters_mgmt/v1/clusters/25t3os3ioe02ma78inrelg3h2a4l82so
      ocm get /api/clusters_mgmt/v1/clusters/25t3os3ioe02ma78inrelg3h2a4l82so/inflight_checks

      There are several problems here:

      1. UI: this breaks ClusterList!  Some other clusters also become inaccessible.
      2. UI: "this page is temporarily unavailable" error doesn't show filters, making it harder to escape (1).
      3. UI: can't render even partial ClusterDetails for this cluster.
      4. What's up with the API? (ongoing Slack thread => OCM-3498)

      This Jira focuses on (1) as the most severe/urgent.

      Steps to reproduce — production

      1. open https://console.redhat.com/openshift.  If it loaded:
      2. type "greg-828" into "Filter by name or ID"
      3. enable "View only my clusters" toggle
      4. disable "View only my clusters" toggle

      Actual behavior

      Can't render a ClusterList page containing such cluster, due to inflight_checks request failing

      If you're "unlucky" to get this cluster on the 1st (or only) page of clusters list, you get grey "this page is temporarily unavailable" page; users that see some other list page before filtering/paging there keep seeing the last "successfully retrieved list", with a yellow triangle on top indicating the problem.

      So if you see few clusters you might get grey error on step 1.
      If not, after typing "greg-828" display name API should return only that cluster, but the previous list will remain — just with yellow triangle.
      Then steps 3–4 work for me to reach the grey error (I guess by making previous list empty?)

      Expected behavior

      For purposes of this card: Not critical whether the one problematic cluster is listed or not; but otherwise, ClusterList should work, and obey all filter changes.

      Steps to reproduce — local dev

      1. Delete file mockdata/api/clusters_mgmt/v1/clusters/21696acc0dbkvh0mh4lranlmvvqb11lg/inflight_checks.json
      2. open https://prod.foo.redhat.com:1337/openshift?env=mockdata — you should see grey "unavailable" page.

      Acceptance Critera

      • Remove in-flight check in Cluster List due to performance and async errors being thrown (revert part of https://gitlab.cee.redhat.com/service/uhc-portal/-/merge_requests/4456)
      • Was primarily added to show inflight validation errors to user from Cluster List, but without it user will see normal Error and can drilldown to cluster details to see any inflight validation errors.

              jswanke@redhat.com John Swanke
              bpaskinc@redhat.com Beni Paskin-Cherniavsky (Inactive)
              Archiver:
              rhn-support-sthamilt Stacey Hamilton
              LAKSHMI SHIVANTHI AMARACHINTHA LAKSHMI SHIVANTHI AMARACHINTHA

                Created:
                Updated:
                Resolved:
                Archived: