Uploaded image for project: 'Red Hat OpenShift AI Engineering'
  1. Red Hat OpenShift AI Engineering
  2. RHOAIENG-4542

Selecting the "Internal Service" link for a failed model deployment leads to an uncoverable error

XMLWordPrintable

    • False
    • Hide

      None

      Show
      None
    • False
    • No
    • No
    • Important
    • Testable

      With modelmesh, when a model deployment fails the dashboard still displays an "Internal Service" link button in the "Inference endpoint" column of the models table (both under a project and on the main Model Serving page). When a user clicks the link button, an unrecoverable error is displayed and the UI freezes. The only way to recover is to reload the dashboard.  I don't know if the same issue exists when Kserve is used.

      The issue is discussed in this Slack thread: https://redhat-internal.slack.com/archives/C05SMJ09DD2/p1710362828353969

      The idea of not displaying a link at all for failed deployments was also discussed earlier here: https://redhat-internal.slack.com/archives/C05SMJ09DD2/p1708013503248129 I talked with kaedward about this and we concluded that for the case of failed deployments we should simply display "–-" in the Inference endpoint column.

            Unassigned Unassigned
            vconzola Vincent Conzola
            RHOAI Dashboard
            Votes:
            0 Vote for this issue
            Watchers:
            2 Start watching this issue

              Created:
              Updated: