You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@flink.apache.org by "Gyula Fora (Jira)" <ji...@apache.org> on 2022/06/23 11:43:00 UTC

[jira] [Created] (FLINK-28228) Target generation logic might skip over specs when observing already upgraded clusters

Gyula Fora created FLINK-28228:
----------------------------------

             Summary: Target generation logic might skip over specs when observing already upgraded clusters
                 Key: FLINK-28228
                 URL: https://issues.apache.org/jira/browse/FLINK-28228
             Project: Flink
          Issue Type: Bug
          Components: Kubernetes Operator
    Affects Versions: kubernetes-operator-1.1.0
            Reporter: Gyula Fora
            Assignee: Gyula Fora
             Fix For: kubernetes-operator-1.1.0


There are currently 2 problems with detecting already upgraded clusters in the AbstractDeploymentObserver:

1. Detecting the initial deployments is not reliable, because the user sends in an upgrade before the observer runs the logic will fail because we have a new higher generation.

2. When an upgrade was detected we should not simply use ReconciliationUtils.updateForSpecReconciliationSuccess because this will mark the current spec as reconciled which is not necessarily true



--
This message was sent by Atlassian Jira
(v8.20.7#820007)