Uploaded image for project: 'OpenShift Over the Air'
  1. OpenShift Over the Air
  2. OTA-964

Define graph-data risk schema for _id filtering

XMLWordPrintable

    • Icon: Story Story
    • Resolution: Done
    • Icon: Major Major
    • None
    • None
    • None
    • None
    • None
    • OTA 236

      In the enhancements repository, we have a registry of cluster-condition types, currently Always and PromQL. As we add the ability to filter by _id for this epic, we have two options:

      Squeeze into PromQL

      We could use the existing PromQL type, with a value of _id that works for existing, stand-alone clusters, like:

      max(
        cluster_infrastructure_provider{_id="",type=~"OpenStack|oVirt|VSphere"}
        or
        0 * cluster_infrastructure_provider{_id=""}
      )
      

      And then cluster-version operators for hosted control planes could learn a command line flag for "when you see _id="", substitute with _id="${YOUR_ACTUAL_CLUSTER_ID}". Or new CVOs could learn a flag for "even though you now default to that substitution, you're a stand-along cluster, and you should leave that block alone". Or something.

      Register a new type like PromQLTemplate

      Open an enhancement pull request adding this type to the registry. Teach new CVOs to understand the new type in 4.y. And start using the new type instead of PromQL for conditional updates from 4.y+ to later releases (both 4.y.z and 4.(y+n)).

      Definition of done

      Pick a direction, and write it down somewhere, either in this ticket if we decide to go with the PromQL squeeze, or in an enhancement-repository pull request if we go with PromQLTemplate.

              trking W. Trevor King
              trking W. Trevor King
              None
              None
              None
              None
              Votes:
              0 Vote for this issue
              Watchers:
              2 Start watching this issue

                Created:
                Updated:
                Resolved: