Uploaded image for project: 'Red Hat Workload Availability'
  1. Red Hat Workload Availability
  2. RHWA-88

NHC add maxUnhealthy config option

XMLWordPrintable

    • False
    • Hide

      None

      Show
      None
    • False
    • Hide
      Cause: Node Healtch Check operator allowed to configure minHealthy nodes, in contrast to MachineHealthCheck's maxUnhealthy, in order to avoid issues with nodes being deleted by remediators.
      Consequence: It was hard to configure a "n-1" scenario, which allows 1 remediation at a time only.
      Fix: Introduce an additional maxUnhealthy field, usage is mutually exclusive to minHealthy.
      Result: Easy configuration of n-1 scenarios

      Rewrite
      Cause: To avoid issues with nodes being deleted by remediators, the Node Health Check (NHC) Operator could configure the minHealthy nodes, but not the MachineHealthCheck's maxUnhealthy nodes.
      Consequence: As a result, it was difficult to configure a "n-1" scenario, which only allows one remediation at a time.
      Fix: To resolve the issue, an additional maxUnhealthy field, where the usage is mutually exclusive to minHealthy, was included.
      Result: It is not difficult to configure a "n-1" scenario.
      Show
      Cause: Node Healtch Check operator allowed to configure minHealthy nodes, in contrast to MachineHealthCheck's maxUnhealthy, in order to avoid issues with nodes being deleted by remediators. Consequence: It was hard to configure a "n-1" scenario, which allows 1 remediation at a time only. Fix: Introduce an additional maxUnhealthy field, usage is mutually exclusive to minHealthy. Result: Easy configuration of n-1 scenarios Rewrite Cause: To avoid issues with nodes being deleted by remediators, the Node Health Check (NHC) Operator could configure the minHealthy nodes, but not the MachineHealthCheck's maxUnhealthy nodes. Consequence: As a result, it was difficult to configure a "n-1" scenario, which only allows one remediation at a time. Fix: To resolve the issue, an additional maxUnhealthy field, where the usage is mutually exclusive to minHealthy, was included. Result: It is not difficult to configure a "n-1" scenario.
    • Feature
    • Proposed

      NHC can currently control minHealthy, but it is either a percentage or absolute value that is based on the total node count, which could potentially change over time.

      Given some use cases (like hyperconverged storage) where we want NHC to enforce single worker remediation action, adding "maxUnavailable" logic like MachineConfigPools use would provide admins better control.

        1. nhc_maxUnhealthy_4_16_connected.text
          21 kB
        2. far_nh_4_18_disconnected_pod_aff_maxunhealt_bmh_annot.text
          213 kB

              slintes Marc Sluiter
              jhopper@redhat.com Jenifer Abrams
              Votes:
              1 Vote for this issue
              Watchers:
              8 Start watching this issue

                Created:
                Updated:
                Resolved: