Uploaded image for project: 'OpenShift Workloads'
  1. OpenShift Workloads
  2. WRKLDS-1717

VM with "evictionStrategy: None" is killed by descheduler when the node is overutilized

XMLWordPrintable

    • Icon: Bug Bug
    • Resolution: Done
    • Icon: Undefined Undefined
    • None
    • None
    • Quality / Stability / Reliability
    • False
    • Hide

      None

      Show
      None
    • None
    • None
    • None
    • None
    • None
    • None
    • None
    • None

      We advice to set "EvictionStrategy: None" to non-migratable VMs, but with descheduler these VMs are simply killed when it tries to do the rebalancing. 
       
      With the new `DevKubeVirtRelieveAndMigrate` profile specific for the VMs, can we adjust the descheduler logic for virt-launcher pods?
       
      Probably, VMs with ES:None should be ignored from the balancing.

              stirabos Simone Tiraboschi
              dshchedr@redhat.com Denys Shchedrivyi
              None
              None
              None
              None
              None
              Votes:
              0 Vote for this issue
              Watchers:
              2 Start watching this issue

                Created:
                Updated:
                Resolved: