Uploaded image for project: 'OpenShift Bugs'
  1. OpenShift Bugs
  2. OCPBUGS-25472

[vsphere] multi NICs machineset scaling leave nodes with taint node.cloudprovider.kubernetes.io/uninitialized

XMLWordPrintable

    • Important
    • No
    • False
    • Hide

      None

      Show
      None
    • Customer Escalated

      Description of problem:

      With the same symptoms of OCPBUGS-18641, but this time creating new nodes with the usage of machineset which have more than one NIC it ends up leaving the taint "node.cloudprovider.kubernetes.io/uninitialized" and misses the ProviderID value.

      Version-Release number of selected component (if applicable):

      As been noticed that in 4.13 in addition to the already explained symptoms, the vmware-vsphere-csi-driver-node goes in "CrashLoopBackOf" due to [Error: virtual machine wasn't found] while in 4.14 this does not happen.

      How reproducible:

      Very often

      Steps to Reproduce:

          1. Install an IPI cluster on VMware
          2. Create/Modify machineset with two network interfaces
          3. Scale out the machineset

      Actual results:

      - Nodes get created with taint "node.cloudprovider.kubernetes.io/uninitialized"
      - ProviderID is missing in the node definition
      - vmware-vsphere-csi-driver-node goes in CrashLoopBackOf" due to [Error: virtual machine wasn't found]

      Expected results:

      The nodes should get created without the taint, with the ProviderID and without pods failing

            mimccune@redhat.com Michael McCune
            rhn-support-csaggin Cristiano Saggin
            Milind Yadav Milind Yadav
            Votes:
            1 Vote for this issue
            Watchers:
            20 Start watching this issue

              Created:
              Updated: