Uploaded image for project: 'OpenShift Bugs'
  1. OpenShift Bugs
  2. OCPBUGS-36463

etcd data store is leftover when infrastructure provisioning fails

XMLWordPrintable

    • Icon: Bug Bug
    • Resolution: Done-Errata
    • Icon: Critical Critical
    • None
    • 4.16.0, 4.17.0
    • Installer / openshift-installer
    • None
    • Moderate
    • No
    • Proposed
    • False
    • Hide

      None

      Show
      None
    • Hide
      * Previously, the etcd data store used for CAPI installs was only removed when either the bootstrap node or the cluster was destroyed. With this release, if there is an error during infrastructure provisioning, the data store is removed and does not take up unnecessary disk space. (link:https://issues.redhat.com/browse/OCPBUGS-36463[*OCPBUGS-36463*])
      ------------------------------------------------
      Cause: The etcd data store used for CAPI installs was only removed when destroying the bootstrap node or the cluster.

      Consequence: If infrastructure provisioning failed, the etcd data store was still preserved on disk despite the fact that it cannot be used, which wastes unnecessary disk space.

      Fix: Delete the etcd data store if there is an error during infrastructure provisioning.

      Result: If there is an error during infrastructure provisioning, the data store is removed and does not take up unnecessary disk space.
      Show
      * Previously, the etcd data store used for CAPI installs was only removed when either the bootstrap node or the cluster was destroyed. With this release, if there is an error during infrastructure provisioning, the data store is removed and does not take up unnecessary disk space. (link: https://issues.redhat.com/browse/OCPBUGS-36463 [* OCPBUGS-36463 *]) ------------------------------------------------ Cause: The etcd data store used for CAPI installs was only removed when destroying the bootstrap node or the cluster. Consequence: If infrastructure provisioning failed, the etcd data store was still preserved on disk despite the fact that it cannot be used, which wastes unnecessary disk space. Fix: Delete the etcd data store if there is an error during infrastructure provisioning. Result: If there is an error during infrastructure provisioning, the data store is removed and does not take up unnecessary disk space.
    • Bug Fix
    • In Progress

      This is a clone of issue OCPBUGS-36284. The following is the description of the original issue:
      —
      Description of problem:

          The etcd data store is leftover in the <install dir>/.clusterapi_output/etcd dir when infrastructure provisioning fails. This takes up a lot of local storage space and is useless.

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

      How reproducible:

          Always

      Steps to Reproduce:

          1. Start an install
    2. introduce an infra provisioning failure -- I did this by editing the 02_infra-cluster.yaml manifest to point to a non existent region
    3. Check .clusterapi_output dir for etcd dir

      Actual results:

          etcd data store remains

      Expected results:

          etcd data store should be deleted during infra provisioning failures. It should only be persisted to disk if there is a failure/interrupt in between infrastructure provisioning and bootstrap destroy, in which case it can be used in conjunction with the wait-for and destroy bootstrap commands.

      Additional info:

              rh-ee-bbarbach Brent Barbachem
              openshift-crt-jira-prow OpenShift Prow Bot
              Jinyun Ma Jinyun Ma
              Votes:
              0 Vote for this issue
              Watchers:
              6 Start watching this issue

                Created:
                Updated:
                Resolved: