Uploaded image for project: 'OpenShift Virtualization'
  1. OpenShift Virtualization
  2. CNV-24784

[2164807] Migration metrics values are not sum up values from all VMIs

XMLWordPrintable

    • None

      +++ This bug was initially created as a clone of Bug #2148383 +++

      Description of problem:
      As mentioned here https://github.com/kubevirt/kubevirt/blob/main/docs/metrics.md migration metrics:
      kubevirt_migrate_vmi_scheduling_count,
      kubevirt_migrate_vmi_running_count,
      kubevirt_migrate_vmi_succeeded_total and
      kubevirt_migrate_vmi_failed_total
      these metrics should represnt the total count - sum up the values from all VMIs
      But instead they are representing values seperated by VMIs

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

      How reproducible:

      Steps to Reproduce:
      1.
      2.
      3.

      Actual results:

      Expected results:

      Additional info:

      — Additional comment from Akriti Gupta on 2022-11-25 09:36:11 UTC —

      As shown in the image attached , metric "kubevirt_migrate_vmi_succeeded_total" should contain the sum up value, that is 3 in this case, instead it shows 3 seperate values as 1 seperated by VMIs

      — Additional comment from João Vilaça on 2022-11-25 10:34:29 UTC —

      We agreed on this for visibility purposes and some technical details.
      But I agree it can be a bit confusing.

      My suggestion would be

      1) renaming kubevirt_migrate_vmi_succeeded_total and kubevirt_migrate_vmi_failed_total to remove the total:

      kubevirt_migrate_vmi_succeeded
      kubevirt_migrate_vmi_failed

      which would still maintain the labels.

      2) and create 2 recording rules

      kubevirt_migrate_vmi_succeeded_total
      kubevirt_migrate_vmi_failed_total

      which would be the sum of kubevirt_migrate_vmi_succeeded and kubevirt_migrate_vmi_failed, respectively

      My question would be, should the total metrics still have the namespace label or not?

      @sradco@redhat.com what do you think?

      — Additional comment from on 2022-11-28 13:05:34 UTC —

      Changing component to metrics as this BZ/discussion appears to be in that domain. Please feel free to revert this if that's not the correct choice.

      — Additional comment from Shirly Radco on 2022-11-29 12:46:33 UTC —

      I agree about the name change to drop the total since the metrics is granular and in on the vm name level.
      I don't see a need for creating the total metrics/recording rules at this point.
      This can easily be done by a PromQL query to get the total for all namespaces or by each namespace depending on the need.

      — Additional comment from João Vilaça on 2022-11-29 14:08:50 UTC —

      waiting for https://github.com/kubevirt/kubevirt/pull/8875 to be merged to cherry-pick the changes into release-0.58

      — Additional comment from Red Hat Bugzilla on 2022-12-15 08:28:48 UTC —

      Account disabled by LDAP Audit for extended failure

              jvilaca@redhat.com João Vilaça
              rhn-support-dbasunag Debarati Basu-Nag
              Votes:
              0 Vote for this issue
              Watchers:
              1 Start watching this issue

                Created:
                Updated:
                Resolved: