Uploaded image for project: 'OpenShift Bugs'
  1. OpenShift Bugs
  2. OCPBUGS-61370

multus-networkpolicy daemonset rolls out one node at a time

XMLWordPrintable

    • Icon: Bug Bug
    • Resolution: Unresolved
    • Icon: Major Major
    • 4.20.0
    • 4.19, 4.20
    • Networking / multus
    • None
    • Quality / Stability / Reliability
    • False
    • Hide

      None

      Show
      None
    • None
    • None
    • None
    • None
    • Rejected
    • CNF Network Sprint 276
    • 1
    • Done
    • Bug Fix
    • Hide
      * Before this update, the time it took to apply updates to the `multus-networkpolicy` DaemonSet scaled linearly with the node count. With this release, this DaemonSet has been updated to allow for a 10% value for the `maxUnavailable` parameter so that the daemonset updates immediately in clusters larger than 10 nodes. (link:https://issues.redhat.com/browse/OCPBUGS-61370[OCPBUGS-61370])
      Show
      * Before this update, the time it took to apply updates to the `multus-networkpolicy` DaemonSet scaled linearly with the node count. With this release, this DaemonSet has been updated to allow for a 10% value for the `maxUnavailable` parameter so that the daemonset updates immediately in clusters larger than 10 nodes. (link: https://issues.redhat.com/browse/OCPBUGS-61370 [ OCPBUGS-61370 ])
    • None
    • None
    • None
    • None

      This is a clone of issue OCPBUGS-60421. The following is the description of the original issue:
      —
      Description of problem:

      When upgrading a large (121 node) cluster the rollout of the networking cluster operator takes a significant amount of time. During the update the cluster operator reports status noting "...56 out of 121 updated". This increments through each node on ~10s interval. For all the nodes to roll out this adds ~20 minutes to the upgrade.

      Looking at the daemonset "oc get daemonset -n openshift-multus multus-networkpolicy -o yaml" the rollingUpdate strategy has a maxUnavailable of 1 (so one node at a time). 

      The maxUnavailable should be set to 10% or 30% similar to other openshift component daemonsets.

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

      4.19.3
4.20.0-ec.5

      How reproducible:

      100%

      Steps to Reproduce:

          1. Upgrade openshift
    2. monitor the network cluster operator 
    3.

      Actual results:

          serial update of daemonset

      Expected results:

          higher level of concurrency

      Additional info:

       

              pliurh Peng Liu
              rhn-support-imiller Ian Miller
              None
              None
              Ying Wang Ying Wang
              None
              Votes:
              0 Vote for this issue
              Watchers:
              5 Start watching this issue

                Created:
                Updated: