Comment 2 for bug 2044115

Reviewed: https://review.opendev.org/c/starlingx/stx-puppet/+/900875
Committed: https://opendev.org/starlingx/stx-puppet/commit/1c39bc6c894edbbd9d9ef0f679e8e0b213789c6c
Submitter: "Zuul (22348)"
Branch: master

commit 1c39bc6c894edbbd9d9ef0f679e8e0b213789c6c
Author: Andre Kantek <email address hidden>
Date: Mon Nov 13 16:03:06 2023 -0300

    Turn off allocate-node-cidrs in kube-controller-manager

    It was detected the generation of the error message "CIDR allocation
    failed, there are no remaining CIDRs left to allocate in the accepted
    range" with high occurence in the log.

    By default allocate-node-cidrs is turned on to allocate address ranges
    for each node. But if cluster-cidr is set to a /64 (or /24 for IPv4
    install) mask with node-cidr-mask-size using the default 64 (or /24
    in IPv4) there will be no available range for the other nodes,
    controller-0 receives all available addresses for the pods.

    But StarlingX uses calico as the CNI solution and it does not use the
    allocated addresses from kube-controller-manager to distribute
    addresses among the pods. That is why this configuration was not
    noticed.

    This change turn off the value on existing installations to remove the
    error message if inserted as a sw patch.

    Test Plan
    [PASS] install an AIO-DX in IPv6 and apply this change as a patch and
           validate that the occurrence of CIDRNotAvailable events in the
           k8s nodes stops growing (kubectl describe node controller-1)
    [PASS] create user pods and validate that the address allocation is
           done to all pods in all nodes, following the ranges determined
           by calico (with kubectl get blockaffinities)

    Closes-Bug: 2044115

    Change-Id: I62b0cb482873703b4b266708fdba279aafd7c5c1
    Signed-off-by: Andre Kantek <email address hidden>