Host-label disable-nohz-full=enabled not working for disabling the nohz_full kernel parameter

[Iago Filipe]

Brief Description

Adding disable-nohz-full=enabled label doesn't remove nohz cpu isolation.

Severity

Provide the severity of the defect.

<Critical: System/Feature is not usable after the defect>

Steps to Reproduce

1 - Install a system with standard kernel

2 - Verify nohz_parameter is enabled, for example:

compute-0:~$ cat /etc/default/grub | grep nohz
GRUB_CMDLINE_LINUX="module_blacklist=integrity,ima tboot=false crashkernel=512M biosdevname=0 console=ttyS0,115200n8 iommu=pt usbcore.autosuspend=-1 selinux=0 enforcing=0 nmi_watchdog=panic,1 softlockup_panic=1 softdog.soft_panic=1 intel_iommu=on user_namespace.enable=1 intel_idle.max_cstate=0 hugepagesz=1G hugepages=42 kvm-intel.eptad=0 hugepagesz=2M hugepages=0 default_hugepagesz=1G irqaffinity=1-43,45-87 rcu_nocbs=1-43,45-87 nohz_full=1-43,45-87 kthread_cpus=0,44 audit=0 audit_backlog_limit=8192 nopti nospectre_v2 nospectre_v1"

compute-0:~$ cat /sys/devices/system/cpu/nohz_full
1-43,45-87
3 - Assign the following label:

system host-lock <worker>
system host-label-assign <worker> disable-nohz-full=enabled
system host-unlock <worker>

4 - Verify nohz_full parameter is still enabled (should be disabled)

Expected Behavior

Nohz_full parameter should be disabled

Actual Behavior

Nohz_full parameter is enabled

Reproducibility

Reproducible

System Configuration

Standard - Test was run on worker node

Load info (eg: 2022-03-10_20-00-07)

BUILD_ID="2022-07-12_18-00-09"
JOB="StarlingX master"

Last Pass

First execution

Timestamp/Logs

Available at:

Alarms

Test Activity

Feature Testing

Workaround

No workaround

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.opendev.org/c/starlingx/config/+/849899

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.opendev.org/c/starlingx/config/+/849899
Committed: https://opendev.org/starlingx/config/commit/705077ed56e778685fcada8598c3c5d4592a38a3
Submitter: "Zuul (22348)"
Branch: master

commit 705077ed56e778685fcada8598c3c5d4592a38a3
Author: Iago Estrela <email address hidden>
Date: Thu Jul 14 18:25:20 2022 -0300

    Nohz_full configuration hieradata needs to be updated

    The grub commandline parameters contains nohz_full grub which needs
    to be updated with something different than a cpu list for kernel
    isolation when the user decides to disable it by assigning the
    disable-nohz-full label. In this manner, the disabled string can be
    used to signal the kernel to restore this configuration to default.

    Closes-Bug: 1981762

    Test plan:
    PASS: Disable nohz_full configuration in a standard worker by enabling
          disable-nohz-full label.
    PASS: Verify that /etc/default/grub doesn't contains nohz_full
    parameter.
    PASS: Verify that /sys/devices/system/cpu/nohz_full was restored to
          default.
    PASS: Enable nohz_full configuration in a standard worker by removing
          disable-nohz-full label.
    PASS: Verify that /sys/devices/system/cpu/nohz_full has the cpulist.
    PASS: Remove label and verify alarms and system overall health.
    PASS: Add label and verify alarms and system overall health.

    Signed-off-by: Iago Estrela <email address hidden>
    Change-Id: I00b68faae2612088b8e2fe7aeb3900babe638ca0