Low-latency worker node pmqos mgmt

[Brent Rowsell]

Brief Description
-----------------
On a low-latency worker with openstack deployed the c-state of the cores is managed.
If a core is assigned to a VM, the c-state is capped at C1 else it can go into a higher c-state if idle.

A similar mechanism is required for non-openstack worker nodes

Severity
--------
Major

Steps to Reproduce
------------------
See above

Expected Behavior
------------------
See above

Actual Behavior
----------------
See above

Reproducibility
---------------
100%

System Configuration
--------------------
All

Branch/Pull Time/Commit
-----------------------
N/A

Last Pass
---------
N/A

Timestamp/Logs
--------------
N/A

Test Activity
-------------
Other

[Ghada Khalil (gkhalil)]

Marking as release gating; this is needed to ensure consistent performance in the containerized env

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.opendev.org/661840

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.opendev.org/661840
Committed: https://git.openstack.org/cgit/starlingx/config/commit/?id=76b1a7a16f536f1187053a22c485d8343e8cc727
Submitter: Zuul
Branch: master

commit 76b1a7a16f536f1187053a22c485d8343e8cc727
Author: Jim Gauld <email address hidden>
Date: Tue May 28 16:34:12 2019 -0400

    Introduce PM QoS cpu latency manager for kubelet

    This creates a daemon 'pm-qos-mgr' that monitors kubelet cpu-manager
    static cpu assignments and modifies PM QoS CPU wakeup latency to
    govern the C-states per CPU.

    Guaranteed pods with exclusive CPUs get "low" cpu wakeup latency
    policy (the c-state is capped at C1).

    Remaining pods (i.e., with default CPUs) get "high" cpu wakeup latency
    policy, so the cpu may go into higher c-state if idle.

    Change-Id: I8470217dc53b6a7912b5023d8c0b04d966357222
    Closes-Bug: 1830545
    Signed-off-by: Jim Gauld <email address hidden>