From b317ad3729eea609910634ce3d8b2e43fb01588f Mon Sep 17 00:00:00 2001
From: David Vallee Delisle <dvd@redhat.com>
Date: Thu, 10 Feb 2022 13:34:18 -0500
Subject: [PATCH] Adding Hugepages role parameter

Hugepages management was always a manual step done by operators via the
TripleO parameter ``KernelArgs``. This is error prone and causing confusion.

The new ``Hugepages`` parameter allow operators to define hugepages as
dictionnary, making it easier to read and follow.

To prevent unvolontary changes, there's multiple validations before
applying a change:
- We convert the current running configurations to an actual dictionnary
  that we validate the new format against
- If no change is necessary, even though the format might not be the same,
  there's no kernel_args update.
- By default, we don't remove hugepages in places except when operators
  specifically set the ``ReconfigureHugepages`` to true.

This change is also opening the door to more automations and automatic
tuning.

Related: https://bugzilla.redhat.com/show_bug.cgi?id=2043588
Depends-On: https://review.opendev.org/c/openstack/tripleo-ansible/+/830869
Change-Id: I1e05a5ea17c858a86acc170cfb91288884664b05
---
 .../kernel-boot-params-baremetal-ansible.yaml | 33 +++++++++++++++++++
 ...leo-kernel-hugepages-424c19a4b1579af8.yaml | 23 +++++++++++++
 2 files changed, 56 insertions(+)
 create mode 100644 releasenotes/notes/tripleo-kernel-hugepages-424c19a4b1579af8.yaml

diff --git a/deployment/kernel/kernel-boot-params-baremetal-ansible.yaml b/deployment/kernel/kernel-boot-params-baremetal-ansible.yaml
index a042a46a73..c20edf40b9 100644
--- a/deployment/kernel/kernel-boot-params-baremetal-ansible.yaml
+++ b/deployment/kernel/kernel-boot-params-baremetal-ansible.yaml
@@ -67,6 +67,33 @@ parameters:
     default: 900
     type: number
     description: Timeout in seconds to specify the wait time for ansible node reboot
+  Hugepages:
+    default: {}
+    type: json
+    description: >
+      Hash for hugepage configuration
+      Example:
+        Hugepages:
+          2048:
+            count: 100
+            default: false
+          1048576:
+            count: 100
+            default: true
+    tags:
+      - role_specific
+  ReconfigureHugepages:
+    default: false
+    type: boolean
+    description: >
+      By default, tripleo-kernel will configure the provided Hugepages as specified by operators
+      but it won't try to remove Hugepages configured on the host that are not part of the TripleO
+      paramater. Enabling this setting will make sure everything is reconfigured exactly like the
+      TripleO parameter, otherwise we just add the content of the Hugepages parameter to what's
+      already in place.
+    tags:
+      - role_specific
+
 
 resources:
   RoleParametersValue:
@@ -80,12 +107,16 @@ resources:
               isolated_cores: IsolCpusList
               kernel_args: KernelArgs
               kernel_args_defer_reboot: KernelArgsDeferReboot
+              hugepages: Hugepages
+              reconfigure_hugepages: ReconfigureHugepages
             - values: {get_param: [RoleParameters]}
           - values:
               TunedProfileName: {get_param: TunedProfileName}
               IsolCpusList: {get_param: IsolCpusList}
               KernelArgs: {get_param: KernelArgs}
               KernelArgsDeferReboot: {get_param: KernelArgsDeferReboot}
+              Hugepages: {get_param: Hugepages}
+              ReconfigureHugepages: {get_param: ReconfigureHugepages}
 
 outputs:
   role_data:
@@ -109,6 +140,8 @@ outputs:
             tasks_from: kernelargs.yml
           vars:
             tripleo_kernel_args: {get_attr: [RoleParametersValue, value, kernel_args]}
+            tripleo_kernel_hugepages: {get_attr: [RoleParametersValue, value, hugepages]}
+            tripleo_kernel_hugepages_remove: {get_attr: [RoleParametersValue, value, reconfigure_hugepages]}
             tripleo_kernel_defer_reboot: {get_attr: [RoleParametersValue, value, kernel_args_defer_reboot]}
             tripleo_kernel_reboot_timeout: {get_param: NodeRebootWaitTimeout}
       upgrade_tasks:
diff --git a/releasenotes/notes/tripleo-kernel-hugepages-424c19a4b1579af8.yaml b/releasenotes/notes/tripleo-kernel-hugepages-424c19a4b1579af8.yaml
new file mode 100644
index 0000000000..558878635a
--- /dev/null
+++ b/releasenotes/notes/tripleo-kernel-hugepages-424c19a4b1579af8.yaml
@@ -0,0 +1,23 @@
+---
+features:
+  - |
+    Adding Hugepages role parameter
+
+    Hugepages management was always a manual step done by operators via the
+    TripleO parameter ``KernelArgs``. This is error prone and causing confusion.
+
+    The new ``Hugepages`` parameter allow operators to define hugepages as
+    dictionnary, making it easier to read and follow.
+
+    To prevent unvolontary changes, there's multiple validations before
+    applying a change:
+
+      - We convert the current running configurations to an actual dictionnary
+        that we validate the new format against
+      - If no change is necessary, even though the format might not be the same,
+        there's no kernel_args update.
+      - By default, we don't remove hugepages in places except when operators
+        specifically set the ``ReconfigureHugepages`` to true.
+
+    This change is also opening the door to more automations and automatic
+    tuning.