docs/doc/source/admintasks/kubernetes/install-ptp-notifications.rst
Elisamara Aoki Goncalves 484ea48a12 O-RAN Spec Compliant Timing API Notification - v1 compatibility (dsR8)
Story: 2010538
Task: 47862

Signed-off-by: Elisamara Aoki Goncalves <elisamaraaoki.goncalves@windriver.com>
Change-Id: I6f38e4e34371dcf9ae71d27286731633a41d3fa8
2023-05-04 10:57:48 -03:00

270 lines
8.2 KiB
ReStructuredText
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

.. xqd1614091832213
.. _install-ptp-notifications:
=========================
Install PTP Notifications
=========================
|PTP| notification is packaged as a system application and is managed
using the :command:`system application` and :command:`system helm-override`
commands.
The application monitors time related services on a host and provides an API
for subscribing to asynchronous status notifications as well as the ability to
pull the state of each service on demand.
.. note::
Changes to a node's |PTP| configuration, applied using the
:command:`system ptp-instance-apply`, requires the ``ptp-notification``
application to be removed and reapplied, using the
:command:`system application-remove ptp-notification`, and
:command:`system application-apply ptp-notification` commands.
This allows the containers to reload the updated configuration files and
monitor the services correctly.
**v1 API**
The legacy functionality of ``ptp-notification`` remains available and is
accessible through the v1 API; v1 is only capable of reporting status changes
for the |PTP| Sync State on a system.
**Limitations**
The v1 API only supports monitoring a single ``ptp4l + phc2sys`` instance.
Ensure the system is not configured with multiple instances when using the v1
API.
**v2 API**
The API conforms to O-RAN.WG6.O-Cloud Notification API-v02.01. Using the v2
API, multiple ``ptp4l`` instances can be tracked for independent |PTP| Sync
State and |PTP| Clock Class notifications.
The application monitors the following services:
- PTP Sync State
- PTP Clock Class
- OS Clock Sync State
- GNSS Sync State
- Overall System Sync State
.. rubric:: |context|
|prod-long| provides the capability for application(s) to subscribe to
asynchronous |PTP| status notifications and pull for the |PTP| state on demand.
You must provide Helm override values indicating the ``ptp4l`` and ``phc2sys``
instances that you want tracked by your ``ptp-notification`` application.
Since multiple ``ptp4l`` instances can be supported on a node, you must specify
the ``ServiceName`` of the instance that the ``ptp-notification`` application
should track.
For example, follow the steps below:
.. rubric:: |proc|
#. Apply labels to nodes that will be running the ``ptp-notification``.
#. Apply the registration label to the controller nodes.
.. code-block::
~(keystone_admin)]$ system host-label-assign controller-0 ptp-registration=true
#. Apply the notification label to each node that is configured for PTP
clock synchronization.
.. code-block::
~(keystone_admin)]$ system host-label-assign controller-0 ptp-notification=true
~(keystone_admin)]$ system host-label-assign compute-0 ptp-notification=true
#. Verify the labels.
.. code-block::
~(keystone_admin)]$ system host-label-list <node name>
#. Locate the application tarball on the system controller.
.. code-block::
~(keystone_admin)]$ ls /usr/local/share/applications/helm/stx-ptp-notification-helm-2.<minor_version>.tgz
#. Upload the ``ptp-notification`` application using the command below.
.. code-block::
~(keystone_admin)]$ system application-upload <path to application>
#. Verify if the application is in the uploaded state.
.. code-block::
~(keystone_admin)]$ system application-list
#. Apply Helm overrides as required.
.. code-block::
~(keystone_admin)]$ system helm-override-update ptp-notification ptp-notification notification --values notification-override.yaml
You can override the default values for the ``ptp-notification`` application:
#. Create a yaml file and update the fields that require Helm overrides.
.. code-block:: none
~(keystone_admin)]$ cat notification-override.yaml
ptptrackingv2:
enabled: True
ptp4lServiceName: True
phc2sysServiceName: True
ts2phcServiceName: True
log_level: INFO
control_timeout: 2
device:
simulated: false
holdover_seconds: 15
osclock:
holdover_seconds: 15
overall:
holdover_seconds: 15
where the values are:
**ptptracking**
ptp4lServiceName: True
phc2sysServiceName: True
ts2phcServiceName: True
- The ServiceName fields are defaulted to True in the application and
generally do not need to be altered.
- A service can be set to "False" in order to disable tracking for that
type. However, if a service type is not configured on a node
(ie. node does not use ts2phc), then the application will automatically
determine this and not attempt to monitor it.
- Use these fields if there is a service that is configured on the node
but you do NOT wish to track.
**log_level: INFO**
Set the logging level. DEBUG can be used for additional logs.
**control_timeout: 2**
control_timeout sets how frequently, in seconds the services are checked.
Value applies to all service types.
**device refers to ptp4l monitoring**
device:
holdover_seconds: 15
poll_freq_seconds: 2
osclock:
holdover_seconds: 15
overall:
holdover_seconds: 15
- ``holdover_seconds`` configures how long each service will stay in the
HOLDOVER state before transitioning to FREERUN. The holdover value
used by the application equates to: holdover_seconds
- (control_timeout * 2).
This is done in order to account for time between the monitor polling
cycles. The ``holdover_seconds`` value should be configured to match the
validated holdover time provided by the device manufacturer.
#. To configure the ``ptp-notification`` v1 API, include the following in
the ``notification-override.yaml`` file. Ensure that values are updated
to match the configured instance names on your system.
.. code-block:: none
ptptracking:
enabled: True
ptp4lSocket: /var/run/ptp4l-instancename
ptp4lServiceName: ptp4l-instancename
phc2sysServiceName: phc2sys-instancename
logging_level: INFO
device:
holdover_seconds: 15
poll_freq_seconds: 2
#. View existing values.
.. code-block:: none
~(keystone_admin)]$ system helm-override-show ptp-notification ptp-notification notification
#. Update and apply the values.
Application values can be added by the user and applied, using the following commands.
.. note::
The application could be in the "uploaded" or "applied" state.
.. code-block:: none
~(keystone_admin)]$ system helm-override-update ptp-notification ptp-notification notification -values <notification-override.yaml>
~(keystone_admin)]$ system application-apply ptp-notification
#. Verify the Helm overrides.
.. code-block::
~(keystone_admin)]$ system helm-override-show ptp-notification ptp-notification notification
#. Apply ``ptp-notification`` using the command below.
.. code-block::
~(keystone_admin)]$ system application-apply ptp-notification
#. Verify application status and pod status using the following commands:
#. Application Status
.. code-block::
~(keystone_admin)]$ system application-list
#. Pod Status
.. code-block::
~(keystone_admin)]$ kubectl get pods -n notification -o wide
.. rubric:: |postreq|
|prod-long| supports applications that rely on PTP for synchronization.
These applications are able to receive PTP status notifications from |prod-long|
hosting the application. For more information see:
- :ref:`PTP Notifications Overview <ptp-notifications-overview>`
- `API PTP Notifications <https://docs.starlingx.io/api-ref/ptp-notification-armada-app/api_ptp_notifications_definition_v1.html>`__
.. only:: partner
.. include:: /_includes/install-ptp-notifications-3a94b1ea1ae3.rest