Yoga

The Yoga OpenStack Charms release includes updates for the charms described on the Supported charms page. As of this release, the project consists of 62 stable charms.

For the list of bugs resolved in this release refer to the 22.04 milestone in Launchpad.

For scheduling information of past and future releases see the Release schedule.

Note

Release Notes contents is superseded by updated information published in the OpenStack Charm Guide (this guide) after the release of any given OpenStack Charms version.

Important

Always upgrade to the latest stable charms before making any major changes to your cloud and before filing bug reports. Note that charm upgrades and OpenStack upgrades are functionally different. For instructions on performing the different upgrade types see Upgrades overview in the OpenStack Charms Deployment Guide.

New stable charm features

With each new feature, there is a corresponding example bundle in the form of a test bundle, and/or a section in the documentation that details its usage. Test bundles are located in the src/tests/bundles directory of the relevant charm repository (see all charm repositories).

nova-compute charm: vTPM support

The nova-compute charm now supports virtual TPM (vTPM) devices and is enabled via a new configuration option: enable-vtpm. When it is set to ‘True’, the Nova and libvirt services will be configured to provide vTPM devices via swtpm.

This requires the swtpm, swtpm-tools, and libtpm0 packages to be available for installation on the Compute nodes. These are available in the Ubuntu 22.04 LTS release and are expected to be backported to Ubuntu 20.04 LTS. In the interim, the OpenStack Charmers team has included a backport of these packages in the ppa:openstack-charmers/swtpm PPA.

glance charm: Cinder iSCSI/FC backend storage support

The glance charm adds support for Cinder iSCSI/FC backend storage. This is managed through three new configuration options:

cinder-volume-types

A comma-separated list of Cinder volume_types that can be configured as Cinder storage backends. The first one in the list will be configured as the default backend.

cinder-http-retries

The number of cinderclient retries on failed HTTP calls.

cinder-state-transition-timeout

The timeout (seconds) of a Cinder volume transition.

keystone charm: new action to query Keystone admin password

The keystone charm has a new get-admin-password action. This action retrieves the admin password of the Keystone service.

cinder charm: new option to expose scheduler_default_filters

The cinder charm has a new configuration option: scheduler-default-filters. It sets the value of scheduler_default_filters in the Cinder configuration.

ironic-conductor charm: new option to expose use_ipmitool_retries

The ironic-conductor charm has a new configuration option: use-ipmitool-retries. This option sets the value of use_ipmitool_retries in the Ironic configuration.

ceph-mon charm: add list-crush-rules action

The ceph-mon charm has a new list-crush-rules action, which provides a list of CRUSH rules defined in Ceph clusters.

The action has a format parameter that accepts the following values:

  • ‘json’ - provides detailed information in json format

  • ‘yaml’ - provides detailed information in yaml format

  • ‘text’ - provides less information in human readable format [default]

nova-cloud-controller charm: new option to set scheduling retries

The nova-cloud-controller charm has a new configuration option: scheduler-max-attempts. This will allow for an increase in the number of retries and hence hosts to schedule on, thereby helping in a successful scheduling of instances. It sets the scheduler.max attempts in the Nova configuration.

nova-compute charm: new option to add extra repositories

The nova-compute charm has a new configuration option for adding extra apt repositories to Compute nodes: extra-repositories. This option takes a comma-delimited list of apt source repository spec entries to add as apt package repositories. The valid values are those accepted by the add-apt-repository command.

Increase KeepAlive timeout for OpenStack API endpoints and dashboard

The HTTP KeepAlive timeout for OpenStack API endpoints and the Dashboard has been changed to 75 seconds. The previous timeout of 5 seconds (Apache’s default) was causing unnecessary termination of client TCP connections, which was also affecting inter-service (OpenStack) communication.

neutron-api charm: new option to enable network-segment-range support

The neutron-api charm has a new option for enabling Neutron’s ‘Network Segment Range’ service plugin: enable-network-segment-range. It allows cloud operators to dynamically manage network segment ranges through the Neutron API. For more details, refer to Network segment ranges in the upstream documentation.

nova-compute charm: new options to set block device timeout

The nova-compute charm has two new options:

  • block-device-allocate-retries

  • block-device-allocate-retries-interval

These options configure the block device allocation timeout. The default values have been set at 300 and 3 respectively, resulting in an overall timeout of 15 minutes. The previous (inherited upstream) timeout of 3 minutes resulted in failures when dealing with large guest images (e.g. Windows or customised Linux).

nova-cloud-controller charm: new option to expose max_local_block_devices

The nova-cloud-controller charm has a new option: max-local-block-devices. It exposes the upstream max_local_block_devices flag. In particular, setting it to ‘0’ will forbid local block devices, effectively compelling users to request volumes instead. For more information, refer to block device mapping FAQ section in the upstream documentation.

nova-cloud-controller charm: new options to filter hosts by isolating aggregates

The nova-cloud-controller charm has three new options:

  • limit-tenants-to-placement-aggregate

  • placement-aggregate-required-for-tenants

  • enable-isolated-aggregate-filtering

These options are useful for limiting host aggregates to specific tenants. For more information, refer to Filtering hosts by isolating aggregates in the upstream documentation.

nova-compute charm: new option to expose reserved_host_disk_mb

The nova-compute charm has a new option: reserved-host-disk. It takes into account available host disk space when scheduling instances. It is similar to existing options reserved-host-memory and reserved-huge-pages.

openstack-dashboard charm: new options to allow for branding

The openstack-dashboard charm has three new options:

  • site-branding

  • site-branding-link

  • help-url

These options are used to set Dashboard parameters that reflect the local environment.

ceph-osd: new options for disk creation

The add-disk action for the ceph-osd charm has incorporated the following options:

  • osd-ids

  • cache-devices

  • partition-size

ceph-osd: new action to remove disks

The ceph-osd charm has a new remove-disk action. This action allows operator to remove previously created disks.

Documentation updates

A summary of the most significant documentation updates is given below.

In the OpenStack Charm Guide:

  • More cloud operations

  • Improvements to the upgrade pages

  • New tutorial for deploying OpenStack

  • New guidelines and resources for documentation and software contributions

  • Add a spellchecker to the build process

  • New page on virtual TPM devices

  • Refactor of the Project and Community sections

New tech-preview charms

ceph-nfs

The ceph-nfs charm provides action-managed NFS storage backed by CephFS. It is a principal charm used in conjunction with a deployed Ceph cluster.

cinder-nimblestorage

The cinder-nimblestorage charm provides NimbleStorage storage backend support for the OpenStack Cinder service. It is a subordinate charm used in conjunction with the cinder principal charm.

cinder-solidfire

The cinder-solidfire charm provides SolidFire storage backend support for the OpenStack Cinder service. It is a subordinate charm used in conjunction with the cinder principal charm.

nova-compute-nvidia-vgpu

The nova-compute-nvidia-vgpu charm provides Nvidia vGPU support to the OpenStack Nova Compute service. It is a subordinate charm used in conjunction with the nova-compute principal charm.

Informational notices

nova-cloud-controller: cpu-allocation-ratio default changed to 2

The default value for cpu-allocation-ratio has been reduced from 16 to 2. The old default was more appropriate for dev, test, or lab type environments but is rarely suitable for clouds running production workloads. If you were relying on the previous default of 16 and start to see VM scheduling failures after the upgrade of this charm, you can opt back into a higher contention ratio by running:

juju config nova-cloud-controller cpu-allocation-ratio=16

rabbitmq-server: options removed

The configuration options modulo-nodes and known-wait have been removed from the rabbitmq-server charm as development work related to coordinated node restarts has made them irrelevant.

Issues discovered during this release cycle

glance-simplestreams-sync endpoint change

The ceph-radosgw charm improves how URLs are processed by the RADOS Gateway. This change however will lead to breakage for an existing product-streams endpoint, set up by the glance-simplestreams-sync application. Manual intervention is required - see the Upgrade issues page for more information.

Nova fails to parse new libvirt mediated device name format

The name format of mediated devices in libvirt was recently changed from mdev_<uuid> to mdev_<uuid>_<parent>. For users of the new tech-preview nova-compute-nvidia-vgpu charm, this will cause new Nova instances to enter into an error state subsequent to a vGPU device being attached to an instance. This is being tracked in issue LP #1951656. A fix will soon be available as an SRU.