[OSA] Failover at the storage level

Dmitriy Rabotyagov noonedeadpunk at gmail.com
Sat Jul 29 16:27:25 UTC 2023


I would say that quite little to do OSA itself, but how to design services
themselves.
Using config_overrides in OpenStack-Ansible allows you to define any
service configuration you want.
So you can leverage nova_nova_conf_overrides for adding any arbitrary
configuration to nova.conf and cinder_cinder_conf_overrides for cinder.conf.

Regarding redundancy - I can't really advice as I'm not really happy with
any of approaches available due to one or another reason. We ended up with
having disaster recovery instead implemented with cinder-backup to RGW (S3)
storage in the different region. As it also supports incremental backups,
which is quite neat

On Sat, Jul 29, 2023, 17:12 Murilo Morais <murilo at evocorp.com.br> wrote:

> Good morning everybody!
>
> I'm trying to raise a scenario with redundancy at the storage level.
> I'm using CEPH with RBD-mirror.
>
> Cinder has a pretty clear documentation about it, but I didn't find
> anything about Glance and the "nova_libvirt_images_rbd_pool" and the
> "cinder_service_backup_driver".
>
> What would be the best approach?
> Only use Cinder with Mirror?
> If by chance the "nova_libvirt_images_rbd_pool" dies, if you are using
> config-drive the disk will die, correct? In this case, would it be better
> to leave local?
>
> I can't see what would be the best scenario for Failover cases with OSA.
>
> Thanks in advance!
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.openstack.org/pipermail/openstack-discuss/attachments/20230729/68c2eb0b/attachment.htm>


More information about the openstack-discuss mailing list