[openstack][quuens][cinder] consistency groups bug ?

Gorka Eguileor geguileo at redhat.com
Thu Sep 22 09:03:19 UTC 2022


On 03/09, Ignazio Cassano wrote:
> Hello,
> some updates.
> The issue appears also using volume group instead of consistency groups.
> Ignazio
>

Hi,

Consistency Groups do not work across Cinder backends:

- Backends from different drivers/vendors.

- Backends from different storage arrays in different geographic
  locations

I don't know if there's a way to make it "work" with generic groups, but
if it were possible it would probably just loop over the volumes and
take individual snapshots, as I don't think it's implemented the option
of having multiple consistency groups (1 per backend) and doing 1
snapshot request per backend.

Cheers,
Gorka.

>
> Il giorno ven 2 set 2022 alle ore 19:06 Ignazio Cassano <
> ignaziocassano at gmail.com> ha scritto:
>
> > Hi everyone, in my openstack installation, for performance reasons, I use
> > a cinder type to address different storage virtual machines on the same
> > device. Balancing takes place via the scheduler. Below is an example of
> > my configuration in cinder.conf:
> >
> > https://paste.openstack.org/show/b0hCTsdniLhinoQiVT7A/
> >
> > As you can see nfsgold1 to nfsgold8 are all part of the volume type
> > nfsgold.
> > Unfortunately if I create a consistency group with volume type nfsgold, it
> > can only contain volumes on the same backend, even though they are of the
> > same type. For example, if in my consistency group there is a volume on
> > nfsgold1 and I add a volume present on nfsgold8 I get the following:
> >
> > https://paste.openstack.org/show/b4XnUeGDF5wYlCBLTcuc/
> >
> >
> > Any help, please ?
> > Ignazio
> >




More information about the openstack-discuss mailing list