Open Stack

Hi Ionut,

Not sure if the 'community' image is a bug, but based on my
understanding. The image must be accessible by Nova/Heat to rebuild the
instance by the Magnum auto healer.

After changed the image back to public, the cluster should be
repairable. The error you're seeing "Updating a cluster when status is
"UPDATE_FAILED" is not supported", I guess it's not reported by the
repairing action, but the health status update action. They are different.

You can manually fix the cluster by a manual resize for the cluster. For
example, if you have 3 worker nodes at this moment, then you can just
try command 'openstack coe cluster resize <clusterID> 3' to do a "fake"
resizing. Then the cluster will be updated to the correct status. Hope
that help.


On 16/02/21 11:16 pm, Ionut Biru wrote:
> Hi guys,
>
> I wanted to "decommission"  the old stable fedora-coreos-32 from my
> catalog and I marked the image with community instead of public.
> That's how the nova team suggested that old images should be
> decommissioned .
>
> A cluster deployed with that image had a node shutdown and magnum
> healer tried to resolve the issue and now the status of the cluster is
> in error with:
>  faults               | {'default-master': "Resource UPDATE failed:
> resources[0]: resources.kube_masters.Property error:
> resources.kube_node_volume.properties.image: Error validating value
> 'fedora-coreos-32-x86_64': No images matching {'name':
> u'fedora-coreos-32-x86_64'}.", 'default-worker': "Resource UPDATE
> failed: resources[0]: resources.kube_masters.Property error:
> resources.kube_node_volume.properties.image: Error validating value
> 'fedora-coreos-32-x86_64': No images matching {'name':
> u'fedora-coreos-32-x86_64'}."} |
>
> In the mean time I have marked the image as public and started the
> node but the healer cannot fix it and  it's the status is still in error.
>
> Magnum api has an exception in the logs:
>  magnum.common.exception.NotSupported: Updating a cluster when status
> is "UPDATE_FAILED" is not supported
>
> openstack coe cluster show:
> https://paste.xinu.at/JxYaw/
>
> What's the right path to fix this issue?
>
> -- 
> Ionut Biru - https://fleio.com

-- 
Cheers & Best regards,
Feilong Wang (王飞龙)
------------------------------------------------------
Senior Cloud Software Engineer
Tel: +64-48032246
Email: flwang at catalyst.net.nz
Catalyst IT Limited
Level 6, Catalyst House, 150 Willis Street, Wellington
------------------------------------------------------ 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-discuss/attachments/20210217/61bf19ab/attachment-0001.html>