netapp copyoffload provider location None

[Friedrich Hiekel]

After creating an volume with copyoffload while have more than one nfs shares, it fails to create an imagecache out of it because volume['provider_location'] is None.

This is the reason why we then get this warning: "Exception while registering image <Image-ID> in cache. Exception: object supporting the buffer API required"
And it is not creating a new cache inside the nfs share.

(https://github.com/openstack/cinder/blob/8b222bdd1df4c0bcfe46f42a5dfb229424dbb9f9/cinder/volume/drivers/netapp/dataontap/nfs_base.py#L381)

By volume creation the volume database entry is created with the field provider_location = null.
When cinder tries to register a image-cache from the new volume it gets the volume information from the database and can not derive the mountpoint for "None" this leads to a crash with the message above.

We are using currently queens.

[Sofia Enriquez (lsofia-enriquez)]

Greetings Friedrich Hiekel,
I'm not sure if this bug is going to have the attention you need because it is based on stable/queens. However, I see that you pointed out a code that is in master, so this issue may be a problem in the current code. I'm going to mention this in the email bug report but feel free to propose your patch to Gerrit so the cinder squad could review it [1].
Cheers
[1] https://docs.opendev.org/opendev/infra-manual/latest/developers.html#committing-changes

[Friedrich Hiekel (fhiekel)]

Good morning Sofia Enriquez,

thanks for you answer. I will create this patch in Gerrit.  As far as I
can see it, this is a problem in all versions, but unfortunately we have
not updated our cluster, thats why I have only reported it for queens.

have a nice day

Friedrich

On 23.03.22 00:06, Sofia Enriquez wrote:
> Greetings Friedrich Hiekel,
> I'm not sure if this bug is going to have the attention you need because it is based on stable/queens. However, I see that you pointed out a code that is in master, so this issue may be a problem in the current code. I'm going to mention this in the email bug report but feel free to propose your patch to Gerrit so the cinder squad could review it [1].
> Cheers
> [1] https://docs.opendev.org/opendev/infra-manual/latest/developers.html#committing-changes
>
> ** Changed in: cinder
> Importance: Undecided => Low
>
> ** Also affects: cinder/queens
> Importance: Undecided
> Status: New
>
> ** Changed in: cinder/queens
> Importance: Undecided => Low
>
> ** Changed in: cinder
> Status: New => Incomplete
>

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.opendev.org/c/openstack/cinder/+/834822

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.opendev.org/c/openstack/cinder/+/834822
Committed: https://opendev.org/openstack/cinder/commit/f9c926fe407ebf3fc8c97e6f02b45c4d32e8c493
Submitter: "Zuul (22348)"
Branch: master

commit f9c926fe407ebf3fc8c97e6f02b45c4d32e8c493
Author: Friedrich Hiekel <email address hidden>
Date: Wed Mar 23 09:14:42 2022 +0100

    add netapp copyoffload provider location

    By volume creation the volume database entry is created with the field provider_location = null.
    When cinder tries to register an image-cache from the new volume it gets the volume information
    from the database and can not derive the mountpoint for "None".
    This fix will set the provider location to prevent the error mentioned in the issue.

    Closes-Bug: 1965952
    Change-Id: I0fe11542d2d5440d1048d388e7a7b09055cf3af8

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/cinder 22.0.0.0rc1 release candidate.