Cinder show does not inform about snapshot

[Maxim V. Yefimov]

MOS 4.1

We noticed that when we delete cinder volumes, it will error out if there are snapshots

Running cinder delete 523e0d8a-0405-4927-b4ab-2e9763ab07a8
ERROR: Invalid volume: Volume still has 1 dependent snapshots

# cinder snapshot-list --all-tenants
+--------------------------------------+--------------------------------------+-----------+----------------------+------+
| ID | Volume ID | Status | Display Name | Size |
+--------------------------------------+--------------------------------------+-----------+----------------------+------+
| 10ed0442-ab0e-462a-a096-317641eb9185 | 523e0d8a-0405-4927-b4ab-2e9763ab07a8 | available | | 30 |
| 19e01a80-dff2-45e0-8ef7-1c4ecbf23a99 | 0d37f162-2d3a-4f9c-87c3-f9c24b4cb388 | available | XXX | 200 |
| 2616a17c-e4e5-4280-87f8-2418346af5f2 | 93a997e7-6bc8-43ed-851f-fc2f5c8814ce | available | XXX | 200 |
| 529c561f-7021-4e69-b51b-06fdd128d2d2 | d7b6e516-2065-4f82-8269-b30eb5e2426b | available | | 10 |
| 64f6be93-3254-4850-8ca0-1cf2fe0f3699 | 42af7955-b7ed-449c-85ee-7099a7017129 | available | XXX | 5 |
| 78c4222b-05f7-49cb-9766-2e9b7c6d30ac | 86b351a5-cbe4-4d15-a261-77e631bcd2ed | available | XXX | 5 |
| a5aff6bf-3e02-4156-8bb6-bf53d96f50d4 | 3b5f182b-cdd4-4ec2-a36e-475e968fe33a | available | | 30 |
| a7f973a7-e09f-4425-907c-87a1d52502eb | ec7b2a8c-78c1-46c3-87af-a3a634137448 | available | | 10 |
| d7f4bb6c-0a1d-4a22-9ff0-874277c21990 | 0d37f162-2d3a-4f9c-87c3-f9c24b4cb388 | available | XXX | 200 |
| faccca8a-9009-48d7-a524-ef0056024f7e | ea52373a-cf78-4fe6-8c7a-859751504912 | available | XXX | 1 |
+--------------------------------------+--------------------------------------+-----------+----------------------+------+

But cinder show does not tell us it's a snapshot:

# cinder show 523e0d8a-0405-4927-b4ab-2e9763ab07a8
+--------------------------------+--------------------------------------+
| Property | Value |
+--------------------------------+--------------------------------------+
| attachments | [] |
| availability_zone | nova |
| bootable | false |
| created_at | 2014-10-10T14:55:10.000000 |
| display_description | None |
| display_name | None |
| id | 523e0d8a-0405-4927-b4ab-2e9763ab07a8 |
| metadata | {u'readonly': u'False'} |
| os-vol-host-attr:host | xxx.xxx.xxx |
| os-vol-mig-status-attr:migstat | None |
| os-vol-mig-status-attr:name_id | None |
| os-vol-tenant-attr:tenant_id | 055df6b2b44845bfb863059fa2358c6e |
| size | 30 |
| snapshot_id | None |
| source_volid | None |
| status | available |
| volume_type | None |
+--------------------------------+--------------------------------------+

[Alexander Rubtsov (arubtsov)]

Related community bugs:
https://bugs.launchpad.net/cinder/+bug/1111839
https://bugs.launchpad.net/cinder/+bug/1276101

[Ivan Kolodyazhny (e0ne)]

Working on fix in upstream for https://launchpad.net/bugs/1111839

[Ivan Kolodyazhny (e0ne)]

This issue requires API changes so it will be fixed only in 7.x+ after patch will be merged in upstream

[Dmitry Borodaenko (angdraug)]

If it's not going to be fixed properly until 7.x, please document a workaround in the bug description.

[Dmitriy Novakovskiy (dnovakovskiy)]

Gentlemen,

Do I understand correctly that ANY volume delete operation will fail in any current Mirantis OpenStack installation if the volume has snapshots? Any backend (Ceph, NetApp), any OpenStack distro (assuming no patch was applied specifically for this problem) ?

[Dmitry Borodaenko (angdraug)]

Dmitriy N.: Yes, this is not specific to Mirantis distro, any current OpenStack installation will have the same problem regardless of backend.

Ivan K.: please document a workaround as requested in comment #4 two weeks ago.

[Ivan Kolodyazhny (e0ne)]

Dmitry B,

the problem it that cinder does not _show_ snapshots if volumes list or volume details is called. Cinder doesn't allow do delete volume if it has snapshot. It's a common behavior to not delete objects if it contains depended data. Volumes could be deleted only if all it's snapshots are deleted.

We can't fix it in MOS <7 because it requires API changes. It could be REST API or python-cinderclient API. In such cases our OpenStack will have different API with upstream.

I can't say that it is workaround, because it's a feature request, not bug. May be it's only UX issue.

you need to call 'cinder snapshot-list | grep <vol_id>' before volume delete and delete all volume's snapshots using 'cinder snapshot-delete <snapshot_id>' command

[Dmitry Mescheryakov (dmitrymex)]

We are not going to fix the issue in 6.x because it requires changes in OpenStack API. And we don't want MOS API be different from the upstream OpenStack. Hence we work in the upstream on fixing the issue.

[Ivan Kolodyazhny (e0ne)]

It's not fixed in Kilo. I hope to get it merged in Liberty/MOS 8.0.

Actually, it's not an but. It is feature request. We've got a very simple and working workaround for it so I change priority ti Medium