Bug 1388119
Summary: | Unable to snapshot image from instance on ceph storage | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Product: | Red Hat OpenStack | Reporter: | Tim Wilkinson <twilkins> | ||||||||
Component: | openstack-tripleo-heat-templates | Assignee: | Jiri Stransky <jstransk> | ||||||||
Status: | CLOSED DUPLICATE | QA Contact: | Arik Chernetsky <achernet> | ||||||||
Severity: | high | Docs Contact: | |||||||||
Priority: | high | ||||||||||
Version: | 10.0 (Newton) | CC: | abond, afazekas, bengland, bkopilov, dwilson, egafford, eglynn, ekuvaja, fpercoco, jtaleric, mburns, pgrist, pmyers, rhel-osp-director-maint, rsussman, scohen, srevivo, yrabl | ||||||||
Target Milestone: | --- | ||||||||||
Target Release: | 10.0 (Newton) | ||||||||||
Hardware: | x86_64 | ||||||||||
OS: | Linux | ||||||||||
Whiteboard: | DFG:Storage | ||||||||||
Fixed In Version: | Doc Type: | If docs needed, set a value | |||||||||
Doc Text: | Story Points: | --- | |||||||||
Clone Of: | Environment: | ||||||||||
Last Closed: | 2016-11-03 20:53:56 UTC | Type: | Bug | ||||||||
Regression: | --- | Mount Type: | --- | ||||||||
Documentation: | --- | CRM: | |||||||||
Verified Versions: | Category: | --- | |||||||||
oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |||||||||
Cloudforms Team: | --- | Target Upstream Version: | |||||||||
Embargoed: | |||||||||||
Attachments: |
|
Description
Tim Wilkinson
2016-10-24 14:19:47 UTC
Created attachment 1213491 [details]
nova.conf
Created attachment 1213492 [details]
cinder.conf
Created attachment 1213493 [details]
glance-api.conf
#show_multiple_locations = false in Glance API conf. We should be hitting on this https://github.com/openstack/glance/blob/master/glance/api/v2/images.py#L290-L294 I will try to find the Nova code catching that/documentation explaining it. This default should have been fixed already on upstream stable/newton commit: 75adc9f031a43effddf0e16bd74f0c9ead9dbf3a Sorry for not clarifying sooner, this is a duplicate of a bug caught recently in automation, but we will want to be certain the fix is working. The bug and fix are actually in openstack-tripleo-heat-templates-5.0.0-0.5.0rc3.el7ost. I'm going to mark it as a duplicate and the fix should be available in recent puddles. *** This bug has been marked as a duplicate of bug 1382737 *** (In reply to Paul Grist from comment #6) > Sorry for not clarifying sooner, this is a duplicate of a bug caught > recently in automation, but we will want to be certain the fix is working. > The bug and fix are actually in > openstack-tripleo-heat-templates-5.0.0-0.5.0rc3.el7ost. > > I'm going to mark it as a duplicate and the fix should be available in > recent puddles. > > *** This bug has been marked as a duplicate of bug 1382737 *** Bug 1382737 is listed as a documentation issue but our issue is not resolved by the permissions fix proposed. The openstack-tripleo-heat-templates RPM exists only on my director node so I assume it requires a redeploy of the overcloud to acquire the fix. Is that correct? Is there any way to diff the package and apply the resolution to my existing overcloud? A redeploy is not preferred given the current scale of our test env. Thanks. In reply to Tim Wilkinson, #c7 The duplicate bug is a fix to the tripleO heat template yaml files to deploy the correct settings? was the is comment intended for this bug? I may be missing the connection. I'm not well versed in the logic behind the settings, but it definitely looks like the same issue so I was expecting the same fix to be the root cause of your issues. We can follow up more with the people verifying Bug 1382737 if that doesn't appear to be the case. The issue I have is the inability to create a nova image via snapshot of an existing instance (nova image-create) in my OSP10 overcloud. Is that the same problem as bug 1382737? That appears to be about the snapshot functionality in the undercloud. I brought this one up again this morning to confirm the fix should address this issue as well. Folks agreed that this is the same problem. Have you been able to try this again on one of the more recent OSP10 puddles? I have added some more folks to the cc list to get more comments if needed too. I've updated the openstack-tripleo-heat-templates in my undercloud (osp10d) to the latest version but I'm confused as to how that will resolve the problem on the overcloud-controller-0 where the snapshots fail. The scale of the test env prevents me from redeploying. Plz excuse my ignorance, I'm obviously missing something. Can someone give us a hint WHICH heat templates would be relevant here? There are approx. 480 files in this RPM. Or give us a pointer to the diff of the git commit specified in comment 4? Even if I do a brute-force diff the 2 RPMs versions, which i can do, I won't know which changes were relevant. If there is some really small change to be done to a config file on the compute hosts, then we can just copy that to them and be done with this. The only clue I see in comment #4 is this: #show_multiple_locations = false in Glance API conf. Is that the only change that we need to make to Glance conf file? So the code indicates that it throws an exception if show_multiple_locations is set to false, which it is by default apparently. It seems harmless to set it to true, is it? If I do this, it won't trigger this exception, right? But will it have any other side effects? I saw https://github.com/openstack/glance-specs/commit/484999de7601117e42c6c0b3ea0db86dce261a72 Unless you object, this is what we'll try, ok? (In reply to Tim Wilkinson from comment #11) > I've updated the openstack-tripleo-heat-templates in my undercloud (osp10d) > to the latest version but I'm confused as to how that will resolve the > problem on the overcloud-controller-0 where the snapshots fail. The scale of > the test env prevents me from redeploying. Plz excuse my ignorance, I'm > obviously missing something. Tim, The THT change indeed will need redeployment to become effective. If that's not an option you will need to change the glance-api.conf on each node manually and restart it. AFAIK this is not affecting OSP10 anymore. (In reply to Erno Kuvaja from comment #14) > > The THT change indeed will need redeployment to become effective. If that's > not an option you will need to change the glance-api.conf on each node > manually and restart it. AFAIK this is not affecting OSP10 anymore. Thanks, Erno. That's just what I needed and it did resolve the problem. root@tim-controller-0:~ # nova image-create --show --poll 1b07c502-6cc7-41a3-a38a-c3fcd80f44d0 rhel72_snap1 Server snapshotting... 100% complete Finished +------------------+--------------------------------------------------------------------------------------------------------------------------+ | Property | Value | +------------------+--------------------------------------------------------------------------------------------------------------------------+ | base_image_ref | 106a15e2-7105-4bc2-9eba-caecb771be0a | | checksum | - | | container_format | bare | | created_at | 2016-11-16T14:57:39Z | | direct_url | rbd://6942aa60-f4c4-4954-88d7-c67e0d0a0df5/images/a963041e-739f-4266-94f1-e4dd6052bd60/snap | | disk_format | raw | | file | /v2/images/a963041e-739f-4266-94f1-e4dd6052bd60/file | | id | a963041e-739f-4266-94f1-e4dd6052bd60 | | image_location | snapshot | | image_state | available | | image_type | snapshot | | instance_uuid | 1b07c502-6cc7-41a3-a38a-c3fcd80f44d0 | | locations | [{"url": "rbd://6942aa60-f4c4-4954-88d7-c67e0d0a0df5/images/a963041e-739f-4266-94f1-e4dd6052bd60/snap", "metadata": {}}] | | min_disk | 20 | | min_ram | 0 | | name | rhel72_snap1 | | owner | b06fce501cb7494b8cdffd6d922c49ff | | owner_id | b06fce501cb7494b8cdffd6d922c49ff | | protected | False | | schema | /v2/schemas/image | | self | /v2/images/a963041e-739f-4266-94f1-e4dd6052bd60 | | size | 21474836480 | | status | active | | tags | [] | | updated_at | 2016-11-16T14:59:05Z | | user_id | 28026fd404c44f7387aa0eecdfe34fa7 | | virtual_size | - | | visibility | private | +------------------+--------------------------------------------------------------------------------------------------------------------------+ |