No way to change the image state to kill if error during nova instance snapshot

Bug #1191115 reported by David Peraza
18
This bug affects 4 people
Affects Status Importance Assigned to Milestone
Glance
Invalid
Undecided
David Peraza

Bug Description

When there is a failure in nova snapshot before uploading the image to glance (for example failure on hypervisor capturing the disk image as a file) there is no way to notify glance that this image is in error state nor the is a way to set it to kill. This will result on uncertainty from a client side checking image state, image state will be in queue until the call from nova to upload which will then go to saving and the active or killed depending on the success of upload to store. If an error occurs before image upload request, nova cannot call update with image to upload and image will stay in queue state. Client will not be able to determine if this is a long running snapshot or something bad has happened.

One suggestion could be to add a new known property named error to the update request. If api sees this it will then set the image to error and add the error to image metadata. I will try a couple of things here, but wanted to get other folks to pitch in on possible solutions.

Revision history for this message
Mark Washenberger (markwash) wrote :

Capturing an irc log

dperaza: hello all, I just added this bug report, will like to get feedback from folks that have more experience. I could be missing something: https://bugs.launchpad.net/glance/+bug/1191115
[1:10pm] markwash: dperaza: Hi
[1:10pm] markwash: dperaza: you've hit on one of our top pain points with nova and glance
[1:10pm] markwash: and its got a bit of a history
[1:11pm] markwash: dperaza: from glance's perspective, a failed upload should put the image back into the "queued" or "pending" state
[1:12pm] markwash: dperaza: more satisfying solutions are going to have to involve greater orchestration from nova

Revision history for this message
David Peraza (dperaza) wrote :

Hello Mark: thanks for response. So how do you differentiate between legitimate long running snapshots and a a snapshot the fails right away, how would you do this today? I think the only gap for glance is to allow errors to be reported, that was we can put image in a state that indicates something already happened. Of course nova virt drivers will need to then report to glance the errors during their error handling, but the would be outside the scope of glance.

Revision history for this message
OpenStack Infra (hudson-openstack) wrote : Fix proposed to glance (master)

Fix proposed to branch: master
Review: https://review.openstack.org/33550

Changed in glance:
assignee: nobody → David Peraza (dperaza)
status: New → In Progress
Revision history for this message
John Bresnahan (jbresnah) wrote :
Revision history for this message
Erno Kuvaja (jokke) wrote :

proposed fix abandoned. Over year old. If this still persists we should open a new bug.

Changed in glance:
status: In Progress → Invalid
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.