Detaching and then attaching a volume to a paused instance has no effect and reports no error
Bug #1427732 reported by
Artem Plakunov
This bug affects 2 people
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
OpenStack Compute (nova) |
Expired
|
Undecided
|
Unassigned |
Bug Description
Steps:
1. Create VM
2. Create volume
3. Attach volume to VM
4. Pause VM
5. Detach volume
6. Attach volume back (without unpausing the VM)
What happens is the volume enters the state "attaching" for a while but then becomes "available" again without being "in-use". Nova-compute and cinder-volume logs report nothing about it even with debug enabled
Nova version: 2014.2.1
I am aware of the bug https:/
Changed in nova: | |
assignee: | nobody → Eric Xie (mark-xiett) |
assignee: | Eric Xie (mark-xiett) → nobody |
Changed in nova: | |
status: | New → Incomplete |
importance: | Undecided → Low |
status: | Incomplete → Confirmed |
Changed in nova: | |
assignee: | nobody → lyanchih (lyanchih) |
tags: | removed: juno-backport-potential |
Changed in nova: | |
status: | New → Incomplete |
To post a comment you must log in.
are you using libvirt driver? is there any log in nova api layer?
please note that the attach is checked something like: at compute/api.py which is API layer code
looks like PAUSED instance should also be able to attach volume, but could you please help to check again?
3036 @check_ instance_ state(vm_ state=[ vm_states. ACTIVE, vm_states.PAUSED, SOFT_DELETED] )
3037 vm_states.STOPPED, vm_states.RESIZED,
3038 vm_states.
3039 def attach_volume(self, context, instance, volume_id, device=None,
3040 disk_bus=None, device_type=None):