Nova evacuate fails with ImageBusy exception on Ceph enabled envs
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Mirantis OpenStack |
Confirmed
|
Medium
|
MOS Nova |
Bug Description
Detailed bug description:
Using nova evacuate leads to changing instance state to ERROR after destroy compute where instance is located. To move instance to another compute nova evacuate should be used one more time.
Details: https:/
Steps to reproduce:
Use environment with ceph(controller + 2compute+ceph)
1. Create net01
2. Boot instances vm1 and vm2 in net01 on a single compute node
3. Shut off the compute node where instances are scheduled (with virsh destroy)
4. Evacuate instances vm1 and vm2:
nova evacuate vm1
5. Check that vm1 is rescheduled onto another compute node and are in ACTIVE state
Details: https:/
Expected results:
Instance is in Active state and recheduled to another compute
Actual result:
Instance is in Error state
Reproducibility:
Always on that cluster
Workaround:
Use nova evacuate vm1 twice. After that instance move to another compute and ping is working.
Impact:
Description of the environment:
9.2 snapshot #632
cat /etc/fuel_build_id:
495
cat /etc/fuel_
495
cat /etc/fuel_release:
9.0
cat /etc/fuel_
mitaka-9.0
rpm -qa | egrep 'fuel|astute|
fuel-nailgun-
network-
fuel-9.
fuel-misc-
fuel-mirror-
shotgun-
python-
fuelmenu-
fuel-library9.
fuel-notify-
fuel-setup-
rubygem-
fuel-release-
fuel-provision
fuel-ostf-
fuel-bootstrap
fuel-ui-
fuel-agent-
python-
nailgun-
fuel-migrate-
fuel-utils-
fuel-openstack
Additional information:
Log from compute: https:/
Changed in mos: | |
milestone: | none → 9.2 |
assignee: | nobody → MOS Nova (mos-nova) |
importance: | Undecided → Medium |
status: | New → Confirmed |
summary: |
- Nova evacuate should be used twice when use cluster with Ceph + Nova evacuate fails with ImageBusy exception on Ceph enabled envs |
description: | updated |
Changed in mos: | |
milestone: | 9.2 → 9.x-updates |