[Openstack] block_device_mapping cause long time detaching
ZHOU TAO A
tao.a.zhou at alcatel-lucent.com
Tue Aug 26 08:39:32 UTC 2014
Hi all,
I used a heat template to launch a server and attach cinder volume to it.
I found that if I use the block_device_mapping parameter in my server
properties, then it will took a very long time to detach the volume when
I delete the stack; without block_device_mapping, the volume will be
immediately detached.
And I don't see any difference from the VM's perspective, whether I use
block_device_mapping or not, the volume will be attached the my VM as 'vdb'.
I am curious about what does this block_device_mapping mean and what
difference does nova behave under the hood.
Here's my heat template:
heat_template_version: 2013-05-23
description: |
jfdkfjdk System
outputs:
instance_ip:
description: the ip address
value: {get_attr: [Myserver, networks, OAM120, 0]}
parameters:
image_name:
label: Image
type: string
default: MCcloudinit
flavor_name:
label: fff
type: string
default: 4x8x160
prinet1:
label: Netffd
type: string
default: OAM120
resources:
Myserver:
metadata:
mytest_id: tzhou002isgreat
type: OS::Nova::Server
properties:
availability_zone: nova
block_device_mapping: [
{volume_id: {get_resource: disk1},
delete_on_termination: 'true',
device_name: vdb}]
flavor: { get_param: flavor_name }
image: { get_param: image_name }
config_drive: 'True'
name:
str_replace:
params:
$system_name: {get_param: "OS::stack_name"}
template: $system_name-0-0-1
networks:
- network: {get_param: prinet1}
security_groups: [default]
disk1:
type: OS::Cinder::Volume
properties:
availability_zone: nova
name: server1_disk1
size: '1'
disk_attach:
type: OS::Cinder::VolumeAttachment
properties:
instance_uuid: {get_resource: Myserver}
volume_id: {get_resource: disk1}
mymultipartmime:
properties:
parts:
- config: {get_resource: 'Myserver'}
- config: {get_resource: 'disk1'}
type: OS::Heat::MultipartMime
More information about the Openstack
mailing list