update to instances/servers fails because ports are still attached
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
OpenStack Heat |
Fix Released
|
High
|
huangtianhua |
Bug Description
if add a new port to ec2::instance in template.,update will fail.openstack version is havana(stable)
template before:
my_instance:
# Use an AWS resource type since this exists; so why use other name here?
type: AWS::EC2::Instance
properties:
KeyName: { get_param: KeyName }
ImageId: { get_param: ImageId }
InstanceType: { get_param: InstanceType }
NetworkIn
template update:
my_instance:
# Use an AWS resource type since this exists; so why use other name here?
type: AWS::EC2::Instance
properties:
ImageId: { get_param: ImageId }
InstanceType: { get_param: InstanceType }
NetworkIn
trace:
014-02-14 16:58:38.493 22709 TRACE heat.engine.
2014-02-14 16:58:38.493 22709 TRACE heat.engine.
when update a stack,heat won't delete the resource until the new resource is created,in this case,the new instance want to use the port which is still used by the old instance;so nova raise a conflict exception.I think will also happen when add a new volume to ec2:instance.
Changed in heat: | |
assignee: | nobody → huangtianhua (huangtianhua) |
Changed in heat: | |
milestone: | none → next |
importance: | Undecided → Medium |
summary: |
- update failed if add a new port to ec2::instance in template + update to instances/servers fails because ports are still attached |
Changed in heat: | |
importance: | Medium → High |
Changed in heat: | |
milestone: | next → none |
status: | Fix Committed → Fix Released |
I think we can make the 'NetworkInterfaces' property updatable to avoid this problem.