Terminate instance button should be grayed out while instance is in suspended state

[David Naori]

 When trying to terminate an instance on suspended status via the dashboard it fails with the following (uninformative) error:
""" Error: Unable to terminate instance: XXXX """

The actual error:
Cannot 'delete' while instance is in vm_state suspended

In my opinion terminate instance button should be grayed out while instance is in suspended state.

[Marcelo Dieder (mdieder)]

I have the same problem.

root@cloud01:~#nova list
+--------------------------------------+---------------+-----------+-------------------------------+
| ID | Name | Status | Networks |
| 8f18f6c2-7ec8-4fb3-b1f8-4fe98300e997 | unisinos_vm01 | SUSPENDED | vmnet=10.1.0.3 |
+--------------------------------------+---------------+-----------+-------------------------------+

root@cloud01:~# nova delete unisinos_vm01
ERROR: Cannot 'delete' while instance is in vm_state suspended (HTTP 409)

but I restart the services (compute, api, rabbit) and not working.

I delete manually all (instance files, volume lvm and entries in mysql database):

mysql> delete from block_device_mapping where id="1";
mysql> delete from security_group_instance_association where instance_id="4";
mysql> delete from iscsi_targets where volume_id=6;
mysql> delete from volumes where id='6';
mysql> delete from instance_info_caches where instance_id="8f18f6c2-7ec8-4fb3-b1f8-4fe98300e997";
mysql> delete from instances where uuid="8f18f6c2-7ec8-4fb3-b1f8-4fe98300e997";

rm /var/lib/nova/instances/instance-00000004/ -rf

/etc/init.d/iscsitarget stop
lvremove /dev/nova-volumes/volume-00000006
/etc/init.d/iscsitarget start

[Tihomir Trifonov (ttrifonov)]

I've tested with the default Devstack setup - no problem to terminate a 'Suspended' instance. No errors, all fine.
Please specify what kind of virt driver do you use, what type is the virtual instance. And even more important - is this the latest dev code(Folsom), or Essex? In the NOVA API current code we have:

    # NOTE(maoy): we allow delete to be called no matter what vm_state says.
    @wrap_check_policy
    @check_instance_state(vm_state=None, task_state=None)
    def delete(self, context, instance):

So it seems no limitation to the vm_state for termination.

[David Naori (davidnaori3)]

Essex

    # NOTE(jerdfelt): The API implies that only ACTIVE and ERROR are
    # allowed but the EC2 API appears to allow from RESCUED and STOPPED
    # too
    @wrap_check_policy
    @check_instance_state(vm_state=[vm_states.ACTIVE, vm_states.BUILDING,
                                    vm_states.ERROR, vm_states.RESCUED,
                                    vm_states.SHUTOFF, vm_states.STOPPED])
    def delete(self, context, instance):

[Jake Dahn (jakedahn)]

I'm not sure this should apply to horizon in folsom and forward. I don't think the openstack api cares at all about whether an instance is in a suspended state, it should terminate when you tell it to terminate.

Other thoughts?

[Gabriel Hurley (gabriel-hurley)]

If that's true, I'm okay with letting this go. @jake can you confirm that and if so mark this as a wontfix?

[Gabriel Hurley (gabriel-hurley)]

Eh, I take it back. I'd rather fix this for now since it *is* a problem with Essex and we're trying to be as backwards-compatible as possible.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/11381

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/11381
Committed: http://github.com/openstack/horizon/commit/56c23b5cdf5a0559c5fda33776c9880159bf67f9
Submitter: Jenkins
Branch: master

commit 56c23b5cdf5a0559c5fda33776c9880159bf67f9
Author: Gabriel Hurley <email address hidden>
Date: Tue Aug 14 18:09:36 2012 -0700

    Clean up instance state checking for actions.

    Better checking for which actions are allowed in which states.
    Fixes bug 1008798 and fixes bug 1009893.

    Change-Id: Id1ed33a09af991ccfd36eafc69b83cf203201d91