Resize openstack icehouse instance at same node BUG

asked 2014-07-22 13:30:55 -0500

totten25 gravatar image

Hi, I believe this problem may be BUG in OpenStack Icehouse Resize Instance. I saw many problem about resizing instance similar me but my problem is difference from others because they cannot resize the instance. I ever installed OpenStack Havana and it can resize instance perfectly but in OpenStack Icehouse, i use same way but i just launch and resize one time and it can resize successfully as same as havana version. If i want to resize more than one time, it will change to resize state then it quickly changes back to active state. everything still be same. I try both of KVM and QEMU but it's nothing happen.

edit retag flag offensive close merge delete

Comments

Without logs it's impossible to understand what's failing. Check your logs, edit your question to add more details.

smaffulli gravatar imagesmaffulli ( 2014-07-22 16:32:34 -0500 )edit

ok I'm sorry I will add important error and warning are shown below after i try resize more than one time /var/log/nova/nova-compute.log in compute node

2014-07-23 06:47:14.411 1610 WARNING nova.compute.resource_tracker [-] [instance: 03971852-647a-4116-b4e7-b65345c1d300] Instance not resizing, skipping migration.

/var/log/nova/nova-scheduler.log

2014-07-22 21:08:26.761 9217 WARNING nova.scheduler.host_manager [req-d46ad3a1-be18-464b-a3b9-11123e481fcc bdb162ee567d4230a988895f2a000a8b9 84ec9bb0ccc34eea84fbf49b557c4a66] Host has more disk space than database expected (43gb > 34gb)

with my respect, i just found 2 useful warning in Controller node and Compute node. Anyway, my settings in nova.conf of Controller node

allow_resize_to_same_host = True, scheduler_default_filters = AllHostsFilter

and Compute node

allow_resize_to_same_host = True

Thank you for your response

totten25 gravatar imagetotten25 ( 2014-07-22 23:17:05 -0500 )edit

From Instance not resizing, skipping migration, it will happen only after i resize more than 1 time. it will be changed to resize state then instance is not resizing and change back to active state as same as i mentioned in first questions.

totten25 gravatar imagetotten25 ( 2014-07-22 23:46:41 -0500 )edit

Oh, it's unbelievable. I can resize instance more than one time but I sometime found new error from nova-compute.log as shown below

2014-07-23 14:17:04.229 1610 INFO nova.compute.manager [req-bc2cc04a-b32b-455b-9112-988b8ad6ab1a00067bdf3023482ebcc6142da5282b36 81e29a87757e4134afd59c1e5292649c] [instance: 77a2e697-451d-41f1-83df-4f2629097751] Setting instance back to ACTIVE after: Instance rollback performed due to: Resize error: Unable to resize disk down.

I deeply research about this error and fortunately, i can figure out it!!!!!. The problem occur from this

# Checks if the migration needs a disk resize down.
    for kind in ('root_gb', 'ephemeral_gb'):
        if flavor[kind] < instance[kind]:
            reason = _("Unable to resize disk down.")
            raise exception.InstanceFaultRollback(
                exception.ResizeError(reason=reason))
totten25 gravatar imagetotten25 ( 2014-07-23 02:55:01 -0500 )edit

I'm not sure that a disk resize down the instance should be whether BUG or prevent the problem from OpenStack older version (Havana) in OpenStack icehouse from code above. it doesn't make sense if root_disk and ephemeral_disk are less than current instance for disk resize down (I understand this point because it may corrupt the filesystem) but it isn't possible in small,medium and large or others flavor will have equally disk capacity. My point is how does OpenStack do when instance is resizing down to smaller flavor? For now, i can comment this condition. Anyway, I never face this behavior when i want to resize down the instance in OpenStack Havana.

totten25 gravatar imagetotten25 ( 2014-07-23 05:03:26 -0500 )edit