Heat fails to create volume if no availability zone provided

[Marcin Zbik]

All nova/cinder hosts are in non "nova" availability zone, I mean they are in, for example "my_az" availability zone. In such case Heat will fail to deploy nova server with block_device_mapping_v2 until on explicitly set availability_zone property for OS::Nova::Server

Part of template:

resources:
  server:
    type: OS::Nova::Server
    properties:
      user_data_format: SOFTWARE_CONFIG
      key_name: { get_param: key_name }
      block_device_mapping_v2:
        -
          boot_index: 0
          delete_on_termination: True
          device_name: vda
          volume_size: 80
          device_type: disk
          image_id: { get_param: image }
      flavor: { get_param: flavor }
      networks:
        - port: { get_resource: server_port }
      user_data:
        get_resource: server_init

Exact same behavior when booting with block_device_mapping (*NOT* v2) and when not specifying availability_zone in OS::Cinder::Volume

Request snippet from cinder-scheduler.log in DEBUG mode:

{
 'request_spec': {
  u 'source_replicaid': None,
  u 'volume_properties': {
   u 'volume_metadata': [],
   u 'availability_zone': u 'nova',
   u 'reservations': [u '0382a27d-ab15-442a-be63-63ec23cc3632', u '111c639b-0948-4c9e-8ef5-c02ba6d540b8'],
   u 'replication_status': u 'disabled',
   u 'snapshot_id': None,
   u 'id': u 'e3019d25-6275-42f8-9dcb-2f73d37e3837',
   u 'size': 80,
   u 'display_name': u '',
   u 'source_replicaid': None,
   u 'display_description': u '',
   u 'project_id': u 'd877486d3dd24be995b07e12285cc089',
   u 'metadata': {},
   u 'status': u 'creating',
   u 'volume_type_id': None,
   u 'multiattach': False,
   u 'source_volid': None,
   u 'qos_specs': None,
   u 'consistencygroup_id': None,
   u 'user_id': u '11a682025d5dceb1a431c02008a0369430f1b32365006691fc526b13da62abe8',
   u 'attach_status': u 'detached',
   u 'cgsnapshot_id': None,
   u 'encryption_key_id': None
  },
  u 'source_volid': None,
  u 'cgsnapshot_id': None,
  u 'image_id': u 'ce17b5db-5d4d-45a9-ad02-ad8e01d5302a',
  u 'snapshot_id': None,
  u 'consistencygroup_id': None,
  u 'volume_type': {},
  u 'volume_id': u 'e3019d25-6275-42f8-9dcb-2f73d37e3837'
 }
}

As one can see there is u 'availability_zone': u 'nova', which will obviously fail with NoValidHost. Workaround for this is to add availability_zone property to OS::Nova::Server (in case of block_device_mapping_v2) or to OS::Cinder::Volume (in case of block_device_mapping).

Heat from OpenStack Liberty (Version: 1:5.0.1-0ubuntu1~cloud0)

# heat-manage db_version
65

[Thomas Herve (therve)]

It looks like a Nova issue? Can you make it work with the equivalent nova boot command line call?

[Marcin Zbik (zbikmarc+launchpad)]

It looks you are right:

Nova boot command:
nova boot --flavor m1.large --block-device source=image,id=ce17b5db-5d4d-45a9-ad02-ad8e01d5302a,dest=volume,size=60,shutdown=remove,bootindex=0 --nic net-id=36674c7e-8bd8-46cf-b3df-2926ed1ef373 instanceFromVolume

cinder-scheduler.log
Task 'cinder.scheduler.flows.create_volume.ExtractSchedulerSpecTask;volume:create' (7a3adbd4-6257-4714-8985-174a666e1eb9) transitioned into state 'SUCCESS' from state 'RUNNING' with result '{'request_spec': {u'source_replicaid': None, u'volume_properties': {u'volume_metadata': [], u'availability_zone': u'nova', u'reservations': [u'e5a1089b-28b7-47f0-a45c-f386925804f2', u'ae4428c6-554a-42cb-b7b9-9e84546c9946'], u'replication_status': u'disabled', u'snapshot_id': None, u'id': u'b80f8c57-7806-4f87-9663-ebcab3052e87', u'size': 60, u'display_name': u'', u'source_replicaid': None, u'display_description': u'', u'volume_admin_metadata': [], u'project_id': u'd877486d3dd24be995b07e12285cc089', u'metadata': {}, u'status': u'creating', u'volume_type_id': None, u'multiattach': False, u'source_volid': None, u'qos_specs': None, u'consistencygroup_id': None, u'user_id': u'1e54cb2fc8734e47984f9d7d3be93d97', u'attach_status': u'detached', u'cgsnapshot_id': None, u'encryption_key_id': None}, u'source_volid': None, u'cgsnapshot_id': None, u'image_id': u'ce17b5db-5d4d-45a9-ad02-ad8e01d5302a', u'snapshot_id': None, u'consistencygroup_id': None, u'volume_type': {}, u'volume_id': u'b80f8c57-7806-4f87-9663-ebcab3052e87'}}' _task_receiver /usr/lib/python2.7/dist-packages/taskflow/listeners/logging.py:178
2016-05-11 16:09:19.254 17300 DEBUG cinder.scheduler.manager [req-3316c438-0f41-469b-ad45-724b3afcd55c 1e54cb2fc8734e47984f9d7d3be93d97 d877486d3dd24be995b07e12285cc089 - - -] Task 'cinder.scheduler.flows.create_volume.ScheduleCreateVolumeTask;volume:create' (4b400d37-400d-48d5-b0d6-0c236cd4f7e4) transitioned into state 'RUNNING' from state 'PENDING' _task_receiver /usr/lib/python2.7/dist-packages/taskflow/listeners/logging.py:190
2016-05-11 16:09:19.271 17300 DEBUG cinder.openstack.common.scheduler.base_filter [req-3316c438-0f41-469b-ad45-724b3afcd55c 1e54cb2fc8734e47984f9d7d3be93d97 d877486d3dd24be995b07e12285cc089 - - -] Starting with 8 host(s) get_filtered_objects /usr/lib/python2.7/dist-packages/cinder/openstack/common/scheduler/base_filter.py:77
2016-05-11 16:09:19.271 17300 INFO cinder.openstack.common.scheduler.base_filter [req-3316c438-0f41-469b-ad45-724b3afcd55c 1e54cb2fc8734e47984f9d7d3be93d97 d877486d3dd24be995b07e12285cc089 - - -] Filter AvailabilityZoneFilter returned 0 host(s)
2016-05-11 16:09:19.272 17300 WARNING cinder.scheduler.filter_scheduler [req-3316c438-0f41-469b-ad45-724b3afcd55c 1e54cb2fc8734e47984f9d7d3be93d97 d877486d3dd24be995b07e12285cc089 - - -] No weighed hosts found for volume with properties: {}
2016-05-11 16:09:19.275 17300 ERROR cinder.scheduler.flows.create_volume [req-3316c438-0f41-469b-ad45-724b3afcd55c 1e54cb2fc8734e47984f9d7d3be93d97 d877486d3dd24be995b07e12285cc089 - - -] Failed to run task cinder.scheduler.flows.create_volume.ScheduleCreateVolumeTask;volume:create: No valid host was found. No weighed hosts available

Should I recreate this bug on Nova ...

[Thomas Herve (therve)]

yes please!