Volume type with extra spec creation failure

[Kyrylo Romanenko]

Tempest tests failure:
tempest tempest.api.volume.admin.test_volume_types_extra_specs.VolumeTypesExtraSpecsV1Test.test_volume_type_extra_spec_create_get_delete[gate,smoke]FAIL
tempest tempest.api.volume.admin.test_volume_types_extra_specs.VolumeTypesExtraSpecsV2Test.test_volume_type_extra_spec_create_get_delete[gate,smoke]FAIL

Possible way of fix found:

1. By additional configuring of deployment:
1.1
On controller node edit file /etc/cinder/cinder.conf
add there strings to [DEFAULT] section:

default_volume_type = hp_msa_driver
enabled_backends = hp_msa_driver

And append there following section:

[hp_msa_driver]
volume_backend_name = HPMSA_FC
san_password = !manage
san_login = manage
san_ip = 192.168.0.2
volume_driver = cinder.volume.drivers.san.hp.hp_msa_fc.HPMSAFCDriver

where san_ip equals osapi_volume_listen variable.

1.2
On fuel node in MOS-Scale rejoined:
Create and edit tempest.conf, add following to [volume] section:
storage_protocol = FC
vendor_name = HP
build_timeout = 196

1.3
In tempest test script tempest/api/volume/admin/test_volume_types.py
change from

extra_specs = {"storage_protocol": proto,
                       "vendor_name": vendor}

to

extra_specs = {"storage_protocol": "hp_msa_fc",
                       "vendor_name": "HP"}

This way also described in issue: https://bugs.launchpad.net/tempest/+bug/1353850

2. Also test will not fail if omit extra_spec from function call
Exclude passing extra_specs argument in test tempest/api/volume/admin/test_volume_types.py
by changing
body = self.volume_types_client.create_volume_type(
            vol_type_name,
            extra_specs=extra_specs)
to
body = self.volume_types_client.create_volume_type(
            vol_type_name)

3. In Upstream OpenStack this testcase passes with Ceph successfully, looks like issue is fixed there
Fetch_to_volume_format calls copy_volume using wrong parameter
Change-Id: Ia3b0f9168235a977a12232e27a5755ad11ec18f5
https://review.openstack.org/150232
Tests log:
http://logs.openstack.org/32/150232/3/check/check-tempest-dsvm-full-ceph/d255607/console.html
find there:
tempest.api.volume.admin.test_volume_types_extra_specs.VolumeTypesExtraSpecsV2Test.test_volume_type_extra_spec_create_get_delete [0.818180s] ... ok
tempest.api.volume.admin.test_volume_types_extra_specs.VolumeTypesExtraSpecsV1Test.test_volume_type_extra_spec_create_get_delete [1.375878s] ... ok

Some suspicious logs fragments from Controller node`s Cinder API:
Date Level Message
2015-01-28 10:07:41 ERR
Failed to run task cinder.scheduler.flows.create_volume.ScheduleCreateVolumeTask;volume:create: No valid host was found. No weighed hosts available

192.168.0.1 - - [27/Jan/2015 17:20:31] "DELETE /v1/3845131a1b2142618500d2b4b29b977f/types/b14b85cf-35cc-47d1-8f06-ff6295f20ab6 HTTP/1.1" 202 206 0.061489
2015-01-28 08:46:53 INFO
http://172.16.0.2:8776/v1/3845131a1b2142618500d2b4b29b977f/types/b14b85cf-35cc-47d1-8f06-ff6295f20ab6 returned with HTTP 202
2015-01-28 08:46:53 INFO
http://172.16.0.2:8776/v1/3845131a1b2142618500d2b4b29b977f/types/b14b85cf-35cc-47d1-8f06-ff6295f20ab6
2015-01-28 08:46:53 INFO
(3921) accepted ('192.168.0.1', 50539)
2015-01-28 08:46:53 INFO
192.168.0.1 - - [27/Jan/2015 17:20:31] "GET /v1/4d37b7a9f8234f0d954e71ee437b68a5/volumes/eb3f37b2-c741-45a1-8c89-b64ea140b443 HTTP/1.1" 404 357 0.070169
2015-01-28 08:46:53 INFO
http://172.16.0.2:8776/v1/4d37b7a9f8234f0d954e71ee437b68a5/volumes/eb3f37b2-c741-45a1-8c89-b64ea140b443 returned with HTTP 404
2015-01-28 08:46:53 INFO
exception thrown: The resource could not be found.

Environments:
Tempest test failure was detected first on following environments:
Release fuel-6.0-58-2014-12-26_14-25-46, Ubuntu + Ceph backend for Cinder, CentOS + Ceph backend of Cinder.

Then issue was discovered on next deployments:

A) Remote deployment
Fuel 6.0 Juno on Ubuntu 12.04.4, 2014.2-6.0
5 nodes: 3 Controller+CephOSD; 2 Computes
Tempest deployed with MOS-Scale

B) Local on VirtualBox
Fuel 6.1 build: fuel-6.1-81-2015-01-27_11-30-12.iso on VirtualBox
Ubuntu 12.04.4, 2014.2-6.1:
3 nodes: Controller, Compute, Ceph-OSD
Tempest deployed with MOS-Scale

In both environments A) and B) driver configuration changes described above allow test to pass OK.

[Yuriy Nesenenko (ynesenenko)]

This bug wasn't reproduced. The specified tests are working.

[Yuriy Nesenenko (ynesenenko)]

It was worked under:
VERSION:
  feature_groups:
    - mirantis
  production: "docker"
  release: "6.1"
  api: "1.0"
  build_number: "201"
  build_id: "2015-03-16_10-48-35"
  nailgun_sha: "874df0d06e32f14db77746cfeb2dd74d4a6e528c"
  python-fuelclient_sha: "cc64fff91fb0d78e5a02e7b93ceff224296f84fb"
  astute_sha: "93e427ac49109fa3fd8b0e1d0bb3d14092be2e8c"
  fuellib_sha: "553cb0cffa40a5f57313f962b6ec6a9bd89306ba"
  ostf_sha: "e86c961ceacfa5a8398b6cbda7b70a5f06afb476"
  fuelmain_sha: "c97fd8a789645bda48d06da224f994f8b52d82f5"