[system_test] Ceph multinode compact sys test fails on deployment

[Tatyanka]

http://jenkins-product.srt.mirantis.net:8080/view/5.1_swarm/job/5.1_fuelmain.system_test.centos.thread_1/22/consoleFull
id | status | name | cluster | ip | mac | roles | pending_roles | online
---|-------------|------------------------------|---------|------------|-------------------|----------------------|---------------|-------
1 | provisioned | slave-02_compute_ceph-osd | 1 | 10.108.0.3 | 64:4d:97:f4:e5:86 | ceph-osd, compute | | True
2 | error | slave-01_controller_ceph-osd | 1 | 10.108.0.4 | 64:c2:06:1f:e3:dc | ceph-osd, controller | | True
3 | error | slave-03_compute_ceph-osd | 1 | 10.108.0.5 | 64:ac:02:90:e3:ea | ceph-osd, compute | | True

in astute node 2 failed with error, and puppet log says that keystone can not be run from first attepmt, but it was successful started from other one, and in puppet log we can see that keystone is operational, and puppet run finish with sucess.
But astute see error and decide that deployment failed, seems we need add some logic to handle errors in keys if we pass some deployment step from several attempts

[root@nailgun astute]# cat /etc/fuel/version.yaml
VERSION:
  feature_groups:
    - mirantis
  production: "docker"
  release: "5.1.1"
  api: "1.0"
  build_number: "28"
  build_id: "2014-10-14_00-01-06"
  astute_sha: "f5fbd89d1e0e1f22ef9ab2af26da5ffbfbf24b13"
  fuellib_sha: "46ad455514614ec2600314ac80191e0539ddfc04"
  ostf_sha: "64cb59c681658a7a55cc2c09d079072a41beb346"
  nailgun_sha: "b9792cb5bbecddfa9c5c3afb4d0f961a2a2776a7"
  fuelmain_sha: "431350ba204146f815f0e51dd47bf44569ae1f6d"

[Dmitry Borodaenko (angdraug)]

Please don't target Medium and lower priority bugs for maintenance releases.

[Vladimir Sharshov (vsharshov)]

Hi. This was as i know problem with bad keystone package. This problem was already fixed by Matt. This is problem do not related with Astute as i can see. Link: https://bugs.launchpad.net/mos/+bug/1391257

 > But astute see error and decide that deployment failed, seems we need add some logic to handle errors in keys if we pass some deployment step from several attempts

Because node 2 was controller node in simple cluster mode, we fail deployment. Is anything wrong with such behavior?

Logs:
     2014-10-14T00:09:48 warning: [424] 946b39ab-b943-4ab3-9e40-28b444dec30a: Critical nodes with uids: 2 fail to deploy. Stop deployment