Fuel Master CI should fail sooner if docker fails

[Matthew Mosesohn]

Fuel Master CI test takes a full ~50 mins to fail instead of responding sooner. Since this test will be run against many patches, it takes too long to fail. Instead, the test should look in /var/log/puppet/bootstrap_admin_node.log for a message that matches the following expression:
ERROR: .*failed to start.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/132582

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/132586

[Tatyanka (tatyana-leontovich)]

We have sometimes here several retries to start some container, so in your case if first attempts fails -> test will fail too(But second one attempt may be successful)

[Matthew Mosesohn (raytrac3r)]

What we should do is if dockerctl check CONTAINERNAME fails, it will print this error. What isn't so clear is should we fail 100% here and abort all deployment, or just write it and exit 1 at the end? This "first attempt" fail is not accurate. We test 255 times to see if the container is ready (about 9 minutes) before marking it as failed.

For the scope of Fuel Master CI, I believe we should just consider it failed immediately by parsing the log for this message and marking the CI test as a failure then and there.

[OpenStack Infra (hudson-openstack)]

Change abandoned by Egor Kotko (<email address hidden>) on branch: master
Review: https://review.openstack.org/132586
Reason: Abandon because of this https://review.openstack.org/#/c/133440/5

[OpenStack Infra (hudson-openstack)]

Change abandoned by Egor Kotko (<email address hidden>) on branch: master
Review: https://review.openstack.org/132582
Reason: Abandon because of https://review.openstack.org/#/c/133440/5