Comment 2 for bug 1564387

This seems to be already fixed since is not reproduced on current master branch.
See details at http://paste.openstack.org/show/493344/

Now the results are:
+------------------------------------------------------------------------------------------------------------------+
| Response Times (sec) |
+---------------+-----------+--------------+--------------+--------------+-----------+-----------+---------+-------+
| Action | Min (sec) | Median (sec) | 90%ile (sec) | 95%ile (sec) | Max (sec) | Avg (sec) | Success | Count |
+---------------+-----------+--------------+--------------+--------------+-----------+-----------+---------+-------+
| yars_action_1 | 1.001 | 2.501 | 3.703 | 3.853 | 4.004 | 2.501 | 80.0% | 5 |
| yars_action_2 | 2.001 | 3.001 | 3.003 | 3.003 | 3.004 | 2.752 | 80.0% | 5 |
| yars_action_3 | 1.001 | 1.501 | 2.002 | 2.002 | 2.002 | 1.501 | 100.0% | 4 |
| total | 5.003 | 6.504 | 8.406 | 8.707 | 9.007 | 6.755 | 80.0% | 5

You see that all atomic actions (but yars_action_3) has 5 runs (this is correct), and 80% (because 1 iteration has failed),
and yars_action_3 has not started once so there are 4 runs and 100% success (all 4 iterations had no errors)