We need to investigate if we can logically test these better (without varying so much timewise).
http://logs.openstack.org/36/116336/36/gate/check-heat-dsvm-functional-mysql/02df233/console.html
http://logs.openstack.org/36/116336/38/check/check-heat-dsvm-functional-mysql/b9c991d/console.html
heat_integrationtests.scenario.test_neutron_loadbalancer.NeutronLoadBalancerTest.test_neutron_loadbalancer 332.804
2015-03-18 10:03:36.421 | 2015-03-18 10:03:35.022 | heat_integrationtests.functional.test_resource_group.ResourceGroupTest.test_update_increase_decrease_count 268.368
2015-03-18 10:03:36.422 | 2015-03-18 10:03:35.024 | heat_integrationtests.scenario.test_server_cfn_init.CfnInitIntegrationTest.test_server_cfn_init 241.258
2015-03-18 10:03:36.423 | 2015-03-18 10:03:35.026 | heat_integrationtests.functional.test_instance_group.InstanceGroupUpdatePolicyTest.test_instance_group_update_replace_huge_batch_size 205.132
2015-03-18 10:03:36.423 | 2015-03-18 10:03:35.028 | heat_integrationtests.functional.test_instance_group.InstanceGroupUpdatePolicyTest.test_instance_group_update_replace_huge_min_in_service 93.108
2015-03-18 10:03:36.423 | 2015-03-18 10:03:35.030 | heat_integrationtests.functional.test_instance_group.InstanceGroupUpdatePolicyTest.test_instance_group_update_no_replace 192.948
2015-03-18 10:03:36.424 | 2015-03-18 10:03:35.032 | heat_integrationtests.functional.test_instance_group.InstanceGroupUpdatePolicyTest.test_instance_group_update_replace_with_adjusted_capacity 160.822
2015-03-18 10:03:36.424 | 2015-03-18 10:03:35.033 | heat_integrationtests.functional.test_autoscaling.AutoscalingGroupUpdatePolicyTest.test_instance_group_update_replace_huge_min_in_service 149.795
2015-03-18 10:03:41.865 | 2015-03-18 10:03:35.035 | heat_integrationtests.functional.test_heat_autoscaling.HeatAutoscalingTest.test_path_attrs_nested 147.330
2015-03-18 10:03:41.865 | 2015-03-18 10:03:35.037 | heat_integrationtests.functional.test_autoscaling.AutoscalingGroupUpdatePolicyTest.test_instance_group_update_no_replace 133.241
Seeing a lot of timeouts on this in the gate. Given that this was initially reported against the RPC nested stacks patch, it seems like there is a good chance that is what has slowed things down to the point where timeouts have become an issue. It's not surprising that this would cause extra load (since we have only a single thread, and it is designed to make things scale *out* more efficiently), but I suspect a comparison with previous runtimes may reveal it to be disproportionate.
The slowest tests from a recent failed run were:
test_ update_ nochange_ resource_ needs_update [1049.335623s] gnalTest. test_signal_ with_policy_ update [832.644550s] pUpdatePolicyTe st instance_ group_update_ no_replace [698.874717s] instance_ group_update_ replace_ huge_min_ in_service [508.550756s]
AutoScalingSi
all of AutoscalingGrou
test_
test_