heat stack creation could use some profiling report

[Removed by request]

Description of the problem:
Long/large deployments (for instance tripleo), can take hours to complete/update and can take longer then usual to work on a given resource. For instance, if a user is used for a certain stack to be created in X minutes and it ends up being created in 3X minutes instead, it might indicate an issue worth investigating.

Current status:
Resource-show can give me the time to create from (updated_time - creation_time) but it does not give me update time on each consequent stack update.
Moreover creating a stack on multiple nodes (for instance using a large resource group) will require a user to go over hundreds of sql lines just to get a sort of understanding of which one of those took the longest to create.

What I would like to see here, is something in the spirit of a time delta report when using resource list. For instance, adding the created_time and adding a new change_time to the resource list output:

trimmed... resource_type | resource_status | creation_time | change_time | updated_time
           parent::resource | UPDATE_COMPLETE | 00:00 | 02:00 | 05:00

And when checking the nested resources:

trimmed... resource_type | resource_status | creation_time | change_time | updated_time
           child1::resource | UPDATE_COMPLETE | 00:00 | 02:00 | 02:10
           child2::resource | UPDATE_COMPLETE | 00:00 | 02:00 | 02:20
           child3::resource | UPDATE_COMPLETE | 00:00 | 02:00 | 04:59 <- ??

[Thomas Herve (therve)]

I may understand what you want, but storing additional data is not a good start to improve performance. We already support openstack profiler, can't you extract this sort of data with it?