heat event-list is really slow

[Steven Hardy]

On my system (SSD equipped laptop), with one example stack generating 21 events on launch, it takes over 5 seconds to to heat event-list.

It seems this is because we're doing a request to nova for availability zones for every single event. I guess we need a smarter way to cache this...

# time heat event-list as108

+--------------------------+-----+------------------------+--------------------+----------------------+
| logical_resource_id | id | resource_status_reason | resource_status | event_time |
+--------------------------+-----+------------------------+--------------------+----------------------+
| ElasticLoadBalancer | 796 | state changed | CREATE_IN_PROGRESS | 2013-06-28T15:57:57Z |
| CfnUser | 800 | state changed | CREATE_IN_PROGRESS | 2013-06-28T15:57:58Z |
| CfnUser | 805 | state changed | CREATE_COMPLETE | 2013-06-28T15:57:59Z |
| WebServerKeys | 806 | state changed | CREATE_IN_PROGRESS | 2013-06-28T15:57:59Z |
| WebServerKeys | 807 | state changed | CREATE_COMPLETE | 2013-06-28T15:58:00Z |
| LaunchConfig | 810 | state changed | CREATE_IN_PROGRESS | 2013-06-28T15:58:01Z |
| LaunchConfig | 811 | state changed | CREATE_COMPLETE | 2013-06-28T15:58:01Z |
| ElasticLoadBalancer | 815 | state changed | CREATE_COMPLETE | 2013-06-28T16:00:25Z |
| WebServerGroup | 816 | state changed | CREATE_IN_PROGRESS | 2013-06-28T16:00:25Z |
| ElasticLoadBalancer | 817 | state changed | UPDATE_IN_PROGRESS | 2013-06-28T16:00:40Z |
| ElasticLoadBalancer | 818 | state changed | UPDATE_COMPLETE | 2013-06-28T16:00:40Z |
| WebServerGroup | 819 | state changed | CREATE_COMPLETE | 2013-06-28T16:00:40Z |
| WebServerScaleDownPolicy | 822 | state changed | CREATE_IN_PROGRESS | 2013-06-28T16:00:40Z |
| WebServerScaleUpPolicy | 820 | state changed | CREATE_IN_PROGRESS | 2013-06-28T16:00:40Z |
| WebServerScaleUpPolicy | 821 | state changed | CREATE_COMPLETE | 2013-06-28T16:00:40Z |
| WebServerScaleDownPolicy | 823 | state changed | CREATE_COMPLETE | 2013-06-28T16:00:41Z |
| MEMAlarmHigh | 825 | state changed | CREATE_IN_PROGRESS | 2013-06-28T16:00:42Z |
| MEMAlarmLow | 824 | state changed | CREATE_IN_PROGRESS | 2013-06-28T16:00:42Z |
| MEMAlarmHigh | 826 | state changed | CREATE_COMPLETE | 2013-06-28T16:00:43Z |
| MEMAlarmLow | 827 | state changed | CREATE_COMPLETE | 2013-06-28T16:00:43Z |
+--------------------------+-----+------------------------+--------------------+----------------------+

real 0m5.430s
user 0m0.252s
sys 0m0.046s

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/36001

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/36001
Committed: http://github.com/openstack/heat/commit/3fa39dd2ccc9013cfba4ec8bd5fb79caeba0350e
Submitter: Jenkins
Branch: master

commit 3fa39dd2ccc9013cfba4ec8bd5fb79caeba0350e
Author: Liang Chen <email address hidden>
Date: Sun Jul 7 14:43:45 2013 +0800

    avoid excessive database calls while loading events

    list_events constructs Event objects through Event.load, which causes
    every event record previously loaded into memory to be reloaded again.
    Furthermore, Event.load will also reload the associated stack, thus
    all its resources and template. Another bad thing with that is
    reloading a stack will lead to many extra stack static resolution
    (including calls to nova to resovle Fn::GetAZs if it's referenced in
    the template).

    Fixes bug #1195793

    Change-Id: I4f14c9a2a4f990b753a6e1405e27eb44c53537c9