Bug 1401445

Summary:	heat resource-list to use more than 1 CPU
Product:	Red Hat OpenStack	Reporter:	Ronnie Rasouli <rrasouli>
Component:	openstack-heat	Assignee:	Zane Bitter <zbitter>
Status:	CLOSED WONTFIX	QA Contact:	Amit Ugol <augol>
Severity:	unspecified	Docs Contact:
Priority:	unspecified
Version:	10.0 (Newton)	CC:	mburns, rhel-osp-director-maint, sbaker, shardy, srevivo, therve
Target Milestone:	---
Target Release:	---
Hardware:	Unspecified
OS:	Unspecified
Whiteboard:
Fixed In Version:		Doc Type:	If docs needed, set a value
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2016-12-05 16:19:44 UTC	Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:

Description Ronnie Rasouli 2016-12-05 09:46:07 UTC

Description of problem:
heat resource-list -n 50 overcloud 
Could be very slow, we noticed that the command doesn't use more than 1 CPU while running it.

pidstat 1 -C 'heat-engine'
Linux 3.10.0-512.el7.x86_64 (undercloud-0.redhat.local) 12/04/2016
_x86_64_ (8 CPU)

01:44:23 PM   UID       PID    %usr %system  %guest    %CPU   CPU  Command
01:44:24 PM   187     30195    0.99    0.00    0.00    0.99     6  heat-engine
01:44:24 PM   187     30229    0.99    0.00    0.00    0.99     7  heat-engine
01:44:24 PM   187     30234  100.00    0.00    0.00  100.00     3  heat-engine

01:44:24 PM   UID       PID    %usr %system  %guest    %CPU   CPU  Command
01:44:25 PM   187     30195    1.00    0.00    0.00    1.00     4  heat-engine
01:44:25 PM   187     30233    1.00    0.00    0.00    1.00     2  heat-engine
01:44:25 PM   187     30234  100.00    1.00    0.00  100.00     3  heat-engine

01:44:25 PM   UID       PID    %usr %system  %guest    %CPU   CPU  Command
01:44:26 PM   187     30195    1.00    0.00    0.00    1.00     7  heat-engine
01:44:26 PM   187     30231    1.00    0.00    0.00    1.00     1  heat-engine
01:44:26 PM   187     30234  100.00    0.00    0.00  100.00     3  heat-engine

01:44:26 PM   UID       PID    %usr %system  %guest    %CPU   CPU  Command
01:44:27 PM   187     30195    1.00    1.00    0.00    2.00     5  heat-engine
01:44:27 PM   187     30232    1.00    0.00    0.00    1.00     0  heat-engine
01:44:27 PM   187     30234  100.00    0.00    0.00  100.00     3  heat-engine

01:44:27 PM   UID       PID    %usr %system  %guest    %CPU   CPU  Command
01:44:28 PM   187     30195    1.00    0.00    0.00    1.00     7  heat-engine
01:44:28 PM   187     30234  100.00    0.00    0.00  100.00     3  heat-engine

01:44:28 PM   UID       PID    %usr %system  %guest    %CPU   CPU  Command
01:44:29 PM   187     30195    1.00    0.00    0.00    1.00     5  heat-engine
01:44:29 PM   187     30234   99.00    0.00    0.00   99.00     3  heat-engine
01:44:29 PM   187     30235    1.00    0.00    0.00    1.00     6  heat-engine

01:44:29 PM   UID       PID    %usr %system  %guest    %CPU   CPU  Command
01:44:30 PM   187     30195    1.00    0.00    0.00    1.00     6  heat-engine
01:44:30 PM   187     30227    1.00    0.00    0.00    1.00     5  heat-engine
01:44:30 PM   187     30229    1.00    0.00    0.00    1.00     5  heat-engine
01:44:30 PM   187     30234  100.00    0.00    0.00  100.00     3  heat-engine



Version-Release number of selected component (if applicable):
openstack-heat-engine-7.0.0-7.el7ost.noarch

How reproducible:
100%

Steps to Reproduce:
1.
2.
3.

Actual results:
heat resource-list very slow performance. 

There is no difference in number of CPU

heat-engine is capping out a single CPU even though > 1
heat-engine workers are available to take the load.

Expected results:
better heat performance use hyper threading / multi core technology

Additional info:

Comment 1 Thomas Hervé 2016-12-05 09:59:55 UTC

I don't think there is anything to do here. We worked on the performance issue of resource-list, which should be acceptable now. Using several threads isn't relevant.

Comment 2 Zane Bitter 2016-12-05 16:19:44 UTC

Python performance is not improved by multithreading.