Bug 1051037

Summary: Vm start on not correct host under power saving policy
Product: Red Hat Enterprise Virtualization Manager Reporter: Artyom <alukiano>
Component: ovirt-engineAssignee: Martin Sivák <msivak>
Status: CLOSED NOTABUG QA Contact: Artyom <alukiano>
Severity: unspecified Docs Contact:
Priority: unspecified    
Version: 3.3.0CC: acathrow, dfediuck, gklein, iheim, lpeer, mavital, msivak, Rhev-m-bugs, sherold, yeylon
Target Milestone: ---Keywords: Triaged
Target Release: 3.5.0   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard: sla
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2014-08-27 09:16:18 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: SLA RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Bug Depends On:    
Bug Blocks: 1142923, 1156165    

Description Artyom 2014-01-09 15:32:44 UTC
Description of problem:
Have two hosts when on first cpu loading 50% and on second 0, start new vm, vm run on host with cpu loading 0.

Version-Release number of selected component (if applicable):
is31

How reproducible:
Always

Steps to Reproduce:
1. Add two hosts, create new vm, run vm, check on what host vm started(for example host_1), poweroff vm.
2. Change cluster policy to power saving(with default parameters), load host_2 cpu to 50%
3. Run vm

Actual results:
Vm run on host_1

Expected results:
Vm run on host_2(because we have power_saving weight module)

Additional info:
Hosts have the same cpu's(same number of cpu's and cores per cpu's)
It's no logs, because no any useful information about internal scheduling process

Comment 1 Artyom 2014-01-22 15:57:30 UTC
After some discussion with Martin, revealed that weight module looks on cpu usage of host, just when it have at least one vm, so you can fast start many vms on host with cpu usage 0%, and after this because balancing policy, vm's starts to migrate on host with cpu usage 50%, one vm each balance round. Now question if it desirable behavior, if it's ok I just close this bug, if not I can close this bug and open some RFE bug.
Martin your decision?

Comment 2 Martin Sivák 2014-02-18 09:22:09 UTC
Probably close or repurpose this bug and file the RFE. This issue only happens when you start your first Vm in the cluster so it can be considered a corner case.

Comment 3 Martin Sivák 2014-08-27 09:16:18 UTC
Technically we do not support loading a host by something else than VMs. The logic will work properly after the first VM is started.

If you want to change this behaviour, file an RFE and propose a new behaviour.