cloudstack-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Prachi Damle <>
Subject RE: [ACS42][QA]Test Plan for "Affinity / Anti-affinity Rules"
Date Mon, 08 Apr 2013 20:56:55 GMT
Hi Sangeetha,

The test plan looks fine. A few comments

1) UpdateVMAffinityGroup API - We should add to the testcases, the scenario where an already
existing VM say on host1, is  updated to have affinity group A1 that asks to avoid host1.

In this case, the VM cannot restart on host1, but should start on some other host within or
outside cluster.

2) And #66 Live migration - Live migration of Vms which are associated with anti-affinity
group failing - The live migration will still succeed since it is Admin's choice to use host2
even if it is not suitable.
We just indicate if the host is suitable or not, but CS should not stop Admin from live migrating.

Affinity Group Processing is a pre-planning step. It will set the scope for the deployment
planners, there is no conflict between the planner strategies and affinity groups. These are
separate steps of deployment planning.


-----Original Message-----
From: [] On Behalf Of prasanna
Sent: Friday, April 05, 2013 6:24 AM
Subject: Re: [ACS42][QA]Test Plan for "Affinity / Anti-affinity Rules"

On 4 April 2013 10:40, Sangeetha Hariharan <> wrote:
> Test cases for " Affinity / Anti-affinity Rules" is posted here -
> Reference FS: - 
> nti-affinity+groups
There are planning decisions that are controlled in the global settings that do similar things
as do affinity groups.
UserConcentration and UserDispersion are two kinds of allocators that affect planning decision
very similar to affinity and anti-affinity.
Some cases surrounding the conflicts arising from these settings would need to be added. Rest
of the tests look good given that the scope of the feature is only a single anti-affinity
processor at this time.
View raw message