cloudstack-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Kishan Kavala" <Kishan.Kav...@citrix.com>
Subject Re: Review Request 12967: Not enough cpu avialable in cluster according to new overcommit ratio but all stopped Vms can be started without any failure
Date Wed, 31 Jul 2013 13:09:58 GMT

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/12967/#review24345
-----------------------------------------------------------

Ship it!


commit ba619ec112170b916f69a7466a779196f74bef7c

- Kishan Kavala


On July 31, 2013, 3:24 p.m., bharat kumar wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/12967/
> -----------------------------------------------------------
> 
> (Updated July 31, 2013, 3:24 p.m.)
> 
> 
> Review request for cloudstack, Kishan Kavala and Nitin Mehta.
> 
> 
> Bugs: Cloudstack-1695
> 
> 
> Repository: cloudstack-git
> 
> 
> Description
> -------
> 
> https://issues.apache.org/jira/browse/CLOUDSTACK-1695
> Not enough cpu avialable in cluster according to new overcommit ratio but all stopped
Vms can be started without any failure(required cpu for stopped vm is >> available cpu
)
> 
> 
> Diffs
> -----
> 
>   server/src/com/cloud/capacity/CapacityManagerImpl.java 05a47dd 
>   server/src/com/cloud/vm/VirtualMachineManagerImpl.java 1fb118f 
> 
> Diff: https://reviews.apache.org/r/12967/diff/
> 
> 
> Testing
> -------
> 
> Tested on old master.
> This happened because we were not recalculating the used and the reserved capacities
based on the new overcommit ratio. 
> 
> Now used and reserved capacities of all the hosts in a overcommited cluster will be recalculated
on overcommit change by the capacity checker. However this will happen only when capacity
checker runs meaning the change in capacities due to updated overcommit values will not be
instantaneous.
>   
> 
> 
> Thanks,
> 
> bharat kumar
> 
>


Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message